Josh Miller, CEO of The Browser Company in his latest post on finally sunsetting Arc:

So when people ask how venture capital influenced us — or why we didn’t just charge for Arc and run a profitable business — I get it. They’re fair questions. But to me, they miss the forest for the trees. If the goal was to build a small, profitable company with a great team and loyal customers, we wouldn’t have chosen to try and build the successor to the web browser – the most ubiquitous piece of software there is.

I used Arc since the beta and I loved it. Arc was the first credible and ambitious attempt to reinvent web browsing that was actually able to get some traction. Unfortunately, it didn’t get enough.

Davie Pierce covered this for the story on the original announcement of Dia:

A strange thing has happened over the last couple of years, Miller says. Arc has grown fast — users quadrupled this year alone — but it has also become clear that Arc is never going to be a truly mainstream product. It’s too complicated, too different, too hard to get into. “It’s just too much novelty and change,” Miller says, “to get to the number of people we really want to get to.” User interviews and data have convinced the company that this is a power-user tool, and always will be. On the other hand, the people who use Arc tend to love Arc. They love the sidebar, they love having spaces and profiles, they love all the customization options. Generally speaking, those users have also settled into Arc — Miller says they don’t want new features as much as they just want their browser to be faster, smoother, more secure. And fair enough! So The Browser Company faced a situation many companies encounter: they had a well-liked product that was never going to be a game-changer. Rather than try to build the next thing into the current thing, and risk both alienating the people who like it and never reaching the people who don’t, the company decided to just build something new.

Many people believe that the venture capital and the “endless сhase for growth” is the culprit here. After all, The Browser Company raised $50M at a $550M valuation in 2024. I’d say these people don’t truly understand how startups and venture capital work.

Less than 0.1% of new businesses in the US receive venture capital funding. Venture capital is intended for companies that would never be able to get a bank loan. It’s an exception, not a rule, reserved only for the riskiest bets, because 90% of startups fail within 10 years. The one and only reason some crazy people are willing to fund them is they might find the next Google or at least an Uber and earn billions. The zero-marginal-cost nature of the tech industry this makes possible: once a product works, scaling it is far easier than scaling something tied to physical infrastructure (Walmart took 63 years to reach a $700B market cap).

Most startups in any given VC portfolio fail. But you don’t know which ones will, so investors place multiple bets, evaluating what it would take for each company to become a unicorn.

Fred Wilson, General Partner at Union Square Ventures, one of the best VC firms in the game, wrote in 2016:

Our first USV fund, our 2004 vintage, has turned out to be the single best VC fund that I have ever been involved in. We made 21 investments. We made money on twelve of those investments. We lost money on nine of them. And we lost our entire investment on most of those nine failed investments. The reason that fund performed so well has pretty much nothing to do with the losses. It was all about five investments in which we made 115x, 82x, 68x, 30x, and 21x. It wasn’t like we were swinging for the fences in that fund. Every single one of those 21 investments seemed like an intelligent investment decision at the time we made it. But many of them didn’t work. We lost all or almost all of our money on over 40% of our investments in that fund.

Mind me, this is the best fund of one of the best investors out there. The majority of funds don’t even beat the S&P 500. Carta published their data on the VC performance at the end of 2024:

For the 2017 vintage, for instance, median IRR fell from 16.8% as of Q4 2021 to 12.0% at the end of Q4 2024.

During that same period, the S&P grew by 2.5x—so you’d be slightly ahead if you’d gone with public markets (with full liquidity along the way).

Indeed, taking venture capital means committing to building a fast-growing product with enormous market potential. It’s not the only way to build a company, and nobody is forced to take it. In fact, anyone who’s raised funding can tell you how hard it is to convince fund managers that your startup is the next big thing.

But it might be one of the few viable paths to building a specific kind of company: one working on a truly novel product with untested technology and unclear product-market fit. One that needs to hire a full team and pay them salaries before generating revenue.

Like the Browser Company.

The same people who lament businesses for looking to grow their revenue probably care a lot about their paycheck coming on time each month. And this team has over 130 people in employment.

Arc was an extremely ambitious endeavour. The team went to rebuild the entire user interface from scratch, something neither Microsoft nor Brave were brave enough to do with their Chromium forks. But this unique interface built for the modern age is exactly what fueled Arc’s growth and recognition. Meanwhile, charging a subscription simply to use the product would kill it in the cradle since most of the growth was happening because of the word-of-mouth and all other browsers from big tech companies are free. Unfortunately, that also means they don’t care much about them.

The middle ground simply wasn’t possible.

***

There are many small myths around venture capital. People made wild statements based on the fact that people invest in companies, as if having Founders Fund on your company’s captable means Peter Thiel is personally making decisions for you.

Or you might point to Zed, an open-source Arc’s copycat as a counterexample to this story. Well, I’ve tried it out and I have to say that lifting 99% of Arc’s design still didn’t fix the natural roughness of consumer open source.

There are plenty of companies, even in software, that never raised venture capital, ranging from JetBrains to Basecamp, and a myriad of others you and I have never heard about. There are even unique companies that switched from a venture path to a cashflow-generating business like Buffer, who bought out their investors, and Gumroad, whose founder essentially got a gift from the investors who wrote the company off. Most successful venture-backed companies like Webflow or Figma raise funding on their own terms just to have a war chest.

Venture capital didn’t kill Arc. It gave us the chance to see what Arc could be. That’s more than most products ever get.

Get into the car, take out the phone, put in into a flimsy piece of plastic that is somehow attached to the vents, connect the charger. Navigate to your destination.

That was the life before CarPlay. Now 79% of buyers aren’t even going to look at a car that doesn’t have it.

Don’t get me wrong, CarPlay is ifinitely better than what we had before. Car manufacturers operate in a very different paradigm than smartphone manufacturers. They source individual blocks and parts years in advance, so in the end your car problaby has a dozen of small computers running different areas, and multimedia was never their strong suit.

But after trying out Tesla, I have to say I prefer their built-in navigator and multimedia experience to CarPlay. Rivian is probably the same, along with other electric-first brands.

CarPlay has the lowest denominator problem. Since Apple doesn’t know what kind of screen the car will have, they have to adapt. In addition to this, Apple is extremely cautious and paternalistic. Because of this, the experience is worst than it could have been.

Here’s a simplest scenario that works great in a Tesla. Look at the map, pinch to zoom out, find a place you want to go but can’t be bothered to remember the address. Long tap, go. I do it all the time when I want to drive to a random place in a city or when I’m traveling in another country and maybe looking for the next sightseeing point. Or when you realize that the road is blocked and you have to force the nav to go a weird route.

You can’t really do that with CarPlay. All navigation apps, whether it’s Apple Maps, Google Maps or Waze, have to operate accoridng to their invisible rules. There’s no pinch-to-zoom because your car’s screen can be shitty, you have to tap a button to change the scale, adjust it and get back. You can’t just long tap on a specific point and drive there—never works reliably for me. And worst of all, even if you’re just at your driveway, you can’t simply take out your phone and do the thing there—your nav will be stucked on turn-by-turn directions, because Apple doesn’t want you to look at the phone. Unless you know the name of the place or an address, you’re screwed.

There’s more. Yes, CarPlay lets you have your music or podcast app, even if you use a third-party one like Castro. But you can’t scroll the song or the podcast episode. Again, because Apple believes you shouldn’t. The apps themselves are pretty bad. Go try and open Spotify simply to play Liked Songs or another recent playlist. It’s not obvious at all. Finally, renting a CarPlay-enabled car in Scotland reminded me how much of your phone’s battery is burned on this, especially with wireless CarPlay. The phone dies in a few hours.

Now, the unique advantage of CarPlay is that you indeed have your phone’s envirionment with all the apps. The world is not limited to the US and Western Europe and other countries have their own maps and music services, which probably have iOS apps. Still, Tesla locks you in with their providers. For a few months it didn’t know they turned a crossing nearby into a roundabout. And I have to use either Apple Podcasts or Spotify to listen to my shows. Well, actually I don’t, because I can continue the same show I had on my morning dog walk by simply tapping the Bluetooth button—everything switches automatically and almost instantly. Still, it’s less native (and I can no longer scroll the episode here either). The audio was a problem we solved years ago and I don’t get why people act so annyoned by having to connect to Bluetooth. The phone is already connected the second I get in anyway.

But the entire experience with built-in multimedia so much smoother I’m willing to lose some things. Plus, electric vehicles in general require good navigation specifically as the app has to know your car’s charge and energy consumption to build the route accordingly and plan for charges. Google Maps won’t cut it. With combustion, most of the time you don’t need to think about gas station, you only do this when you realize you’re low on gas already.

I’m not actively opposed to CarPlay. It would have been interesting if Tesla and Rivian used it. Although I doubt Tesla would be willing to drop their Mobile Connectivity subscription. The $10 I pay every month enable the built-in “phone” inside the car to stream music and even videos when you’re stopped. But to their credit, basic navigation and remote control are free, which can’t be said about many legacy manufacturers like Kia.

In case you haven’t been following, there’s a huge drama in the WordPress community. Matt Mullenweg, the creator of WordPress, launched a crusade against WP Engine, a company selling managed WordPress hosting and the competitor of his own entities. You can catch up on the story here.

But the funniest part of this WordPress drama is that WordPress is irrelevant.

Yes, it still powers around 40% of websites. But it’s not the future of the web. You *can* be both big and irrelevant. IBM created PCs as we know them and is currently worth $214Bn, yet nobody cares much about them. In fact, this scandal is probably the most attention WordPress received from the public in the last few years.

WordPress usage has plateaued. It’s not going to die (although Matt is surely trying a lot), but it’s unlikely to grow. Unfortunately, this is exactly when people realize it’s all a zero-sum game, and the fighting starts. Automattic probably believes that taking users away from WP Engine is a more reliable growth channel than trying to acquire new ones.  

Technology products don’t get replaced by new, slightly better products. More often than not, they’re replaced by a completely different paradigm that makes certain use cases and applications radically simpler while throwing away aspects previously deemed important. 

Back in the day, there were countless content management systems (CMS). PHP Nuke and its countless forks, Drupal, Joomla, and many others. As an aspiring developer, I coded my own (but I had no idea how to make other people use it). WordPress did actually win. But then WordPress wasn’t replaced by an improved WordPress. Instead, fully managed proprietary solutions took over.

You want to sell things online? Shopify. You need a landing page? Webflow or Squarespace. You want a blog? Substack or Beehiiv. 

Yes, there are downsides. You’re locked in, and you might have to pay fees on your revenue. These platforms are less customizable, and the lock-in enables them to raise prices over time. 

But they are also dramatically easier to launch and operate. Doesn’t matter if you’re an individual or a small business, the ability to start with a template and edit things without coding experience is amazing. 

So, who uses WordPress? Organizations that prefer to host things themselves. There’s an army of WordPress developers and agencies who can build anything for them, and that alone creates the demand. If you want to self-host, there are no comparable solutions out there, exactly because people rarely develop for the outdated paradigm. People don’t trust WordPress the project and are asking around what are the possible alternatives with the same philosophy. There are practically none. One exception is Ghost, which is an open-source blog engine. But even Ghost moved toward a far more lucrative space of newsletters and paid communities, and it’s not a fully-fledged competitor to WordPress.  

At my previous job, we had an agency build us a website on WordPress. And while we could edit texts on the landing page and add logos and news entries, adjusting anything else was near impossible. My current company has a Webflow website, and I have already rebuilt half of it and added an entire blog. We pay $340 annually for that privilege. Frankly, not much more expensive than premium WordPress hosters.

I know serious tech companies with complex software products and an army of developers who use Webflow for their landing pages. This way, they can keep developers focused on their actual product and let marketers and designers edit the pages as much as they need to.

I have read dozens of excited reviews of Orion like these pieces from The Verge and Ben Thompson. I have also seen a comparable number of people surprised by this wave of affection, like John Siracusa in the most recent ATP episode. 

And I can understand this. Orion can be an impressive demo, but it’s still a demo. A prototype. A concept car. It’s reasonable to assume that Apple might have a similar pair of glasses in their labs somewhere and faced the same cost issues. “Apple doesn’t release prototypes”.

But I think we should stop giving Apple the benefit of the doubt. It’s their job to convince us. We’re excited about Orion precisely because Meta showed them, and Apple didn’t. I honestly do not care if they have something cooking. Yes, they have a good track record overall. But I don’t believe in a second we should buy in on their secrecy. They do it for their benefit, not ours. Otherwise, it becomes a cult with blogger shamans trying to interpret the divine signs. 

Also, Apple Vision Pro is exactly this – an expensive prototype. And probably a failed one. Yes, this might not age well, like the early stories about the death of iPhone. Apple can turn this ship around. But it’s telling that even the people most excited about Apple Vision Pro acknowledge they haven’t touched it in months. It’d be perfect for an airplane, it’d be perfect for a workplace, and yet practically nobody bothers to carry them on their next flight. Who’s the Federico Viticci of Apple Vision Pro, i.e., the person willing to go to unimaginable lengths to use the device they love so much? I don’t know, I haven’t seen one. 

Apple focused on a great passthrough and came out with a compromised device that is too heavy and expensive and is struggling to engage either users or developers. Meanwhile, Meta now has a $300 device with Oculus Quest 3S. It has passthrough that enables you to see around, not bump into furniture or play AR games. Nobody is going outside in these headsets anyway. But in the end, Meta owns the VR market and has an army of users and developers building all kinds of experiences. They have extremely popular Meta RayBan glasses that I actually see people wearing. 

Now, each of these developers understands that the experiences they build might end up on unobtrusive AR glasses in a couple of years. Meanwhile, nobody is pressured to build apps for Apple Vision Pro, which still has neither Netflix nor YouTube and not much immersive content from Apple itself. 

Meta showed us a convincing vision of the future, and they seem perfectly positioned to get there.

Hyperlinks are the foundation of the internet. People say that Tim Berners-Lee invented the internet as we know it when he connected hypertext with TCP/IP. And yet, in 2024, hyperlinks are second-tier citizens at best, as the most popular social platforms don’t support them or penalize you for trying. 

Facebook started deprioritizing links in 2017. Zuckerberg confirmed this later by announcing a broader effort to prioritize “meaningful social interactions” on users’ news feeds. The stated goal was to focus on posts from friends, family, and groups over content from businesses, media outlets, and publishers, particularly those containing outbound links. Twitter joined the party in late 2020, as they also wanted to reduce their reliance on external content.

This shift forever changed the internet. 

Platforms dislike links because they are, by definition, links to other websites. You click a link, and your attention goes elsewhere. But that’s how the internet works. Links are the foundation, and people use links for all kinds of reasons. A desire to start a discussion about an article or a video is one of the incentives to post on social in the first place. 

Meta and Twitter imagine a future where all content is directly syndicated through their properties. Even if they don’t support this kind of content natively. Have you tried using Twitter Articles? They’re primitive, can take 10 seconds to load, and in the end, the engagement figures are not too different from an ordinary post with a link. Twitter videos can’t replace YouTube in their dreams. Most importantly, as an ordinary user, you can’t just reupload someone else’s video or a news article to Twitter to share it natively. 

So you paste a link, and nobody sees it because social platforms push users towards the algorithmic feeds they control. As a result, people post less or nothing at all. 

It doesn’t mean there are fewer posts. But more and more content is instead coming from other accounts that leverage the native tools and mechanics to be favored by the algorithm. And I don’t think their content is particularly good or tasteful. 

One interesting signal in social products is people using weird hacks to overcome the platform’s limitations. In early Instagram, all photos had to be square. But people wanted wide and portrait instead, so they used apps to add bars on the side until Instagram enabled this natively. 

“Link is in the first comment” became such a hack with the hope that the network won’t supress it, even though the conversion would drop drastically anyway. In fact, threads on Twitter are another example of a hack that eventually became a part of the product. Another alternative people used was posting a screenshot of a note instead. They wanted the ability to post long-form content, at least on certain occasions. By the way, Twitter of 2024 definitely suppresses the reach of images that contain lots of text, you can easily see this from the view counts. 

It’s almost like we need a new social network for links. Unfortunately, we did have one for a while, and it died. I’m talking about Artifact, an AI-enabled news app launched by the founders of Instagram, operated as a social network with nothing but links. I loved the app, but there weren’t enough people enamored by this concept. 

The real question is whether we go next.

Telegram is a secretive social media platform with 900 million users. Their employees aren’t allowed to talk about their jobs. It only has one product manager in its founder. And they got to 900 million users without turning into Facebook but also aren’t earning as much.

Telegram is not really a secure messenger app. It’s one of the largest social media platforms. But certain product choices and the location of its audiences make it almost invisible to people based in the United States or Western Europe.

When you first open it, Telegram looks similar to WhatsApp. WhatsApp’s founders famously focused on keeping it simple and clean. For them this meant not adding anything else: “no ads, no games, no gimmicks”. So they added nothing to the app, even years after Meta bought it. About a year or two ago I noticed WhatsApp started adding very specific features that were obviously lifted from Telegram, like channels. But still much slower.

Telegram can get incredibly complex. On top of DMs and group chats, it can host super groups of up 200,000 people. Anyone can create a channel, a publicly-facing broadcast similar to a Facebook page. The largest one now has 55 million followers. You can save anything in your own personal space, and you can organize chats by folders. In addition to channels, Telegram pioneered other UI paradigms, like video circles, which are vastly superior to voice notes (lifted by WhatsApp again).

But despite all this complexity, Telegram hasn’t turned into the infamous Facebook main page. All these features are carefully hidden. Pavel Durov says that he is the only product manager and personally oversees how the app looks. It certainly looks like it. Despite being a social platform, there’s no unified feed, even though it’d provide the best possible ad space.

Compared to Signal or any Meta’s messaging app Telegram provides the most seamless and polished experience. You can instantly switch between conversations. You can quickly find anything shared inside these conversations, including images, files and links. And there are specific searches for all of these categories. Back in 2017, I often had to look up some old files and images in Facebook Messenger, and it was atrocious. Most importantly, you can easily use it on multiple devices with great desktop apps (there are two, for some reason), and everything is synced through the cloud.

Meta provides you with half-baked web-based apps. And with WhatsApp, you have to relink it to the phone constantly. But Telegram can only do this because the messages aren’t encrypted end-to-end. Telegram started as a “secure messaging app” (they launched quite a bit before Signal). But they aren’t using this tagline anymore since it’s not a differentiator, and it’s not true. With Telegram, you have to trust the team not to peek and keep their servers safe and secure. You can create dedicated E2E chats, but only manually, and their functionality is as limited as you’d expect; they’re tied to a particular device and don’t sync at all.

But most users don’t care. This trade-off makes Telegram easy to use, and the team focuses much more on social features anyway. In regions where Telegram is popular, channels replaced newsletters as a form of revolt against algorithmic feeds. They’ve recently revamped user profiles to make them more public-facing and let you pin your channel right there. You can write long texts, share images, and not worry about links being suppressed. On the other hand, you will never get any algorithmic boost. You must proactively look for these channels or get a link elsewhere. In a world where social platforms demand specific content and try to commoditize all content, Telegram feels like a remnant of the old web.

Telegram did add Stories, and I think that was one of the rare misses by the team. Having someone as your Telegram contact is a weaker signal than following on Instagram. Imagine discussing your upcoming partnerships while your stories show how you left the bar at 4 am. Considering the “messaging” part with groups and DMs, they should have made stories visible to “Close Friends” only and asked you to choose the list manually. Then, it could have worked.

Telegram has a notoriously lenient approach to moderation, so you can see accounts from both sides of the Russia-Ukraine war. But unlike modern Twitter, Telegram doesn’t shove it in your face, so even Zelenskiy is happy to use it. While each channel has a URL, the reading interface on the web is extremely basic, and aggressively pushes you to go to the app instead. They definitely could have improved it but chose not to. Telegram is its own kingdom. That’s one of the reasons you rarely see links to its content. Social media apps often use built-in browsers to keep you there, but Telegram goes beyond and provides reader views for most websites, especially news media. No reason to leave the app. Last month they even added its own multi-tasking to Telegram, where you can collapse the browser and continue talking to people or reading channels.

Telegram is the app people use for everything. Talk to their partner. Set up a dinner party. Discuss work projects. And read the news, of course. You can create a group with dedicated threads that look like a Discord/Slack server. I know companies with a few thousand employees that use Telegram as their internal comms tool. It has downsides, of course, as you don’t control your employee’s accounts, and they can mix and match their personal conversation or easily forward messages. Companies write bots to accommodate (like deleting the person from all chats, but it’s still clunky).

With Telegram, you need a phone to set up your account, but you don’t have to share it. Instead of giving out your phone with iMessage, Signal or WhatsApp or sharing your personal Instagram page, you can send a nickname. This made Telegram especially popular among crypto people who prefer to stay semi-anonymous. It is the default communication app for crypto projects, both for their internal ops, connecting to other businesses and community management. They could use no other single service for all of these things instead.

How does Telegram make money? It’s complicated. I’ve seen Durov’s previous project, VK, and have used Telegram since its early days. I think this team is opposed to ads and what it takes to make them work on some philosophical level, and they don’t want to learn. That’s part of the reason Telegram feels nice. There are ads, but they’re pretty unobtrusive since there’s no global feed, and they operate mostly by context with minimal targeting. Telegram also doesn’t have much data on you, so I can’t imagine that CPM is too high. You see these ads in channels, and most of them are for other Telegram channels.

Telegram also has a Premium subscription at $5/month. The subscription removes ads and raises some artificial limits in the interface (the number of chats in a folder); it also gives you a unique check mark and animated emojis. Some people buy it because they like it, and people who use Telegram heavily in their work buy it because it’s just easier than fighting the app. Back in January, Pavel reported 5 million paid subscribers. If we don’t take churn into account, that’s an annual runrate of $300 million. Very impressive, but I’m not sure if it still covers the expenses alone.

Running Telegram takes a lot of money. Partly because they store everything in the cloud, so your mother potentially doesn’t have to figure out how to move her WhatsApp messages from her Android phone to a new iPhone (saying from a personal experience). In 2017, Telegram raised $1.7 billion from investors and announced they would launch their own blockchain TON. The SEC considered their tokens to be unregistered security, so Telegram backed off. They spun off TON, made it open-source, and integrated with it later.

Crypto is a big part of Telegram right now (but you might never see it like other complications). Both Twitter and Instagram canceled their NFT avatars. But in Telegram there’s a crypto wallet, there are web-based mini-apps, and a very basic appstore. The Hamster Kombat game was launched on this platform and gathered 300 million users.

There aren’t too many useful apps yet, but it’s interesting to watch it develop. For years, people were looking for the WeChat of the West. I don’t believe there will be such an app, for many reasons, but Telegram is the closest thing, and it’s curious both use the web stack for these mini-apps.

We don’t know much about the team. Telegram says they have only a few dozen employees working on the core app. The fact that they were able to get to 1/4th of Meta’s userbase is incredibly impressive. Still, we can only rely on their words, as nobody else can share anything about their work. And if you read any stories about Elon’s dreams for X, Telegram might be much further along this path. Telegram Premium probably has more paid users than X Premium. But that’s not all.

Pavel Durov is obviously unhappy with Apple and Google. He’s been at them for years. iOS and Android are critical platforms for Telegram, as the app completely depends on their rules and whims, from content moderation to limitations on what and how they can sell inside the app. You can buy Stars to pay for goods and services inside the Telegram ecosystem, but Apple takes 30%.

But inside this crypto wallet (which isn’t available in the US as many crypto things), you can store USDT, which is the next best thing to a digital dollar. And it’d take a flick of a switch to let users pay for content and services with USDT that they can buy anywhere. This would simultaneously solve Telegram’s problem and push its audience toward crypto.

Telegram is in an unstable equilibrium. They need to perform a careful balancing app with SEC, which already destroyed their original plan and forced them to return most funds, and with Apple, which can destroy their app as they did with Fortnite on iOS. I believe they’re betting they can do this, and their lives will be made easier with the rising pressure on Apple from regulators and the growing adoption of crypto itself. And Telegram might have the best chance to ride both waves.

We’ve all seen these posts about how Twitter fired 80% of its employees, and then nothing happened. Seems it ultimately did affect their revenue:

Internal documents obtained by The New York Times show that, in the second quarter of this year, X earned $114 million in revenue in the United States, a 25 percent decline from the first quarter and a 53 percent decline from the previous year. The company aims to reach $190 million in U.S. revenue during the third quarter, bolstered by advertising associated with the Olympics, football and political campaigns, the documents said — but that target would still set the company’s quarterly earnings at 25 percent less than they were last year.

These people might argue that these two items aren’t correlated but I doubt it. Twitter didn’t have a great advertising engine, so most of its revenue came from generic brand advertising, and companies have brand safety teams for a reason. Elon, a “libertarian capitalist”, still seems to have a problem with the free market when it’s against him.

I think you can believe that Twitter was grossly over-staffed, badly managed and dysfunctional before Elon bought it and ALSO believe that he fired far too many people in the wrong way without fixing anything much. Benedict Evans

I really loved Twitter. I think it changed a lot, mostly for the worse. But because Elon has the worst fans and the worst critics, it’s extremely difficult to find non-polarized opinions on this. Being an eternal centric, I wanted to gather all the details. So stay with me if you’re interested.

The Good

Let’s start with good changes. Twitter was an anemic company. It always seemed like they were terrified to change anything in their product. Probably, because they didn’t understand why it worked and were too terrified to break it (well). And even when they came up with something new, like the predecessor of Twitter Articles, they’d make it available to ten people they knew in San Francisco for two years.

Elon came in and made them ship all those things to the point where the audience largely attributes Community Notes to him. And this is almost reasonable. Community Notes is a fascinating tech product for the social media era. The main problem with fake content is that it spreads much more easily than any rebuttal. Community Notes effectively chase down and clarify the original posts and even notify the users who interacted with them earlier.

There were all these stupid ideas about marking and signing AI content. It’s not like people want to verify if something they’re seeing is true. It’s confirmation bias: they seek something that confirms their pre-existing beliefs. You don’t need AI to create an engaging fake, like that photo of Kamala Harris with Epstein. And the problem with forcing OpenAI or Midjourney to stamp their content is that the next day there will be a Github repo for removing that signature. In addition to a dozen open-source alternative image generators, you can download from anywhere.

Community Notes solve this problem and allow citizens to keep corporations and politicians in check. You can’t just post obvious falsehoods on large accounts anymore (well, it’s still as complicated as everything in this world).

Under Elon, Twitter rolled out a reasonable paid subscription (although I think three tiers are too complicated). They released a bunch of things they built before and even some completely new ones, like X Hiring. Even little things, like finally putting your pinned lists at the top of the feed on the desktop.

After losing most of its advertisers, Twitter has seemingly revamped its ads team. One representative even reached out to my company. However, most ads I see are still from Twitter natives, like content creators promoting their accounts and newsletters. I’m not sure if there’s a lot of money in this.

Twitter now has a powerful and engaging algorithm that should excel at onboarding new users, something they ultimately struggled with before. Many of my friends tried Twitter but didn’t understand the point of it. Twitter would only work if you follow the right people. Now it’s basically automatic and on par with Meta’s tools.

Their goal was obviously to keep users on the website, and in this context, shutting down Revue and suggesting authors use Twitter Articles instead makes sense. The feed on the mobile app also changed; now, if you open a video, you can just swipe to the next one, bringing it closer to TikTok.

This is also why they prohibited alternative clients. I know many hate them for this, but purely from the business perspective, this change made sense. The old Twitter knew they had to do this but preferred the tactic of boiling the frog. The new Twitter comms on this were far from ideal, but they pulled through.

Finally, monetization created obvious incentives to set up and run accounts that posted various content on Twitter. Some of them good, some of them… well, there’s the next section.

The Bad

Twitter botched down their update to the blue check program. Look, this thing was flawed. Twitter employees would sell you a blue check for a couple of grand if you knew where to ask. At the same time, I met companies that struggled with impersonators but couldn’t get verified. The blue check was supposed to mean you’re who you are, but it became a power status symbol instead. Sometimes, when people abuse your product’s limitations, it’s a sign that you need to change it.

However, the logical solution would have been to keep the existing accounts and then offer all who are willing to get a blue check if they subscribe to Twitter and provide their ID for verification. There are KYC-as-a-service companies that do this for fintech startups, so it’s pretty easy. Don’t want to show your documents? Nobody has to have a blue check.

Instead, we got the current mess. Through several iterations, Twitter still got to the point where they awarded the blue check to notable accounts. But the important change is that it can be a DogTurd69, a famous “thought leader”. At first, blue check became coded pro-Elon and even right-wing. Now they don’t mean anything, except that your replies are boosted to the top.

You probably have seen all these companies with gold checks. Don’t believe they paid $12 000 for a year. Twitter gave these to the most notable companies, probably with the expectation that they would start paying next year. We’ll see how many do (or if Twitter itself changes course). Proving the affiliation of specific people (like writers or media reporters) with a badge sounds like a great idea. However, most marketers I talked to believe it’s just too expensive for the little value it provides now.

In addition to losing third-party clients, Twitter has become more closed overall. You can’t browse it without logging in. Often, you can’t even reliably link to it on other platforms, and there are services like vxtwitter whose entire goal is to provide rich previews. In the process, they closed down the API and killed even some of their own tools, like the Card Validator, provided by all social platforms. Finally, outside links are getting suppressed to a degree you shouldn’t even bother posting them. I’m pretty sure they now even track links in the first reply from the author, at least based on my own tests. They did this before Elon, but now doubled down.

Surprisingly, all of this didn’t help beat the bots, which was one of Elon’s key goals as he was getting into it. Just try tweeting the word “metamask,” and you will see a bunch of blue-check accounts trying to scam anyone reading it. It is becoming increasingly difficult to distinguish these bots from regular users since they look just the same, and LLMs are getting very good.

Old Twitter’s algorithm focused on ranking the people you follow with an occasional mixin. The new algorithm is much closer to TikTok. Even if your account is empty, it will create a pretty engaging feed. The problem is this engagement often comes from the worst aspects of human behavior. I’ve never been keen on the idea that social media thrives on this alone, but this particular algorithm might. The Twitter of today is toxic.

Since firing their entire trust and safety team, Twitter has been far more hands-off in moderation. People flocking to Mastodon would refer to the app as “filled with Nazis”, and at first might have been an exaggeration. But now I constantly see racists, incels, and right in my feed, even though I try to have zero interaction with them or their tweets. They hate minorities and ask if women are sentient. I don’t want to be around these people and don’t think free speech is the panacea in this case. The problem is not they’re not banned, but they’re clearly getting amplified, since I don’t follow them.

Everyone who has moderated a forum knows you must do this very diligently if you have even a small audience. If you don’t set the standard and have new people coming all the time, it’ll fall to the lowest common denominator. Is it possible at the scale of hundreds of millions? Well, Twitter could do this before, and not because it banned all conversations. But it certainly wasn’t promoting fringe ideas to everyone’s feeds. Easy to see why brands might not be happy with it.

The new algorithm is very polarized. It gathers the most engaging feeds and leaves everything else to rot. If you’re not getting into the For You feed, 90% of your followers are unlikely to ever see your content. Elon made it very obvious by putting the view counts up. So, authors are incentivized to go to the extremes.

Let’s get to the original thought. Is Twitter operating smoothly? Certain minor things have been broken for months for me. For instance, if I click on any tweet in my bookmark, the page jumps to the top. There are many places across the product that work like this now.

The Ugly

I spent the last weekend without my phone, avoiding all the feeds. When I opened Twitter afterward, it just felt so unnecessary.

There were several reasons I liked the old Twitter. It was the most open social network where you could actually post links and easily copy text. There was an ecosystem of services and bots built on top of it. And it also was the most real-time network for news junkies like myself.

I don’t think I’ve ever befriended someone in real life after following them on Instagram, but with Twitter, it did happen. I think it’s because you can get acquainted with someone much more easily by reading their thoughts instead of looking at their sunset photos.

Well, Twitter was actually a great platform to share photos as well! People were always stingy on likes, but the engagement was comparable to or better than Instagram. And the resolution and quality were much higher. I’d rather share updates about my trip on Twitter than try to jam it into a short description under a gallery of 10 photos that nobody will scroll through.

As for being real-time, I think it still holds this role, and alternatives haven’t caught up (and some probably don’t even want this).

That text-centric nature has always limited Twitter’s appeal to most people. It was never a truly big platform. However, it also self-selected the users who were interested in this—the people I wanted to follow, read, interact with, and potentially learn from.

Twitter is basically a large forum where you own all of your replies. I was an active user of many thematic forums that just died, and every time, it was almost like I lost a part of my identity. Twitter allowed you to keep this identity between “forums” and discussions. But this aspect is only as important as the network itself.

Now, most people who were on Twitter are still there, but many have left, scattering to Mastodon, Threads, BlueSky, and others. Of these, only Threads seem like a reliable alternative for me. Mastodon by itself still feels like something made by geeks and for geeks, and the lack of any algorithms and recommendations really limits your community. I’m not talking about “growing your followers”, it’s hard to simply find interesting people to follow. But, if Meta makes Threads fully federated, there will be a unified community where you can follow people from celebrities and your friends to geeky developers.

We will see what happens to Twitter, but I doubt it will have a big turnaround.

I guess growing up means watching the things you love die.

The Digital Markets Act is a bad law. I can understand the idea that led to it. Digital tech companies have a lot of power, and existing regulations don’t adequately cover them. But instead of targeting specific examples of overreach, the EU tried to come up with a universal and far-reaching framework that freezes product development for the largest incumbents and prohibits vertical integration.

Ben Thompson explained this well:

The DMA, though, is much more far-reaching than that; the excerpt above rightly identifies the interoperability requirements as the more meaningful factor affecting Apple’s decision. Consider iPhone Mirroring: given that the iPhone is a designated gatekeeper platform, is Apple allowed to enable iPhone screen-sharing with a Mac but not with Windows or Linux? Or SharePlay Screen Sharing: can Apple allow you to jointly view apps via Facetime, but not other VoIP services? Or the big one, Apple Intelligence: can the semantic index be available to Siri, but not other voice assistants?

Just this week, the European Commission tried to allege that Twitter’s new blue check system violates the DMA. Yes, I believe the transition was stupid. Twitter should have kept the old verified accounts (or at least most of them) and offered new users a blue check through a subscription and ID verification. But it’s baffling for a government structure to intervene. If Twitter makes bad decisions, it can fail on its own. It’s not critical infrastructure by any means.

I also think Apple put itself in this position by being so stubborn. They got themselves to a point where the customer experience on iOS is consciously diminished. Who benefits from the fact users can’t sign up for Netflix or Spotify on their iPhone? Or buy Kindle books? This should have been a signal their approach doesn’t work. Now, they will have to contend with different regulations across the world, as we’re already seeing it developing in the EU, South Korea, Japan, and even the US. This gets us closer to the fragmented internet.

A big part of the problem is that the DMA doesn’t provide firm guidelines. It’s like its authors wink at Apple and other “gatekeepers” but don’t want to say what they really want out loud, even though the EU had the capacity and the experience to require specific adjustments.

So, what could they have done instead of DMA?

First, Apple should allow apps to lead users to a webpage where they can buy digital goods or subscriptions directly with their credit card. This option will always carry more friction, so many existing apps will likely stay on Apple’s payment system. It’s a balance of earning more due to smaller fees and converting fewer customers. If you have a mobile game, you might feel compelled to keep the player immersed (unless you’re Fortnite or Roblox). Since they will also have to compete with alternative payment methods directly, there’s a good chance Apple will beef up their own system and make the fees more competitive. Ultimately, I love that I can see all my subscriptions on a single page and delete them. But under the current regime, it will never be a comprehensive list.

Second, Apple should enable sideloading for applications. There are a lot of advantages to the AppStore model of vetted apps. But there are problems. Just recently, Apple deleted multiple VPN apps from the Russian AppStore. Why they feel obligated to submit to the laws of the country, where they have no direct sales, puzzles me. Still, without sideloading, their users are trapped. By allowing sideloading, Apple can shed this responsibility so that their decisions and censorships would have minimal effect. I don’t understand who actually needs entire external stores. Hide it under a checkbox with scary wording; as long as it’s there, it’s OK.

Everything else is honestly pretty minor. Yes, you can demand Apple to open its NFC chip to third parties. This will allow the likes of Wise and PayPal to save some money on Apple Pay and work directly instead. However, the two major changes I outlined earlier will be enough to act as a steam release valve. Both companies that can’t work under the AppStore model (Spotify or Amazon Kindle) and companies that really don’t want to and can create strong relationships with customers directly (from Tinder to Basecamp) will stop criticizing Apple almost immediately.

Now, you might say that DMA regulates the digital markets in their entirety and covers many other gatekeepers. But it was already carefully designed to affect all the companies they wanted. If you want specific changes to be made, it’d be better to require specific things.

You might say that Apple is a private company and the government shouldn’t be involved at all. I disagree. Apple built one of the most important computing platforms in the world—your phone is the actual personal computer. They’ve benefitted tremendously from this and will continue earning billions from it. But it makes sense for society to intervene and limit negative externalities.

The internet is probably my favorite invention in the whole world.

It’s hard to compare it with penicillin or electricity, of course, but I still remember the awe I experienced when I realized I could talk to some strangers right on my computer or look up anything I wanted. Well, mostly texts since music and videos weren’t exactly available on my 45.2 kbps dial-up connection.

This was around 2005. The internet was paid by the hour. One month my parents got such a huge bill they realized I spent around 100 hours online, and believe me, it was expensive.

They didn’t think much of this new toy but believed it’d change everything. I was reading the news, learning how to code and design, and how to shoot better photos and edit them.

I didn’t have any money, but thankfully, after you got online, everything was practically free back then, supported either by volunteers or online advertising (and piracy, yes). Everyone was on the same Internet, especially if you knew any really popular languages, such as English, Spanish, or Russian. There were no borders, and if any laws were applied at all, they’d be American, so quite liberal regarding what you can and can’t write or say (but click a checkbox that you’re over 13).

Since then, the internet has reshaped economies and enabled entirely new ways of work. My entire career was built through the Internet. I learned everything I needed to land my first job and have worked remotely since then. I work in communications and the internet is where most people consume the news, without even thinking about it. It’s just there.

The internet was only allowed to emerge and develop as it did because nobody was paying attention to this silly little thing. Societies are slow to react, so when they did, it was already too late. If you think about it, many aspects are truly crazy. Anyone can publish anything, Google copies content from all websites, and entire industries have been decimated by the shift online.

As the government woke up, it all started crumbling. First came the authoritarians, who saw a direct threat to their rule since the internet (especially mobile) allowed opposition leaders and protesters to organize or spread alternative views. Because it was a new medium, they would often decide to create a new watchdog or set up innocuous laws they could leverage to erase the content they didn’t want.

Then, to my enormous surprise, the developed countries started slowly adopting practically the same laws as China or Russia (which is rarely a good thing).

First was the economic regulation, as they saw the giant digital economy growing in front of their eyes and adopted things like the “digital” VAT that must collected from online platforms even for cross-border sales (which seems to conflict with the very concept of VAT but OK). I call this digital protectionism.

Then they came for content. Certain governments granted themselves the ability to quickly block websites without ever going to court. Required tech platforms to “land” in their country or region, set up local subsidiaries, and hire people (who, in certain cases, would be just hostages to the regime). Demanded “data retention, so ISPs and online platforms store copies of their customs data.

For some reason, the UK and Australia were at the forefront of offensive laws leading to wide-scale government surveillance (under a veil of protecting children, of course), as both tried to ban end-to-end encryption. Now the entire European Union is trying to do the same, the decision has been delayed but the possibility is out there. In 2023, Signal confirmed they’d leave the UK if they passed this law.

Australia (again), Canada, Spain and France introduced the link tax, a weirdly worded subsidy/tax combo that takes money from Google and Meta and gives it to the largest local news organizations (but not any other organization or small newsrooms). Benedict Evans has a great post explaining why it’s a terrible idea.

But if you do accept the novel theory that links being free for 25 years is a market failure, there’s a further breach of basic logic: if all links have value, why should only newspapers be paid? If all links were paid, then newspapers’ share would be a pittance.

Of course, the European Union is famous for its initiatives, starting with GDPR and now going all-in with DMA. GDPR created a requirement to store “personal data” for EU residents, which led to a noticeable number of websites blocking access from the Union.

Sidenote: Apple and the EU are fighting each other but turned out to be surprising allies against targeted advertising, which funds the free internet for hundreds of million people

Digital Markets Act (DMA) was the latest huge milestone. We still don’t know where it will ultimately lead, but Meta and Apple have already had to adjust their products specifically for the EU. Threads blocked access to all accounts created from the EU for many months. Now, we have learned that Apple won’t release its Apple Intelligence features in the EU.

All these regulations bring us closer to a fragmented internet that will be nothing but a shadow of what we had.

China famously has its own intranet cut off from the rest of the world by the Great Firewall. Russia is building similar capabilities after being sanctioned and ostracized to the point most companies leave its market voluntarily.

The United Internet of America is collapsing. We’re driving head-on into a world where every major country and major economic block carved out its own fragment of the internet with different rules, regulations, and local champions. You might say this is the natural order of things that works in all other industries, and big tech companies will adapt. And you’d be right. But the Internet didn’t work like this, and I believe it was better for the users.

***

What does it mean for us, exactly?

There’s a good chance that, with time, you won’t have access to that popular app or website people are raving about. Both as a consumer or a business, you’d have to use local tools that aren’t necessarily better (most likely worse), especially if the regulations prevent them from having to compete on the world stage.

It’s likely you will find it extremely challenging to break the laws here. Threads’s ban of Europeans from accessing the app was quite bulletproof, and even the VPN couldn’t help. This is just like in the physical world, where a ban or tariffs on certain goods can make it extremely challenging to buy them.

Arc is an alternative Chrome-based browser with a unique interface. It’s become so popular that The Verge reports on pretty minor updates. I switched to it quite early and haven’t looked back.

Browsers are the most important apps we have on our computers right now. Whether you like it or not, the application layer has shifted to the web. But as I wrote back in 2020, browsers haven’t caught up.

Look at Google Chrome right now. It’s essentially the same browser that launched in 2008. The only recent major update was Tab Groups, which happened in 2020 and they are still subpar. For one, groups aren’t persistent and I can’t understand the logic behind this. You create a “Work” group, open a bunch of tabs, close them… and the group disappears. Want to “work” again? Create it, name it, and choose the color. Every time.

Google Chrome was invented for web surfing: it lets you open a bunch of almost static web pages and read their content. Now, we have entire apps like Figma, Linear, and Spotify running in web browsers, and Chrome’s interface hasn’t been adapted at all. All Chrome can offer is pinning them as tabs. No considerations have been made to help people manage their tabs.

The development of Chrome’s interface was also remarkably slow and inconsistent. Chrome on Android would get Reading List, iOS wouldn’t, and it’d then take them years to bring this to the desktop and set up syncing. I’ve just tried Chrome’s current Read Mode on desktop, and it can only show you text side-by-side with a page for some reason.

You can also turn PWAs into “desktop” apps. But this option is a bit hidden, and most of Google’s own apps, except Photos and Maps, don’t support it. Where are PWAs for Gmail, Calendar, and Docs? I suppose Google would prefer you to use Chrome, where you’re always a click away from Google Search and its ads.

Now, Chrome isn’t the only browser. But Firefox’s UI is not too different. Safari has its own issues, but at least it offers a cohesive experience. You get a complete browser out of the box, with a fantastic Reading Mode (still best in class), a synced Reading List, and much better Tab Groups. Now, because Apple isn’t great at cloud, the two previous features just stopped working properly for me for a few major MacOS versions, but right now, they do work. Plus, browser extensions are quite important to me, and even though Safari adopted the same standard as Chrome and Firefox, developers still need to do separate work to distribute them.

But at its core, Safari can only offer pinned tabs (which are incredibly small squares). It can also suddenly kill your Google Meet tab with an active call because it “consumes too many resources” (this happened to me).

I often hear the claim that companies don’t need to invest in their iOS browsers because they still have to use Safari’s Webkit. I don’t buy it at all. For one, Brave has been pretty good at creating a nice experience across all platforms, as I outlined in a different post.

What Makes Arc Different

Arc became the first credible and ambitious attempt to reinvent web browsing that was actually able to get traction. It wasn’t the only one or the first one. I listed some options here. Some of them died, some dragged on.

Arc’s most important part is its sidebar. And vertical tabs! At first, you might feel like you’re losing too much space. But most websites right now don’t take as much width anyway, except for the likes of Webflow or Figma. Everything else looks fine, even on a 13’ MacBook. In return, you can keep lots of tabs open and still see most of the titles. Out of major browsers, only Edge and Brave offer vertical tabs now.

In the sidebar, you get three distinctive groups of tabs.

The first is for the favorites, but I mostly use it for applications: email, calendar, Spotify, Notion, Readwise Reader, Google Docs, and anything I use constantly. When you close such a tab, its instance is terminated, but the icon stays there in the same position as a bookmark. Mouse targets are generous and don’t take up too much space.

Next go the bookmarks. Anything you’d like to have handy, but now you get the titles. And you can put them in folders. One of the adoption hurdles for Arc is that it doesn’t get traditional separate Bookmarks, so I put mine right here. The persistence is also there.

Only then do you get traditional tabs that you lose when you close them. Arc treats them so harshly that they are automatically closed after some time, which can’t be more than a month (a bit excessive if you ask me).

On top of this, Arc offers Spaces, which are essentially tab groups. You can create multiple spaces that would share your Favorites but have dedicated Bookmarks and Tabs. I have one for primary browsing, one for sales, one for reading the media and one for software development.

There’s a joke that there are two groups of people. The first has no more than 5 tabs, the second has no less than two hundred. Arc is perfect for both, but especially for tab hoarders.

On top of this, Arc offers multiple quality-of-life improvements:

1. You can switch between tabs in a loop with a quick preview, just like you switch between apps with Cmd-Tab (Alt-Tab).
2. The autocomplete in the new tab popover prioritizes currently opened tabs, so you can switch by typing part of their name.
3. There’s a shortcut to copy the current URL, which I probably use 20 times a day.
4. You can click a button and sort all of your opened tabs by categories, creating order out of chaos.
5. You get little custom benefits, like picture-in-picture for Google Meet, always available media controls, countdowns to your next meeting with Google or Notion Calendar, etc.

One criticism I sometimes hear about Arc is that it’s based on Chromium. Nobody cares. Chromium is the most advanced browsing engine, powering most apps people use. And it’s definitely not IE6, especially not the one I remember. IE6 was a bad browser that held the entire web development industry hostage. Chrome was the first to implement many standards. It won. Give up.

Then, we have the obvious challenge of monetization. Safari has Apple, Chrome has Google, Mozilla has… erm, Google, and Brave has their own ads on the blockchain. Arc is built by the Browsing Company, a venture-backed startup that needs to grow and either IPO (seems ambitious) or sell. The only way I see this happening is a subscription for advanced features, probably with a team collaboration angle.

But I’ve worked with a company utilizing this exact path and can say it’s not easy. Arc indeed got a great PMF score from me, but I’m wondering if they will be able to offer such compelling features enough that people would pay.

I have to say, I don't know what I'd use if @arcinternet disappeared tomorrow

Would be extremely disappointed

— Yury Molodtsov ⚡️ (@y_molodtsov) October 24, 2023

***

I haven’t been this excited about the browser in a long time. But at the end of the day, we have just one outlier, and I wish Arc all the best.

The European Union and Apple are locked in a fight nobody wants to cede. First was the Digital Markets Act (DMA), which made Apple create an entirely new subset of rules for iOS and the AppStore. And yesterday, Apple was fined almost $2 billion for their anti-steering provision affecting companies like Spotify and Amazon’s Kindle.

All digital purchases on the iPhone must go through In-App Purchases, which involve a 30% commission to Apple (goes down to 15% for long-term subscriptions). While certain services like Netflix and Spotify (classified as “reader apps”) can ditch IAP, the resulting user experience gets weird. You cannot sign up if you download such an app as a new user. And the AppStore rules prohibit developers from even suggesting to their users that they should go to the website instead and start their subscription there.

This is probably one of the most egregious examples of how Apple’s strategy leads to suboptimal experiences on their devices. Apple feels entitled to this tax, which you can clearly see in their statement.

Despite that success, and the App Store’s role in making it possible, Spotify pays Apple nothing. That’s because Spotify — like many developers on the App Store — made a choice. Instead of selling subscriptions in their app, they sell them on their website. And Apple doesn’t collect a commission on those purchases.

Spotify didn’t really make a choice because they didn’t have one. Some companies, like mobile game development studios, can shed 30% of their income because their marginal costs are more or less zero. This isn’t the case for Spotify since they have to give back most of their revenue (which amounts to 74%) to copyright owners (i.e. labels and artists). The same is happening with Books. You can’t buy books on Kindle or Audible on your iPhone because then Amazon would have to pay 30% to Apple. And for any book, there’s already a pre-defined distribution between the publisher, the author, and the seller. Amazon would be losing money on every book then.

One of the main problems with the AppStore fees is the rules don’t differentiate between goods with marginal costs and those without them.

What makes Apple look particularly bad and should have motivated this case is the fact they operate competing services in the face of Apple Music and Books. And Apple doesn’t have to pay themselves 30%. So, they enjoy better economics and a better user experience since you can sign up right away. One could build a legal case that Apple is extending its monopoly power from one industry to another, which is prohibited in most major jurisdictions.

The Digital Markets Act has many provisions. Apple carefully crafted such a response that most developers would not even dare to use the updated EU regime. They did open certain things, like access to the NFC chip, and later, they removed some of the limitations. However, they still want to control the overall experience and charge money.

In my opinion, this is indeed a case of an unstoppable force facing an immovable object. And it gets ugly. The European Union shouldn’t make tech product decisions. What is even worse is that most EU regulations are written in a particular way. Like with the infamous GDPR, businesses have to guess what exactly they need to do for the government and regulators to assess whether it’s compliant. It’s also noticeable that the EU’s fines are getting arbitrary by this point. Writing a big number because Apple is a wealthy company and simply saying “the tech giant’s terms were “unfair trading conditions” without providing clear guidance on what exactly you want them to change isn’t a great way to regulate.

And Apple should see a writing on the wall by now. The longer they hold on so profoundly to their entrenched position, the worse the outcome will be for themselves and their users. The company had an easy way out: let all developers share a link to their website instead (or in addition to) using an IAP. Lots of developers would have likely stayed with the IAP just for the sake of conversions alone. Most (over 70%) of AppStore revenue is coming from games. A mobile game likely won’t suddenly ask you to go online, set up an account and buy some gems there because it will be afraid of losing a potential sale.

Of course, some companies might prefer this approach anyway, especially if they have the brand power to pull it off. Epic Games’s Fortnite is probably the best example (if it was allowed back in the AppStore). Tinder would also use this, and dating apps surely drive a significant part of AppStore revenue as well. But most games and apps aren’t like that. Such an approach would make things easy for businesses selling goods with notable marginal costs. And most importantly, it’d provide a better user experience for the users.

Apple can talk a lot about how their system provides safety, but they have no problem with you giving your credit card to Uber right in the app. And what’s actually the difference?

The current system creates perverse incentives for Apple and forces the company known for building great products to provide subpar experience. Even Apple might be better if they get off the needle of AppStore fees. When Tim Cook provides an update on the “Services” revenue growth, he wants you to think about AppleTV+ and Apple Music. Yet most of this revenue category is provided by just two things: AppStore fees and Google’s default search payment. And none of this makes things better for actual customers.

I support sideloading, and I don’t believe Apple should ultimately decide what software I can run on my own device. This limits innovation and puts them in a position where they can be forced to run censorships for authoritarian governments, just like they had to remove VPN apps from the Chinese AppStore.

But I also understand that one of the reasons iOS has a thriving app ecosystem is the lack of alternatives. If you have an Android device, there are countless websites and forums filled with cracked apps you can install immediately.

Instead, Apple might see governments worldwide legislating their product experience, and the result will likely be far worse.

Most politicians are populists, which is the natural result of the basic rule of communications: the larger your audience is, the simpler your message should be. And with politicians specifically, there’s a straightforward line from the size of their following to their actual power.

Politicians are incentivized to find a slogan that resonates with people and run with it regardless of their personal opinion on the matter. As an off-topic, I doubt certain US Republicans have a significant issue with LGBT themselves, but they adopt the message that works for people who can vote them in (or out), however cynical this might sound.

The same is happening to AI. And it’s accelerated by AI becoming a major trend covered by practically every media publication. Not only most have dedicated AI writers, but there are entire outlets (e.g. VentureBeat) that have largely shifted their focus to AI.

And sometimes, they’re just broadcasting the opinion of crazy people like Elizier Yudkowskiy, who isn’t an expert in any modern area of AI. He’s a self-proclaimed expert on “AI safety”, which seemingly means regurgitating ideas from sci-fi novels with specific outlooks to convince everyone we’re doomed. Balanced opinions are complicated, and slogans are easy, which is why both doomers and proponents of AI (i.e. “e/acc” movement) dominate the discourse even though all practical discussions are somewhere around the middle (and should be happening in a completely different plane).

Whether Biden or Sunak actually believe AI is dangerous is irrelevant (although Biden might have indeed been spooked by an underperforming action movie). If politicians think doing something publicly will help them score, they’d do this. OpenAI, Anthropic, Google, and all others will be the first to tell you they support strict regulation of AI precisely because this will help cement their hard-earned position at the forefront of the industry.

There are copyright and safety questions, but all practical concerns fall into a completely different bucket than what pundits say. Generative models won’t drop a nuke on Los Angeles (“The Creator” is a great movie, though), nor would they automate everyone’s job. But they raise the same old problem of algorithm-based decision-making and pose a new problem that doesn’t map well on our current copyright laws (both in the letter and the spirit).

Safety

Algorithms have been used to make decisions affecting people’s lives for decades. Sometimes, they’re basic; sometimes, they’re pretty advanced. Credit scores, mortgage assessments, and background checks aren’t manual but run by algorithms.

The actual problem worth discussing is the potential of inherent bias AI might ingest from the training data and negatively affect someone’s life while obfuscating the process.

Old OCR algorithms used to recognize printed texts were highly complicated and cumbersome and relied on computer vision and advanced math. We use AI to solve specific complex problems differently and abstract away the solution when it’s too complicated to build. We use data to train more-or-less generic networks and then run primitive arithmetical operations many times, ultimately providing good results we wouldn’t achieve through classic algorithms.

Because of this, most modern AI networks are essentially black boxes. You know the input and the output, but you don’t see how the decision was made, and it’s challenging to dissect the model while it’s “thinking” to understand why.

So, if a financial institution uses AI for credit checks and fraud assessments, one could imagine a scenario where flawed training data would lead it to discriminate by race or other parameters. And since it’s a black box, only long-term statistical observations would allow us to see the problem, though most likely, the people in charge would claim the discrepancies are accidental.

This problem is more tedious but is way more accurate. AI won’t do it because it’s terrible. It’s emotionless and doesn’t have a goal. This is a popular trope in sci-fi, but I don’t believe Artificial General Intelligence could be a cold, emotionless beast. Emotions give purpose, which is why ChatGPT doesn’t think anything until you tell it to. It’s simply not there yet. What we have now is way more basic.

Copyright

Neither large language models nor image generative networks fit well into our copyright frameworks. The line between plagiarism and inspiration was always blurry. All human artists, including writers and painters, learn by observing the works that came before them and then produce something of their own.

Vanilla Ice used a guitar riff originally composed by Queen and David Bowie. He then claimed he added a beat between the notes and that the original melody was different from his. It was ultimately decided that this was not the case. Ultimately, he had to pay up, but only because it was too transparent. Musicians get inspired by others’ music all the time.

Generative networks aren’t too different from humans in this regard. But since this is machine intelligence, they can do this much better, at scale, and then provide everyone with a quick and easy way to create pieces that might copy someone’s material too directly.

Writers like George R.R. Martin claim that OpenAI’s GPT4 was trained on their material, which could very well be correct. Before, we had a simple delineation. A writer consumes existing material and hopefully uses it to write their own Song of Ice and Fire. Nobody just starts writing one day without reading a single word imagined by others.

If you use others’ material too directly, we call it fan fiction and mainly don’t pursue it as long as you aren’t trying to sell it. Then it’s illegal.

But the ultimate use case of GPT4 is not someone using it to retell some author’s book but to write new stories under a user’s command. Yet the reason it’s capable of doing so is all the writing that was consumed during the training stage. So, does it break the copyright?

Generative networks are stuck between humans and plain computers, and until we develop new frameworks and the common law system in the US and the UK create precedents, nobody is sure what to do.

***

Now, the emergence of generative networks might have already affected the job market globally. As usually happens with technology, the biggest threat is to the low-level contractors, such as people who accept $5 orders on Upwork since they’re based in a cheaper (and poorer) country.

The scandal around the intro to Marvel’s Secret Invasion is a perfect illustration. The VFX people working on it used purposefully poorly-made AI-generated images due to their Uncanny Value quality relevant to the story. Half of the Internet reacted as if the producers fired everyone and went to write the prompts themselves (and manually did all the VFX on top to make the animation, I guess).

In fact, actual commercial writers, designers, and other content creators were the first to start using tools like Stable Diffusion and MidJourney in their work. A game development company I know uses these tools to accelerate its image pipeline. This specifically helps designers spend less time on mundane things. For them, AI hasn’t replaced a single worker; it simply allows people to produce more content and, most importantly, elevate the floor of the overall quality.

Sometimes, it simply acts as a reference. I’ve heard a music artist explaining that he can now use MidJourney to explain what he wants for title art from his designer. And I myself use it to produce covers for my Spotify mixtapes. Again, technology always raises the floor, and MidJourney, along with other tools, allows people who can’t draw anything to turn their fantasies into reality (well, a real image).

You probably do need to think about AI when making life-long choices. My aunt recently told me her friend’s daughter had just enrolled in a university as a translator. Google Translate was shit ten years ago, got pretty good after this, and then was blown by GPT4. Does this mean we will not need translators soon? No, but there surely will be fewer jobs, and they will primarily focus on high-value tasks, such as adaptations of movies, popular books, etc. Ultimately, this means fewer translators and less money for them.

What happens when AI goes upmarket and captures way more? We’ll probably have to deal with this than by finding a balance. On the one hand, if we don’t automate things, the economy won’t grow, and everyone who is poor stays poor. On the other, as a society, we sometimes make decisions that might be suboptimal for the greater good. We’ll see what happens here, but even if one country severely regulates AI out of existence, others might embrace it.

Technology has replaced thousands of professions, all while creating new ones. We no longer have human “calculators” or telephone operators. Caution and the desire to preserve the living standards for some people is welcome, but imagine if we used the same approach a hundred years ago. We’d probably still have actual humans running elevators.

Blogging became popular in the late 1990s and early 2000s, with the rise of platforms like Blogger and WordPress, which made it relatively easy for people to run their blogs.

Then came social media platforms and managed services like Medium that ultimately gathered hundreds of millions of people who’d never bothered to set up their own blogs. Traditional blogs have been almost abandoned during this time.

And even though we’ve seen the resurgence of email newsletters and Substacks, they operate a bit differently: newsletters are usually more professionalized, and authors stick to one or two areas and generally write to make money out of it.

It’s not like all blogs died. I’m subscribed to a few dozen of them via RSS. But it’s certainly not what the majority of people are doing. Social media provided a straightforward way of broadcasting to a big audience, but as intermediaries, they could change the rules of this game (many media orgs felt the result).

The same is happening to Twitter now. Before Elon, its algorithmic feed was primarily focused on people you follow, bringing their most essential tweets and conversations higher while injecting a few “heroes of the day” for your amusement. I liked this feed and preferred it to the chronological one. The current “For You” feed is more like a textual TikTok, gathering anything engaging from your language’s segment of Twitter.

It’s indeed very engaging. Sometimes, I accidentally switch and notice myself scrolling through posts from unfamiliar people and all the emotions they produce in me. At the same time, I realize posts from the people I follow don’t always make it into my feed. The “Views” counter Elon introduced showcases this even better.

Unless you’re already big or willing to commit to this game of engagement farming fully, anything you write on Twitter there will likely stay irrelevant.

And here we come back to blogging. I started writing here four years ago, and at the beginning, it felt like shouting into the void. Coming up with content ideas was difficult. But everything I wrote was compounding. I can never predict which posts will do well, but in the long run, a few will attract attention. And sometimes, specific things and concepts I wrote about previously suddenly become popular, and people go to read up on them. Even coming up with ideas has become much easier. There are at least 10 posts I’m yet to write.

Compounding is the name of the game. Anything you write is searchable instead of being drowned or pushed outside everyone’s feeds. At this point, Google and Direct traffic vastly outnumber any Twitter engagement for me. Nobody cares about your old tweets (unless they want to cancel you), yet people regularly find the posts I wrote a few years ago.

There’s a concept called “Create Once, Publish Everywhere” and having your own blog helps, even though most centralized platforms suppress posts with outside links. When you lead people to your domain and provide them with some ways to follow you (email, RSS, and social), the conversion rates might suffer, but any of these followers is worth so much more. This was the key reason for newsletters’ popularity – not because everyone loved email, but the fact you could reach people directly without an algorithm between you (well, Gmail anti-spam and categories are a bit of an algorithm).

If you want to set up your blog, I have a guide here. Ultimately, it doesn’t matter what platform you use. What matters is that you stick with the practice and continue writing no matter what.

Superhuman is a productivity-focused email app initially designed for founders and executives that has massively expanded its focus since then. Its “the fastest email experience ever made” tagline is well-deserved. It’s also known as an email app that charges $30 a month and amassed over 200,000 on its waitlist, which made many people believe it’s obnoxious.

I’ve been a user of Superhuman since its early days (in fact, we invested in the company). I think it’s indispensable for people who deal with email a lot. This is enabled by many different things, forming a unique software experience I haven’t seen anywhere else.

Superhuman is fast. It’s a web app, but it’s the fastest web app I have ever seen. I mostly use it as a pinned tab, and there’s an Electron wrapper if you need one. All interactions are rapid yet smooth. But it goes deeper. Superhuman breaks the age-old concept of GUI and consciously eliminates UI elements like menus and buttons unless they’re critical. Instead, it pushes you to learn keyboard shortcuts and operate like a power user. In case you forgot or if you need to access a rare, more convoluted function, there’s a powerful Cmd-K Command Bar that has become a standard for powerful productivity-focused apps. And Superhuman’s natural language processing is genuinely one of the best.

As a result, the app is clean and streamlined. It’s almost impossible to miss an important email. One of my friends working in VC said this was the reason he’s paying for Superhuman, simply because missing a particular email could become a major problem for them one day. And if you need to deal with something later, it has fantastic Remind Me functionality.

Superhuman has a self-explanatory Split Inbox feature, allowing you to filter all emails into dedicated folders. If you’re asking how this is different from Gmail labels, you should understand that a) all splits are right in your face at all times, and b) you can cycle between them with one keystroke. It’s more similar to Gmail categories like Forums or Social, except you can create them yourself and as many as you need.

One thing many apps miss the mark with is their mobile apps. Superhuman doesn’t. Its iOS and Android apps are, of course, different, but they have all the necessary functionality for you to do things on the go.

***

Everything I described above can be helpful for anyone dealing with email. Now, one of the reasons I’m willing to pay that much is specifically because I use Superhuman in my work when pitching stories and announcements to journalists. It was just as relevant when I was talking to founders, other investors, and Limited Partners when I worked in VC. And it could come in handy for many other areas.

Superhuman has Read Status tracking. It even started a controversy a few years ago, which, in my opinion, was way overblown (but it’s good they removed IP and geography tracking from it). Now, one aspect of pitching reporters is that they get a ton of emails and rarely reply. If I see that they read my pitch and ignored it, I don’t have to follow up. Of course, there is similar functionality in other email apps (which aren’t free as well) and Gmail extensions like Mailtrack. But these extensions never integrate well. This was one of the original ideas behind Superhuman. Its founders built and sold Rapportive, a Gmail extension connecting with LinkedIn, and realized that many people used half a dozen of similar extensions at any time. Superhuman is a brand-new app that comes with these functions baked in.

And then there are little things they never even boast about. For instance, if you worked with Gmail a lot, you know its text formatting is a mess. You write something in Google Docs, paste it into Gmail, it looks completely natural, and you send it only for the receiver to get an email with a giant font you never intended (sometimes it only breaks in mobile Gmail which is even worse). Cleaning all the formatting could help, but what if you need the text to use bold/italic and links? When you paste text into Superhuman, it preserves that and removes font and size inconsistencies. The entire app is filled with these wonderful little things.

***

Is Superhuman perfect? No. For me, there are still issues. The primary one is the pace of development. The team shifted focus to money-making audiences of enterprise clients, sales teams, and Outlook users; as a result, the app has barely seen any improvements for the single-player mode in years, with AI being the only major update.

There are things I’d love to see in Superhuman. One is more advanced attachments search and management functionality, similar to what you can find in Outlook and Hey. I often had issues finding “that lost version of the agreement signed by both sides”. Certain aspects of the email workflow remain the same as they were a decade ago and I don’t believe they can’t be improved.

The main reason Superhuman is able to charge so much for an email app is that there isn’t any real competition. Mac users can try out Mimestream, a lovely Gmail client with a native app available on a much cheaper subscription. But it has no iOS app and probably won’t have one any time soon. Apple’s Mail.app is conceptually outdated and a bit slow since it relies on IMAP. Gmail’s iOS app is barebones. Outlook for iOS is actually nice, but it had the same bugs driving me crazy for several years, and the desktop version is much more convoluted. Spark was OK but wasn’t that much better, and I lost my outgoing emails on too many occasions for me to trust it again. Twobird is an interesting newer entrant, but I’ve seen so many services emerge and die over the years I don’t have much confidence yet.

Frankly, there isn’t a single iOS-capable email app I could recommend to someone. And this is pathetic.

The truth of the matter is building a capable email app is incredibly difficult. Creating a sustainable business on top of one is even more challenging. At least Superhuman seems to be doing both of these things together.

If you want to try it, here’s my referral link.

There’s a widespread notion that Uber is getting more and more expensive. You can find multiple series of articles produced on this topic each year and entire SEO-optimized webpages for this particular query.

Some people even believe that Uber is worse than conventional taxis right now. Uber’s only advantage to them was the price, and since those prices have soared, they see no value. In fact, these people believe Uber could only grow because of its initial low prices, artificially attracting customers and taking market share, only to leverage its semi-monopoly position later.

I think they completely misunderstand what Uber actually is and what makes it an excellent service.

Uber isn’t the only ride-sharing company even though it was the first. Depending on the country, you could switch its name with Lyft, Bolt, Grab or something else. I’d keep using “Uber” throughout this story.

***

While Uber and its clones dominate most markets, some have been left out due to insufficient volume (imagine a remote island) or government intervention. Bulgaria prohibited global ride-sharing apps. Barcelona (Spain) instituted an artificial 15-minute limit to force you to hail a conventional taxi on the street.

Every time I visit such a place, I realize what Uber is and how difficult it was before.

First, you have to find a cab. You likely need to be in a popular city area to do this, or you will have to wait a long time (or walk to a busier street). Maybe you can order via phone, although you’d need to know the language and have roaming enabled on your carrier plan.

Sometimes, there are entire lines of yellow cabs. When I went to Madeira, I saw a stereotypical line of two-decade-old Mercedes. They spend most of the day waiting, so when you take one, your fare also includes this empty time. You don’t know if they’ll have a terminal to pay with a card. The A/C might be busted, the car can reek of cigarettes or something similar, and if there’s a line, all the drivers expect you to take the first car, regardless of its quality.

A conventional taxi operates very similarly to a tourist-trap restaurant. There’s no expectation you will visit again. They probably don’t want to be completely terrible, but they also have zero desire to make it comfortable for you or invest in this beat-up Mercedes they bought for some unknown reason.

Uber changes this equation by becoming the arbiter. By leveraging their power as an aggregator, they can assign ratings to drivers and riders and execute control over them. Each particular passenger is no more likely to end up in the same car. But each passenger is a rider provided by Uber, something they could give or take back if a particular car isn’t great for their business.

And suddenly, fares go down a bit because Uber cares a lot about supply utilization. Drivers rush to buy a hybrid Toyota Prius with its great mileage (it also has lots of space in the back). And they never just stand in a line waiting for the next customer to walk to them.

The taxi experience before Uber and its clones was miserable. Did they subsidize the rides initially? Yes, probably. But its critics should at least stick to their line of reasoning. Because for many years, they were sure Uber would never be profitable until it suddenly was.

I don’t use Uber because it’s cheap. I’m OK if they raise prices to ensure drivers and the company make enough money. I use Uber because the experience is generally better, and if something happens, I have the company to deal with it.

***

Is Uber perfect? Probably not, as any big company. But we must accept reality and deal with externalities while considering consumer benefit. Do you seriously think a corrupt system artificially limiting taxi licenses to the point that they cost a million dollars in New York City is great? I find it laughable.

Basecamp and DHH specifically know how to trigger people and use this all the time to drive awareness without spending a dime. They take stuff that’s happening anyway, like when they had issues with Apple because of Hey, and sometimes they write something edgy out of the blue.

They juggled the names of Basecamp and 37signals, with the latter being the current one, but most people know them as Basecamp, so this is what I’m using here.

Basecamp is known for building very opinionated and slightly old-fashioned products as a remote and bootstrapped company. There are many similar companies out there, but Basecamp is certainly the most recognizable. Why?

Several reasons. First, they started back in 2004 (the year they switched to building products) and have survived to this day, which is an impressive feat for a tech startup and helped generate awareness along their journey. Second, they built open-source products, making them popular in the community, and an entire tech stack in the form of Ruby on Rails, which once was a very reasonable choice for web development. Now, Ruby’s share is about 5%. People used to it might continue working with RoR, but practically nobody is getting into web development through it.

Most importantly, they knew the role of marketing and communications early on. They showcased their uniqueness. Wrote multiple books on the topics they were into. Ran a blog. Hosted a podcast. Ultimately, advertised their company as a great place to work. Writing a book in the 21st century won’t make you rich unless it’s a bestseller, but it creates status. 

You see, Basecamp isn’t unique in being a bootstrapped company. The entire Indie Hacker community wants to get the same – become a millionaire and live off the income from your product, ideally without having any investors to report to. The challenge is it’s incredibly difficult. Only 5% of people there see any success. Andrew Wilkinson shared the failure story of Flow, his project management tool. This would likely happen if you tried building such a company yourself, although your mistakes would be different to his.

Their level of awareness and brand recognition is what makes Basecamp different from all of these indie hackers and small teams. This “secret sauce” enables them to sustain tech products operating in very competitive sectors. It’s practically free marketing. And because of their loudness and strongly-held beliefs, they manage to attract customers who are OK with their products the way they are. I heard a phrase that perfectly describes them: “Basecamp is a marketing agency with a software business on the side”.

Somewhere along the line, they realized that being loudly contrarian gets them attention. And what they actually do might be irrelevant, as it happened with the latest thing. DHH made a snarky commit that removed TypeScript support from their framework Turbo. Surprisingly, this coincided with their announcement of Once, which seems to be a new brand for software products you buy once and then figure out how to run and manage yourself.

And by this point, it truly feels too often they’re being contrarian for the sake of it. As if they’re specifically looking for a way to be different to everyon else. At the same time, they might be pushing people off with this approach Even their newer open-source projects, like Hotwire and Stimulus, are way less successful than Ruby on Rails was. Svelte is probably a good peer for their frontend framework, it’s also much less popular than React and Vue, yet still has 10x more stars on Github. 

Basecamp’s products are opinionated and old-fashioned. The description for Basecamp itself, their project management and communications tool, sounds perfect for a small team like ours. I tried it several times, but it always felt like you had to commit to its vision, as it seemingly lacked any flexibility or customization. I get a feeling it’s primarily used by a small number of people who click with it or non-tech companies. Especially, considering how proud they are for supporting older versions (these are customers who literally don’t want updates).

Hey, their email service, made a big splash, primarily because Apple wasn’t letting it into the AppStore. That wasn’t the smartest idea, considering the people they went against. For DHH, it was like he fell into a well of free chocolate He even went to the US Congress to talk about it. 

Hey is extremely opinionated, locked down to the point you can’t import your email history or use third-party apps with it, while their own apps are mediocre at best (largely because they’re cheaply made wrappers over their web version). I know it works for some people. It is ambitious, has an interesting story, competes with Gmail, and comes from Basecamp. I can tell you that they’d nice media coverage for this. But because of this “war” with Apple, they got 10x of what they would have if they just launched it peacefully.

I wanted to get back to the cheaply-made apps. This is a repeating theme here. Basecamp cares a lot about costs. One of the loudest recent crusades they went through was against cloud services like AWS. Don’t get me wrong, in 2023, it can be quite reasonable for a company like theirs. You have very predictable loads and aren’t planning to grow 10x; the tech stack is mature enough that running your own servers isn’t too different from virtual servers in the Amazon cloud, etc. But most importantly, they care about this because Jared and DHH get all the profits. They view it as a very reasonable and traditional approach for the business. And since their business is unlikely to start growing much faster, cutting down costs is very important. 

It’s just unusual for the tech industry. Employees of publicly traded tech companies get RSUs that complement their income and align the incentives with the company. Employees of privately held tech startups get stock options, which align their incentives and might potentially make them a lot of money (while being worth slightly over zero on average). Employees at Basecamp get nothing other than their salary, which is unusual for the tech industry.

Basecamp might be a great place to work. I heard the usual tenure for their employees is about 6 years, which is a lot for the tech industry. Although, this is the number I heard before they fired a bunch of people for discussing “politics” (or, being tired of being challenged). This arrangement might be perfect for them, they understood this in advance, it can be “priced in” their salary, plus they’re getting other perks, such as an asynchronous and remote workplace. 

I don’t have anything against this approach, as long as this was the agreement and both sides keep it. Mailchimp did a much worse thing: they didn’t offer equity, the founders boasted they’d never sell, then they sold to Intuit, and the employees who spent years building the product got nothing. 

Just don’t tell me being bootstrapped is some kind of a blessing here. It’s just another way to run your company, no more or less noble than raising venture capital or crowdfunding on Kickstarter. And it doesn’t prevent you from sharing revenue. Todoist, which was also bootstrapped, eventually started doing revenue sharing via stock options. 

For the same reason, I don’t understand companies that proudly say they will never attract venture capital or go public. It’s just not a great option for the team members, that’s all. 

Jason and DHH are genius marketers. All I’m saying is, maybe we need to be a bit more skeptical before we rush to write this tweet or a news story and give them what they want for free. P.S. Yes, I get the irony here.

There’s a certain community in tech that’s very vocal about their preference toward native apps. I share that sentiment, yet sometimes people take this idea too religiously.

First, what is a native app? It’s an app built using the native frameworks of a platform. For MacOS it could be SwiftUI. Such apps leverage the native platform’s interface, including windows, buttons, text areas, and everything else. They look right and familiar, and they behave this way. And most of the time, they’re smooth and fast. Historically, this has been the way to build applications for any platform.

But the Web has advanced so much from the early pages with hyperlinks that it’s now a platform for applications. Modern JavaScript runtimes, reactive frameworks (e.g. React), and platform APIs allow developers to use web technologies to build apps running on desktop operating systems. And this has largely become the preferred way to create new apps by now. If you use Slack, Notion, Spotify, or Todoist, then you’ve seen it. Their desktop applications are effectively just wrappers, usually built with a technology called Electron (Spotify isn’t though). And people hate Electron.

Here are the main problems people point out in web apps.

***

Memory consumption

Each Electron app basically carries a large part of Google Chrome in it and has to keep it running. Naturally, this isn’t great for memory consumption. Of course, Notion is a complex app by itself, but in comparison, the native app for Apple Notes usually takes around just 120Mb for me.

This particular problem is impossibly difficult to solve. In fact, you might save a bit of resources by simply running all the same apps in your browser, since it will at least be able to share some resources and offload tabs when they’re needed.

They might be slow

Electron apps are known to be slow. There are just too many levels, and JavaScrip is an interpreted language, not a compiled one, so by default, it will be orders of magnitude slower than an app written in Swift, Rust, or C++.

Still, in my experience, this isn’t necessarily the defining characteristic of a web app. Superhuman is insanely quick. Todoist is also seamless enough. While VS Code is sluggish, it still performs better than Atom did. Unfortunately, such apps are rare exceptions. Yes, the foundations are indeed slower, but at least some people are capable of writing quick web apps.

Feel free to share positive examples you’re aware of.

They have unique UI

Since web apps can’t just use the components provided by the operating system, they have to recreate everything from scratch. And this creates a lot of burden for developers and, I think, lowers the quality floor. Creating beautiful, compelling apps is possible, but it requires so much work. People building native apps get it all “for free”. In fact, this could be one of the reasons Apple still has such a vibrant ecosystem of great artisanal apps since developers can build most things out of nice ready-made components.

This also means users often must learn your app’s UI instead of relying on their muscle memory.

***

I’ve been focusing on desktop, but the same is happening on mobile, albeit at a smaller scale. Tools like React Native allow developers to build apps for iOS and Android. I’d say the “experience gap” is a bit larger on mobile. Notion famously started introducing more native components in their app, which instantly made it quicker and nicer to use. Still, their apps are largely web-based, which limits their functionality. For instance, it’s quite difficult to implement complete offline mode (Notion caches recently seen pages but nothing else).

Basecamp has a particularly good visual example of how this works, even though they use quite a “basic” approach.

***

Web apps are here to stay. There are fewer and fewer artisanal Mac apps these days, even though sometimes a jewel appears out of thin air, as it happened with Craft. Quite unusually, they have native apps for all of Apple’s platforms and a web app with the same functionality, which is useful for people on Windows, ChromeOS, or Linux.

Native apps do feel great. But they have their own challenges. The first and the most important one is making them cross-platform. If you’re a small developer, you might not have enough resources for this. And if you’re a big company, you don’t want to struggle to keep all your apps’ development in sync. Building a web app and wrapping it with Electron ends up being the most reasonable approach for most people.

So most of the time, our choice isn’t between native and web apps. You either have that useful app or you don’t because it didn’t make much sense for the developer. And while I like native apps and enjoy them when I have such an option, I’d rather have a web app than nothing.

If I use the app all the time, I’d want it to be at least quick (and ideally, native). But if I need it occasionally, I’d be OK with pretty much anything, as long as it’s useful and there’s nothing better. Also, we have seen evidence it’s humanly possible to build quick web apps. The fact most people don’t bother is a different question.

It seems that we’re finally getting out of this weird period of collective gaslighting where people tried to convince everyone Zoom was the best conference app out there.

I prefer Google Meet. It’s just better. I understand it was terrible pre-COVID as if Googlers never used it themselves. But they’ve caught up in a major way since then. Now it’s a very competent web app that gets out of your way and lets you talk to people. You open the link, and you’re right in the call. Meet doesn’t ask you to update anything.

Look, I’ve been working remotely since 2014. We used to have calls with the US, and most of the time, people would suggest calling their cell (despite the terrible voice quality). Then I started noticing Uber Conference and GoToMeeting, which were a bit clunky but at least used VoIP to provide clear voices.

When Zoom appeared, it made things simpler, in large part by simply becoming a common standard first for the tech industry and then for everyone else. As long as you had the app installed, you could quickly join the call, and most importantly, it was reasonable to expect your invitees wouldn’t freak out because of this link.

But Zoom was never great. It still has a very clunky and ugly interface for scheduling and calls. Try quickly figuring out how to share the invite link when you’re already on the call. And the best option to schedule a call was using Zoom add-ons in Google Calendar or another app. Try doing it from the web or in the app, it’s like a flight control system.

The worst part about Zoom is its nature as a native app. I have no idea who decided it’d be a good idea to check for updates when you open it 30 seconds before the meeting and then block you from joining the call until it’s done. Zoom does have a web app, but it doesn’t want you to use it and hides the very possibility (most people don’t even know it exists). Why does a simple app need blocking updates so often? Oh wait, it’s no longer as simple as it was.

Zoom is worth $20Bn now and was over $160Bn at its peak. Between its annoying install prompts, forgetting you’re logged in, and the add-ons nobody asked for, there’s a simple truth that Zoom’s functionality is a commodity. If you run a company, you already pay for Google Workspace or Microsoft 365, which provide your employees with email, calendars, and video calls. So Zoom is desperately trying to justify its valuation and its entire existence by introducing things you don’t really need from it. I had to get Zoom Premium a few days ago, and it was almost $15 per user. That’s a lot of money for basic functionality you can find practically everywhere. And I’m not talking about their advanced features for webinars and such, most users don’t need this.

Zoom is going down the road of Dropbox. A company that once packaged known tech in a nice way and grew on it only to realize their product has become a commodity and every tech giant has a better-integrated version. So Dropbox decided to build features that would help justify companies paying for it and, in the process, ruined the experience for regular people (I pay for OneDrive now).

Right now, Zoom is very far from that old idea of an app that lets you simply join calls in one click. And despite all Google’s fumbles in messaging, Meet is just that. I know people have problems with Microsoft Teams, but I visited meetings that used Teams and it was fine. Again, no need to install anything, just open the link in your browser and click “Join”.

We also haven’t seen too much innovation in the space. Personally, I don’t like video calls too much (especially because I like to pace when I think). The only innovation we’ve seen is switching from big rectangles to smaller circles, which might be quite nice since it removes a little bit of anxiety and you don’t feel glued to the camera as much. SpatialChat is a good example, they’ve also tried putting everyone on a virtual plane so you can move between discussions as if you were in a room.

It seems to be quite difficult to build a successful startup in this space. I’ve seen Whereby that since shifted into offering whitelabel calls to other products. And there’s Around, which was acquired by Miro, and focuses on remote and hybrid teams that have to collaborate a lot while also supporting conventional calls.

But don’t tell me Zoom is the best, please.

I’ve just watched a funny sketch about Threads from Ryan George (look him up!) and noticed one thing.

He was calling Threads an app by Meta.

When Zuckerberg renamed the company, people laughed at this as an attempt to avoid the negative publicity associated with Facebook, or a corporate reporting stunt similar to Google calling itself Alphabet.

There was one key difference.

Google barely uses this name for anything other than their corporate reports on one weird website and still makes its money from ‘Google’.

Meta actually uses its new brand throughout their products. It’s also safe to say it no longer focuses on Facebook, instead relying a lot on Instagram, which it used to implement new formats such as Stories and Reels, effectively blocking the potential growth of Snapchat and TikTok. And they’ve slowly been building out revenue-generating capabilities into WhatsApp, without making the app more complicated.

So they truly deserve a new name.

The launch of Threads, Meta’s alternative to Twitter, has reinvigorated discussions about privacy and data ownership. A number of smaller Mastodon instances have preemptively defederated Threads, people have concerns about Meta launching yet another social app and getting even more data, etc.

Personally, I believe that we’re in the midst of a moral panic our grandkids will be laughing about.

A moral panic is a widespread fear, most often an irrational one, that someone or something is a threat to the values, safety, and interests of a community or society at large.

I often see people confusing different privacy-related things and using weaponized language to convince you something is bad. Just look at the terms thrown around. You need to protect your privacy and fight the surveillance as if these people are living under Stasi or trying to fight the statewide network of spies—the words we choose matter. By framing the problem in the right context, you can position yourself as the winner without even making a move.

‘Facebook sells your data’ sounds bad. ‘Surveillance’ sounds bad. But it doesn’t sell your data, and it’s not surveillance. If you can only articulate why you dislike sometime with rhetoric that isn’t actually true, that tells us something in itself. Benedict Evans

Meta isn’t selling your data. Meta is selling an ability to show precisely-targeted ads to relevant audiences. Some companies want to sell you their product. This list includes everyone from Unilever to a local car shop that needs people who drive Chevrolet and live nearby.

Show Me The Harm

Most of the time, privacy proponents cannot provide evidence of harm experienced by anybody involved. You could come up with some examples. After several US states made abortions nearly illegal, people realized the government could demand Google to share its search and location data to prosecute women using those clinics. It’s true that if Google, Meta, and others didn’t keep this data, this “attack vector” wouldn’t exist. While this situation is extremely unfortunate, we must agree that the actual bad actors here are state governments, not Google. It’s very reasonable for services to keep the history of your searches and location data to provide better recommendations (even if this allows them to target ads better), and they have to follow the local laws (even if they’re barbaric). The reason people blame Google is twofold. Some dislike them, while others trust Google more than their government, which is why they put pressure on the former.

Other than this, all you can likely find is the old anecdote about a parent learning their daughter is pregnant because of targeted ads. This story doesn’t seem to have any solid foundation, though.

Finally, there was Cambridge Analytica. In 2018 everyone was raving about the dataset scientists gathered from Facebook and shared with third parties that basically “allowed either Trump or Russia to change the outcome of the US elections”. Except it didn’t. Because the U.K. Information Commissioner’s Office released a lengthy report that found Cambridge Analytica’s work didn’t affect either of the elections. Marketers can promise you a lot, doesn’t necessarily makes it true. Cambridge Analytica used the fact Facebook had a more open API at the time and couldn’t enforce what users would do with the data. Yet when they closed it down, a lot of people were unhappy.

Instagram Isn’t Listening To You

Some have even been convinced Meta’s apps are listening to everything you say via your smartphone’s microphone! How else would they know to show me an ad for a thing I discussed with my friends but never searched for?!

They don’t listen to you, because:

• It’s technically-challenging
• It’d be trivial to detect
• They don’t need to

Because if you talked to your friend offline, you likely were at the same place, which allowed Meta or Google to put you in the same lookalike audience and try showing you the same ad. This is why you might see something your partner is looking for online – because you spent a lot of time at the same location (be careful with your gift research). The modern adtech stack is simultaneously complex and much simpler than you imagine.

Some could say even this concept is creepy. I get that. There’s a difference between an intelligence officer watching my every move and a soulless black-box algorithm that doesn’t care about me and is only trying to optimize ROI for a certain creative. What makes me worried is exactly the governments collecting data they have no legitimate use for from their citizens.

You Aren’t the Product, You Use The Product

You’re not just losing privacy because you sinned. In return, you get access to enormous free services. I’ve long advocated for advertising-funded services because of how egalitarian they are. It’s easy for a middle-aged person born in the US to claim they’d rather pay for everything than watch ads. The only reason I got into tech is that as long as my parents paid for the Internet access (dial-up at first), I had access to all the content out there.

There’s value in targeted advertising. It’s one of the best vehicles for businesses to reach their customers. There’s a reason Apple uses two very different warnings for ATT and its own advertising engine. They try to make it sound like what they’re doing is fundamentally different, but it’s not. Apple believes that anything you do on your iOS device is their “first-party” for them but “third-party” data for Facebook.

It’s “Personalized Ads” for Apple, but scary “Tracking” for everyone else, and Apple turns their option on by default.

Sometimes I myself prefer the ad-free experience. I pay for the ad-free tier on Netflix. But there are services, like social networks, that only truly work if everyone can use them for free at all times. And you likely underestimate how much services like Instagram or Gmail would cost you in their ad-free version. In 2022, Facebook’s (the “Blue App”) ARPU worldwide was around $39.65 (per annum), but in the US & Canada, some of the most profitable regions for them, it was $206.44. Are you ready to pay $17 per month for Facebook?

But if they make so much money on us, they should pay us!

I’ve seen countless attempts to build startups aiming to do this, both for the conventional web and in crypto. All have failed. Your information doesn’t cost that much by itself (if you don’t believe me, try finding a buyer). It becomes valuable when ingested into black-box algorithms at Google and Meta that operate petabytes of data and know how to extract value from it. Oil isn’t the best metaphor for this, the closer one would be sand when it’s turned into advanced microprocessors by TSMC.

Confusing Private Things and Behavioral Data

To make things even more confusing, people conflate some very different concepts under the same umbrella. Personally, I’m OK with targeted advertising, and I understand what it requires. I’ve bought many things off Instagram ads and specifically permitted it to track my activity.

There are cases where I care much more about the security of my data. I wouldn’t want my home address to be shared with strangers. I don’t want my medical or financial records to be publicly available. Sometimes I get notifications from 1Password about yet another hack of some services that potentially includes basic details on me, such as my name, email, and my password. These rarely include credit card details because not every website keeps them, and the ones that do have to care much more about this due to reason and regulation.

I don’t like such events because they could very easily lead to harm. It’s not about algorithms reviewing my “data” to generate revenue, it’s about humans using it against me to their advantage. Nobody can just bribe Facebook to get your data out. All big platforms care a lot about their data, this is one of the things that actually improved since Google and Facebook started dominating the online advertising market.

Some things are almost borderline. They usually happen because of incidents or miscommunication. In 2010, Google admitted that the cars used by Google Street View projects collected 600Gbs of data from open WiFi networks in Germany. The cars were fitted with antennas that look for WiFi networks and use the data for their location services. I doubt they could have used that payload data, but they should have been more careful.

Another thing that comes to mind is tech companies storing voice command recordings to assess and improve the quality of speech recognition. I’d expect something like this to happen, but they should warn about it in advance and give you an opt-out if you want it.

***

I’m not against privacy, and I understand why people might not want someone to track web pages or apps they use. But it’s important to remember that there’s a trade-off and be prepared for the outcome. While I don’t like most of the GDPR applications, which just make many reasonable things illegal and solidify the monopoly of the largest tech advertising companies, allowing users to export their data or demand their account to be deleted is a good development. Creating a framework that incentivizes companies to safeguard users’ data and punishes them for leaks is also important, although I think we should be reasonable. In their desire to punish the American tech companies, the EU is too trigger-happy to issue fines exceeding what oil companies paid for the largest spills.

People like to say they prefer chronological feeds to algorithmic ones. I’ve seen countless prayers to Meta to re-enable the former option in their apps. And when Twitter made the algorithmic “For You” tab the default one, entire Chrome extensions were built, and petitions started to disable it.

I don’t think a pure reverse-order chronological feed is the best option. The actual problem here is we rarely see *good* algorithms built to help us and not drive engagement. But I’d love to see a social network giving me tools to catch up on the people I care about.

Strictly speaking, chronological feeds are based on an algorithm. A very stupid one. They work well when you only follow a few people, but they can be gamed. By design, they prioritize people who post more. People like professional bloggers and influencers. And if you only have a few minutes to check up on your contacts, you see a small slice of everyone’s recent posts – not necessarily the most important ones.

Kevin Systrom, Co-Founder of Instagram, who is now building Artifact, described their disadvantages in his interview with Alex Kantrowitz:

You would friend people, and your social network became your filter for relevant content. If I follow you and you post something, I see it. These networks grew to be slightly too large for that to be manageable.  If, on Instagram, we just sorted your feed chronologically, the people that posted the most would get the most attention and people would self report seeing too much of one person while missing other people.

He talked on the same topic in his interview with Ben Thompson:

We used to say internally, “Your feed is already ranked, it’s just ranked by order by created at descending.” Who’s to say that that’s the correct ranking for actually getting to most of your friend’s content? <…> “Can we help you stay up to date on what your friends are doing on Instagram”? That was the core of when it made Instagram special, and it was getting totally lost in a chron-ranked feed, because your friend posted 8:00 PM and in between then and the next morning, the brands you follow, the influencers posted fifty times because they have a whole team to help them post, and that friend post was totally buried.

I agree with this assessment. In fact, it mirrors my own experience with Facebook and other early social networks. If you are using RSS, you’re seeing the same problems. If you try subscribing to a major news website, they will drown you in content, dominating everything else. And if you’ve been away for a week, you’ll probably glance at the last headlines and just mark everything as read.

(Two years ago, I wanted to build an RSS service that would rank articles and posts for you. Thankfully, they launched Artifact, and I just use it.)

In anthropology, there’s a concept called Dunbar’s number. It is a cognitive limit to the number of people with whom one can maintain stable social relationships. And Robin Dunbar himself estimated this number to be around 150. I’ve always felt that conventional social networks and chronological feeds break around this exact number of follows.

Some people (especially in tech) claim that you must manage who you follow much more strictly. Follow carefully, remove swiftly, and mute the ones who post too much noise. I did this myself. And there are two very clear problems:

1. You clearly can’t expect everyone to do this.
2. When you mute a person who posts too much, you lose them ‘forever’. Unless you manually visit their page and give them a second chance, they will never re-appear in your feed. Maybe they just went to Coachella but ultimately are still reasonable people?

Social networking apps have tried to solve this in different ways. A now-defunct Path only allowed you to be friends with 150 people (the same Dunbar’s number). This would lead to a weird dance of trying to find someone to throw away to make space for the new person you want to add. Others, like BeReal, solve it through their UI and by limiting the volume of content. All of these options seem too artificial.

Twitter’s previous algorithmic timeline was actually quite useful in terms of surfacing the stuff I might have missed and mixing it with just a few percent of potentially interesting authors I don’t follow. The reason I (and many people) dislike the current “For You” tab is that it’s pretty much a TikTok for text. It’s extremely addictive, but when I pull away I realize that I don’t care about any of these people.

A useful algorithm would work to surface things I shouldn’t miss from the people I already follow. No matter how long you were offline, it’d highlight posts you need to have a look at. One of the common criticisms of algorithmic feeds is people refreshing the webpage and losing their position. This can be solved if the algorithm tracks the “seen” status (just like RSS readers do), so someone who is a completionist could read everything.

That’s what makes me slightly hopeful for BlueSky. One of the reasons they weren’t satisfied with ActivityPub and decided to build AT Protocol is precisely because the former doesn’t support algorithmic feeds.

At Bluesky, we’re approaching this challenge from a marketplace perspective. Our goal isn’t to create every algorithm in-house, but to enable the developer community to bring new algorithms to users swiftly and effortlessly

***

Imagine being able, for the first time, to choose the algorithm building your personalized feed. Focusing on what you value. Mixing it with popular content or keeping it to the same list of N people you follow.

I don’t know who will actually ever build this. But it’d be wonderful.

Hating on media paywalls is very common. Whenever you see a paywalled article shared on Twitter, you can see people asking what’s inside and if there’s a way to access it.

One solution many people propose is micropayments. This implies being able to pay a relatively small fixed amount to access a single news story without subscribing. This is what Elon wants to implement at Twitter.

Rolling out next month, this platform will allow media publishers to charge users on a per article basis with one click.

This enables users who would not sign up for a monthly subscription to pay a higher per article price for when they want to read an occasional article.…

— gorklon rust (@elonmusk) April 29, 2023

Ben Thompson has a good explainer on why this wouldn’t work in the latest episode of Sharp Tech. The key reason is that while this might sound enticing for consumers (we’ll get back to this later), it’s technically impractical and terrible for businesses.

First, micropayments simply don’t make sense for publications. They can’t afford to acquire customers each and every time. They’ve invested in their teams and facilities. There are no marginal costs in a single article, but it’s also not an .mp3 file you could be selling for decades. Even the most successful publications, like The New York Times, aren’t terribly profitable (NYT had a profit margin of 15% in 2022)l. Most others fare much worse. So they’d much rather have a predictable recurring revenue instead of trying to entice customers to pay every time.

Second, card transactions are expensive. In the US, you’d pay at least $0.29 + 3% (most likely more). So if you want to sell one article for $1, over 30% would evaporate. And this is before we start talking about Apple’s 30% if you’re selling on iOS. There are ways around this. Twitch has tipping, so you buy a bunch of internal “tokens” in advance and then use them to award creators. This way, you only pay the credit card fees once. But you need a central actor to run this, and Twitter could have been such a platform, considering its role in the news ecosystem. Well, in its previous life, until Elon burned all the bridges and started replying to press requests with a poop emoji.

But now we’re getting back to the first point – news orgs (and everyone else) don’t actually benefit from micropayments. Twitter can build it, but I’m not sure who’d use it. It’s not just incremental revenue – it’s also lost recurring revenue from all those subscribers who’d subscribe if they couldn’t access the content another way. If you don’t want to subscribe, you aren’t the target customer.

One could argue that putting all your content behind a hard paywall could lead to a declining reputation since fewer people can read it (and be influenced). There are all sorts of factors here. This is a complex question, so let’s leave it for another day.

It’s impractical to expect people to subscribe to all news media. Well, I have to, because I work in PR, but I’m an outlier. But everyone has been saying the same about video streaming. The average person subscribes to 2.8 streaming services. You heard about that show on Disney+, so you subscribe for Disney+, and then you cancel it later.

Why won’t we do it with news websites?

Ok, there is no “Buy for $5” button, but if you really want to read it, why not subscribe for one month?

I think part of the reason is that people don’t actually trust their paywalls and subscription management. Netflix is the gold standard of subscriptions.

Look at its paywall. Clean, simple, no weird trials, no pricing change after the first month/year.

When you do subscribe, they send you an email before the next charge to remind you about it. Feel free to cancel in advance. And if you want to, it’s very straightforward.

With news media there are two key problems that I see over and over go: unclear pricing and a difficult cancelation process.

Now we’re slowly getting out of the terrible age when the only way of canceling your The New York Times subscription was to call someone. It maybe took FTC to get them them. But most news website have finally agreed you should be able to cancel your subscription online. Still, there’s still an expectation among consumers that you can’t. And with The Wall Street Journal it’s actually still the case (unless you’re covered by a local law). You don’t do this when you know you’re delivering value to you customers. You do this when you hope they don’t notice the charges for two years.

Then you have the “paywalls” themselves. Marketers at these companies try very hard to get you in and then start charging the full price afterward. The paywalls are designed to trick you.

Here’s Insider.

And here’s Financial Times.

You’d think having small grey font is something more akin to a sleazy insurance policy and not an news org you should trust, but here we are.

Just give us straightforward pricing and promise it’s easy to cancel online. I subscribe to The Information, and even though the subscription is expensive, the decision was easy, because I expect everything to be smooth and simple.

If you already paid for the first month, there’s a clear incentive to try using it while you can. The same you’d do with Netflix, Hulu or another streaming that shouldn’t have been started in the first place. And then, maybe, you realize you’re getting value and don’t want to lose it. Plus, nobody is subscribing to a news website to binge old news in one month just to cancel it right away.

Reading all those discussions about the updates to Instagram and how people are turning disillusioned with the app, I began thinking: what would an ideal anti-Instagram service look like?

Let’s look at the most common complaints and try to find an alternative. Some of them are specific to the way Instagram works right now, some are true about any social network.

At last, we will try to figure out a potential business model that might work for this product and still onboard all of your friends.

Why Instagram?

However popular TikTok is becoming, I still think that Instagram is the primary social network for most regions in the world. This is where people keep their online profiles, this is what you ask if you meet a new friend (unless you go directly to messengers). Nobody is using Facebook to proactively meet and befriend people. Instagram offered a very simple way to produce content through photos and this is what most people can actually do. Especially compared to elaborate TikTok videos or endless snarky tweets.

***

The Feed

The thing people vocally hate about the updates to Instagram is the changes to your feed. Instagram is pushing more and more content from people you don’t follow into your timeline because their data shows people tend to watch it.

The solution would be to only include content from people you’re following. There can be a dedicated discovery tab if you want to expand but it should never intersect with the feed.

It goes further. The OG Instagram was indeed about your friends. There were no influencers. The square photo format and filters were added for many reasons, but one od them was probably to disincentivize professional-grade content.

Is there a way to build a social network with no influencers? That’s a difficult question. You could copy Path and limit the number of follows or you could build it on top of phone numbers. Still, I could easily imagine services providing burner phone numbers so celebrities wouldn’t have to give up theirs.

Influencers are the power users of any network. They produce the most content per capita, likely the best content and generate the most engagement. Removing them might harmstring your growth but that would an interesting development.

Content

I don’t think people have problems with video per se. Except for the fact Instagram is favoring it algorithmically – which we “solved” in the previous paragraph.

Even more, get us more content types: text, entire galleries of photos, map locations, etc. There are apps like Dispo that experiment with the way you create content but I feel it’s just a gimmick. There’s no reason to make low-quality photos with your latest iPhone.

Photos are the easiest way to document your life and each of us now has a pretty good camera in our pocket. Let’s not limit it artificially.

When people are performing weird hacks to overcome the limitations of your platform it’s a sign that there might be something wrong. Instagram used to be quite good at noticing such trends and adapting to them. I remember how people started uploading vertical and horizontal photos by adding white bars. Eventually, Instagram submitted and let us upload freeform images.

When people on your platform are posting long texts as screenshots of Apple Notes in a gallery then something is wrong. When I’m making a screenshot just to use Apple Photos OCR to extract a link or a phone number then something is wrong. Such use cases should be assessed and processed. It’s weird that Instagram is worse than WordPress for a travel blog – it should be 10x better.

Engagement

There was a short period of time when it looked like Instagram cared about your screen time. You’d get a prompt that you have looked through all the new posts. But very quickly they started showing suggested posts instead of the previous ones.

This is a common claim towards any social network – they want to take over your time.

There’s an option to solve this I dream about.

Abolish the feed as it is.

Either use an RSS reader-style list of posts with read status explicitly shown or switch to a daily digest instead of a time-sucking endless feed.

Let’s say the user can choose the time of day and get their digest. Change it not too often.

BeReal found something similar but I don’t think it’s the content creation moment that matters. Most of the time I’m literally doing nothing spectacular. But if I only can catch up once a day it’ll be a much healthier experience.

The Walled Garden

Instagram is a walled garden. At least we now have a web version, which you can use on the iPad in the absence of a native app. You can’t select and copy text. You can’t add links to posts, only stories. You can’t reliably read all comments, especially on the web.

I’m following all kinds of volunteers and while Instagram is a good place for them to build their audience, it’s terribly difficult to raise funds or even leave a CTA.

So let’s open it. And that might allow us to solve the chicken and the egg problem of any emerging social network.

Users have no incentive to start a new social network if it doesn’t have their friends already. So they should be able to easily share their content outside. And if we’re adopting the digests model, we can use email to distribute it. We still have to acquire and onboard the user who creates the content, at first their followers could sign up via email or third-party messaging apps.

Position it as a life journal so there’s an incentive to post at least for yourself while the network is trying to grow.

The app would provide a more complete experience, comments, DMs and all other things so people interested in it might join eventually.

Monetization

The most important part. We can come up with the ideal app, but if it doesn’t have a scalable and inclusive business model it will fail and lose ground to Meta or anything that comes after them.

Glass is an interesting example. It’s also a photos app. But because of their chosen style and a $5 subscription only photographers are using it. It’s unreasonable to expect everyone to pay a comparable fee, especially since the world has other countries in addition to the US and Europe. Micro.blog is similar. It is very close in terms of its ethos. But most of my friends will not be paying $60 a year for it.

WhatsApp experimented a bit with a $1 annual subscription in select regions. On paper, it might work. Build an app that has almost everyone and you will have at least over a billion dollars. Still much less compared to Meta. And in the early stages, this will likely be unsustainable.

In fact, I don’t even have a problem with ads. Ads made Internet free and available to millions of kids, including myself. But all the ideas introduced here, from daily digests to focusing on real friends, clash directly with the advertising monetization model and the engagement and data it requires.

***

I would call this the Social Network Dilemma. A social network is only good if it actually has at least most of your friends. To do so it has to be free for the end user and have a viable business model that powers its growth flywheel. But then we end up with an app that has dramatically different interests and doesn’t really serve us. And subscription access limits the numbers of people who’d onboard further decreasing the value of such a network.

I do believe that with a sort of a distinction between content creators and content viewers we can accommodate some of these downsides. If the content is spread beyond the network and people have multiple ways to follow without being active participants it is still viable. But who knows?

The only option we seem to have is some kind of a premium tier for people who might want a bit more in terms of features. Similar to what Telegram is doing (with an unknown level of success). If you want more photos per post, unique visuals and appearance, and other power user features – you pay. Only by letting people join for free and still earning money through subscriptions the network I described is likely to be viable. Still, the subscription by itself doesn’t solve all the underlying problem. You still want more users, you want them to be engaged, so you might end up in an eerily similar place. Or let business users in and charge them for their accounts. But you need to be able to generate value for them.

Building a social network is probably one of the hardest problems in consumer tech world and I’m not going to pretend I know anything about this. But as a user I do often feel that the apps we use let us down and don’t offer an experience designed specifically for our goals.

I think the lack of sideloading has been a tremendous advantage for iOS and Apple’s ecosystem overall. I also need Apple should open it up.

The fact you can only install apps from the AppStore on iOS gives users, developers, and Apple itself tremendous benefits. Users can install any app with no fear for their wallet or data (hm, most of the time). In turn, developers get more revenue because users are willing to install and buy new apps. Developers also don’t have to worry about piracy – it doesn’t exist. That’s different on Android, where I can go to an online forum full of cracked .apk files visited by 30 million people a month. Google even offers special anti-piracy protection to developers. I think it’s more than just about the lost share of their revenue. This status quo has likely affected the very business model of thousands of apps on Google Play Store, pushing developers to advertising or from developing an Android app in the first place.

iOS doesn’t have these issues. There’s a notion that all Apple platforms don’t really suffer from piracy or malware but I can’t really agree with that. There are websites with cracked apps for MacOS (I won’t link to them but you probably can find them) and there is malware targeting Macs. The scale is much smaller, of course, but that is just a correlation to the number of Mac users. This is just iOS.

The fact you had to go through the closed AppStore and give up 30% of digital revenue is an issue for many developers and publishers and will likely be one of the key points of any future anti-trust action against Apple. Epic Games played a game of chicken with Apple and working things out in court right now. If iOS allowed sideloading without jailbreak many developers might have chosen to stay out of the AppStore, just like they don’t distribute their Mac apps through Mac App Store. But that would also destroy some of the advantages iOS creates for both users and developers.

I believe there’s a very particulate user where sideloading is reasonable. To clarify, where Apple relinquishing control over what apps you install is reasonable.

It’s censorship.

People throw that word all the time now but I’m specifically talking about the government-level censorship. Some people believe that the block of Parler by everyone in the tech industry from Apple to Amazon AWS was an act of censorship but that’s a different discussion that I’m not going to participate in right now.

Clubhouse, a voice-based social app value at $1 billion in their most recent Series B, just got blocked in China on multiple levels: ISP, Apple AppStore, carrier networks. This means that in order to run Clubhouse there you need:

1. A VPN
2. An alternative iCloud account for the AppStore
3. A Non-China phone number

That’s more than enough friction. VPN is probably the easiest part, though most of the well-known ones are blocked as well. AppStore is a bit trickier – you can create a new account without a credit card. But it’s a huge hassle to switch between them and you’d lose some things e.g. like tracks downloaded in Apple Music. And if you need to buy an app or make an in-app purchase you’re out of luck – you need a foreign credit card as well.

Clubhouse is hardly alone. Most of the Western social networks and apps are blocked in China in a similar manner, although at least as far as I know they don’t always block confirmation texts. But you can’t access Facebook, Google, Youtube, or a myriad of other services in China. Apple follows the demand of China and removes entire categories of apps from the Chinese AppStore. They even built a unique Apple Podcasts app for the Chinese market.

India banned TikTok in June 2020 alongside 59 other Chinese-made apps, citing national security concerns. Russia blocked LinkedIn and has been developing the technical and legal capabilities to enforce such bans. Benedict Evans called the end of the American Internet in 2020 since 80-90% of Internet users are now outside the USA while the Internet itself became vastly more important. This means that countries with a tendency to show off their sovereignty begin to both favor their tech champions and heavily regulate the global entrants.

The Internet was born as something truly global but in a sense it was American. US-based companies dominated since they were able to generate more revenue and spread their fixed costs on a much larger userbase than any platform in a smaller country could. As a result, the US partially exported their legal system and values onto those platforms. And I think that had a positive effect on most of the world since those values often were more open and democratic. But now it seems that process was a fluctuation and not something we should take for granted.

Apple still wants to sell its products in China, India, Russia, and all other countries. But having the centralized switch to control what apps you can install and how you pay for them becomes a much larger liability for them when governments get involved. And it’s very detrimental to their users. There are bypasses for free apps but only of them. If iOS allowed sideloading even in some limited form alongside the main AppStore that pressure would be removed.

Potentially, getting access to a blocked app might be the only reason I’d ever buy an Android phone if there is no other way.

Our browsers are astoundingly outdated and their developers seem to be oblivious to that. We went from basic Hacker-News-style HTML pages sprinkled with a little bit of Javascript to running full-scale applications like Figma or Descript – something you wouldn’t believe to be even possible in a browser ten years ago. The whole definition of the web browser changed – it’s not just something you use to “surf the web”, it’s where people often do all of their work: email, calendars, documents, design, even code with things like Github Codespaces or Replit.

Now, look at this Chrome from 2008 or so.

And now look at the modern one.

It certainly looks better. Yet the actual UI barely changed in all these years.

I’m mostly using Chrome as an example because it’s at least 57% of the modern web and is expected to be the most advanced but you could look at most other browsers and you’d see the same picture.

We went from web pages to web apps but the UI paradigm hasn’t really changed. Everything is a tab and all tabs are the same and each belongs to some window – which are also the same.

You can pin tabs. Most of the time I have 3-4 most important apps pinned right there and I’ve seen a lot of people do that. Still, they live in the same area and occupy the same shortcut space (Cmd-1,2,3,…,N).

PWA and in-built Chrome’s functionality to “Install” apps could have helped. It creates a local shortcut for this webpage and launches as a dedicated app within your OS. But in order for the experience to be smooth you really need the app to be a PWA, otherwise, it’s too finicky for a regular user. For some reason, only a couple of Google services, such as Google Photos, are actually available as a PWA and look like a well-behaved local app. All of their most popular apps, such as Gmail, Calendar, or Docs don’t. Even if you “install” them, manually change the icon to look good, you’ll soon face issues, for instance, it’s impossible to change an account in Google Calendar without being thrown back to the browser. Google could have led us and turn all OS into literal launchpads for their own web apps would be highly beneficial but instead they dropped the ball.

In general, there are two distinctive ways people deal with browsers tabs:

• People who mostly have a few tabs
• People who have two hundred tabs at least

Personally, I’m in the first category as too many tabs immediately make me anxious. But I doubt browsers actually work well for the second category – or do anything about this scenario at all. Go hunting with your mouse or use shortcuts to switch between the first 9 tabs and left-right. There are extensions that help with it: from The Great Suspender that intelligently unloads the unused tabs to OneTab that helps you save sessions to things like Tabbs that just help you find anything in your browser. Unfortunately, most of these extensions are created by single developers, can be dropped at any moment and have terrible design (not all of them, but most). Basically, you’re at their mercy.

It reminds me of the way early Mozilla Firefox worked – its early versions were so barebones, you’d have to install extensions just to get you some basic functionality like saved sessions or downloads manager. I think we can expect more from our browsers in 2020.

Sometimes they give us an experimental feature that could really help, like recently added tab groups, but based on the past performance there’s no indication they’d continue working on it and provide an API to extension developers. Imagine if you could save a tab group for later, push any existing tab into a tab group you saved, if they could freeze automatically when you collapse a group so they wouldn’t take memory. If only.

In fact, these are the things I would want to be present in a browser:

1. Advanced tab management functionality
2. Session manager to quickly switch between different contexts
3. Command menu to quickly switch between tabs and look up items in history
4. Better support for proper web apps

You can achieve some of that with a combination of extensions but it’d all be supported with a duck tape and look pretty bad.

By the way, Reading Mode that Chrome added some time ago is a joke.

Solutions

In the past few years, I’ve seen a few attempts from venture-backed companies to create a new “browser” or augment the existing ones. Why venture-backed is important? Well, these are companies, not individuals, and they have enough cash to build out a product that might become financially viable. I’m not yet sure most of them would be able to justify venture-scale returns in the end though, except maybe a few.

Mighty

Mighty is a cloud streaming for Chrome. You can buy the lowest-end MacBook Air and then casually browse 500 tabs filled with heavy apps with no issues. Our laptops are highly optimized for video streaming so it should last long and eat way less RAM.

Initially, I thought it was a weird idea, but then I realized that a) we all work in web browsers b) if you solve Chrome streaming, you solve streaming for any app.

Mighty is in limited access right now. I haven’t tried it but as far as I know they add a few niceties here and there to make Chrome a bit better. Maybe in the future they’d heavily upgrade it as they had to fork it anyway but we don’t know yet. I suppose it doesn’t actually solve the issues I’ve been talking about but would solve the innate “heaviness” of Chrome.

Browser Company

The elevator pitch for the Browser Company is exactly about building a new modern browser from the ground up. They are very early so you can only see certain experiments they showcase from time to time.

Despite the ways our internet use has evolved, the browser has remained relatively unchanged. While all of our other software tools are changing for the better—with more collaborative features, flexible interfaces, and powerful functionality—the browser largely still does what it did twenty-five years ago.

That’s why we’re building a new browser—we believe it could do so much more to empower us. We’re imagining a browser that can think as quickly as we do, take work off of our plates, and pull our creativity forward. A browser equipped for the way we use the internet in 2020, and foundational for how we hope to use it in the future.

Building a new browser is insanely hard. They are extremely complicated apps. Even if you just fork Chrome and try to add some UI you’d likely fall behind very quickly. I’ve been using Edge for some time and it already falls behind Chrome in a few areas: no rich audio control, tab groups can’t be collapsed, etc.

Station

Station started as an alternative to browser – a command center for your web apps. Its biggest advantage was that tabs were tied to an app so you could have a hundred of them opened and you wouldn’t even see that. Also, things like being able to quickly switch between accounts in a more reliable way. Station was built on top of Electron, a framework developers use to bring web apps to the desktop. Each app or service has to be added to Station directly by the team. On top of that they had to rebuild all the features users would expect in the browser.

With time the developer realized it was a dead end. Users liked the concept but the performance was pretty bad so they’d churn. The team made a heavy decision to drop it completely and turn Station into a regular Chrome extension. Now they don’t have to recreate the entire browser but at the same time they don’t have control over it and their functionality will be forever limited by the APIs provided to them. Also, I recently switched to a new MacBook with M1 inside it, but on my previous Mac it was insanely slow and I couldn’t use Station at all.

There have been a number of similar apps such as Wavebox, Shift, and a bunch of others. I think the ones trying to rebuild the browser based on Electron won’t have much luck. It just doesn’t make sense if people still have to use another browser and it’d be quite hard to catch up.

Building a new web browser from scratch using a framework (Electron) that was never designed to do it meant iteration cycles were slow. Very slow. We were not learning fast enough. On top, we had to build features you simply expect as standard in a browser and fix bugs that users wouldn’t typically experience in a regular browser. That taught us nothing at all.

Workona

Workona is a Chrome extension that adds deeper integrations with apps and enables search across their histories, proper sessions and tab management, as well as collaboration features for teams. It also faces all issues of being a heavy extension replacing the new tab – on weaker machines it’s a bit unpleasant to use. When I was using it, it also had limited integrations with apps – it couldn’t really do more than search your browser’s history for a respective domain.

Sidekick

The team behind Sidekick decided to outright fork Chrome. That allowed them to perform far deeper modifications and keep it quick.

• Sidekick has a separate apps bar on the left that is visible on all pages. So if you always want to have Gmail, Notion or Calendar open you don’t have to keep them pinned and constantly stumble upon them. They’re here.
• Sidekick has a very powerful search across your history and open tabs. If you want to open a document, a Slack channel, or a Figma mockup it’s likely you opened it before – so they can provide a ton of value by just surfacing your history items.
• Sidekick has user sessions and you can quickly switch between contexts. At the end of your workday you could just switch to a personal session and have a clean slate, while all your countless work tabs are safely preserved.
• It has some nice additions such as an embedded ad & tracker blocker and ML-enabled tab suspender to save RAM.
• What’s also important is that their modifications are rather limited. You can enable Chrome syncing and other Chrome instances would believe it’s a regular Chrome, so you can send a tab to it from mobile, for instance. Still, the features may lag behind, it doesn’t have a build for Apple Silicon, for instance.

I really hope we will soon see ever more significant advancements in the browsers’ UI.

Upd. Based on the feedback I’ve adjusted the description of Wavebox as it’s not an Electron app but a Chromium fork.

Probably, one of my most controversial opinions is that online ads are actually great for the society and the public narrative shifting to paid subscription software, services and content might prevent a lot of people from accessing them in the first place.

You see people in tech being very vocal on the topic of monetization models of Google and Facebook. Some claim they’d prefer to have a paid ad-free experience and ready to pay however. As of September 2020 the ARPU for Facebook is almost $12 per month and it’s mostly driven up by the US and Western Europe – that’s how much they’d have to pay. Tech Twitter welcomed Hey, a paid email service that costs at least $99 a month, as well as a number of other apps and services. That effect is accelerated by the whole industry of apps switching from one-time purchases to subscriptions – a whole different question. And paid premium software is getting way more expensive, something I covered before.

It’s one thing to live in San Francisco, work at a tech incumbent or even a startup, and have enough resources to throw some for an exclusive email service. Or you can live in Utah. Or maybe in a small village in Italy. The average dollar-adjusted income varies in different developed countries but it’s still high compared to the rest of the world.

But there are people who have a much smaller income or maybe none of it. I didn’t have a debit card when I was growing app. They were a thing but my parents didn’t really have them. Even if I had some cash it would have been extremely difficult to buy something online with it. I was only able to start doing that when I started earning a bit of money doing web development – and accepting WebMoney, a product with a truly terrible experience but too few alternatives at the time.

Despite all of that I was able to enjoy the same global Internet as basically anyone else. We used the same Google. We all raved about Gmail when it launched. I learned a ton of stuff by reading tutorials and chatting on forums. Early social networks allowed us to communicate with our friends and classmates. All of that was free for us and funded by our mere potential to buy something. And it was beautiful.

I enjoy seeing great artisanal software created for people who are willing to pay more money to get nicer things. I’ve been one of these people on multiple occasions. And yet I worry that this might lead to the segregation of the Internet and radically different experiences for people with an opportunity to pay and the ones without it.

There’s nothing nefarious with collecting data we share online in order to serve better, more relevant ads. It’s not your crazy stalker watching your windows through a pair of binoculars. It’s a set of algorithms which don’t care about you at all. Radically limiting that data collection would lead to much lower CPM/CPA and fewer people and companies willing to pay for your attention. Online ads aren’t always great: they might lead to a bad reading experience or throttle your browser. I support fighting that part of the issue. But I’ve never really hated the experience that ads on Facebook or Instagram give me. It’s not ideal but I wouldn’t pay for that service.

By the way, Hey now has 30,000 paying customers, which is very impressive but also shows how many people are actually ready to pay for software when there are free alternatives.

And of course, there’s a ton of edge cases in tech, just like in any other industry. People might use social networks to spread political lies. Or target real estate ads to only white people. This is still a new industry and we need to deal with it on a case by case basis, learning and adjusting the policies to avoid this. But threatening to throw away the concept altogether is just impractical.

Google was founded in 1998. Altavista was created 3 years earlier and although it was probably the most popular search engine for the English Internet it lost its relevance simply because Google was better at indexing the web and ranking the pages to extract the most relevant webpage for your query.

There’s an infamous negging question that some investors ask:

“It’s cool you folks are doing that but what if Google enters the space?”.

That question is mostly useless because Google isn’t really a single entity. It has myriads of products that you could compete with:

• There are core products where Google is practically invincible: search and advertising network.
• There are products where Google’s advantages in scale make it harder to compete. Google Photos is probably one of their best services. Google can easily absorb storage costs and has enough CS PhDs to augment it with basic AI. Try building something better – and free. Google Stadia requires very specific competencies in data centers, custom hardware, and exquisite backend. Even Android phones – Google might not be the best at hardware, but the example of Essential showed us how difficult it actually is.
• And there’s everything else, from Google Tasks to a myriad of canceled products in the Google’s graveyard.

You can definitely compete with Google’s non-core products. Gmail’s former lead designer explained why even with all their resources Google would have issues with shipping better software.

The obvious ways to tackle this problem is from the inside. I’ve tried that. As much as I loved Google Inbox (a product and team I co-founded but left before launch), what launched was a well executed and polished but watered-down version of the original vision.It isn’t that the teams behind these products don’t care. They care deeply. But big products have big teams and are often inside big companies with their own bigger (and often unclear) priorities. It is messy.

Can you tackle their core business though?

Facebook managed to build a compelling advertising business due to their enormous userbase and innate ability to capture data to target ads better.

Nobody was able to beat Google search though. The problem is that like many kinds of software there’s a positive feedback loop. The more people use Google the better results it can provide. This leads to even more people using Google since they believe it’s a superior product. All the data goes back to Google and improves its algorithms. With all that revenue Google is able to hire the smartest people in the world and pay them way above the market to ensure its algorithms would be advancing faster than any competitor is able to.

Yes, there’s Bing and privacy concerns push some people to DuckDuckGo but almost all of them admit that its search quality is at least slightly worse – it depends whether you care more for the relevance of the results or visual simplicity, lack of ads, AMP pages and search snippets that alternative engines provide.

10x or Nothing

In order to really beat Google, you need to build a compelling product. And since users’ habits are really hard to change, this product needs to be not just marginally better, but 10x better – and without all that data that Google already has. But wait, how can you achieve it? It’s not like Google’s search results are trash. They’re generally very good. When I was a kid I had an entire book on how to use various search engines (most of the services it taught are dead already). But now you don’t even have to learn, it basically knows what you need.

The current understanding of technology trends say that new platforms emerge during paradigm shifts.

• Microsoft became a dominant tech giant because it entered the market with the right product at the right time. As the hardware architecture became open, it quickly was commoditized and IBM with their old mainframes couldn’t compete.
• Then Microsoft lost its relevance when modern smartphones appeared and turned out to be more important devices than personal computers. Apple and Google defined that market and benefited greatly from it. In 2019 Apple made 54.7% of its revenue on the iPhone alone.

That gives us the direction to look into. Google takes our queries and provides us with webpages. But the fact is, we don’t really need webpages, at least in most cases. We need answers. That’s why Google has been adding all those search snippets that give you an answer without making you follow a link. But these snippets aren’t automated at the high-level and there’s an army of developers of Google building them out.

The paradigm shift would be some sort of AI that gives you answers to your questions. That could convince people to actually move to your platform – because the advantages are 100% clear. And this is an improbable task. We’ve taught Alexa how to answer the birthdate of Lincoln and the distance between the Earth and the Sun. Imagine a service that has to answer any question. What’s the best chair? What drugs to take if you have the flu? There’s an enormous set of challenges, not even just technical but also related to legal questions and basic safety.

Open AI recently released their GPT-3 language model that can generate answers. But people already figured out reliable ways to crack it. We’re still not able to create something that would pass a Turing test and it seems in order to replace the search engine with the answer engine we’d need a proper strong AI.

Google understands that. They invest enormous resources in AI, so many that a $500M acquisition of DeepMind pales. But the thing about the paradigm shifts is that incumbents aren’t always able to overcome them and keep their position at the technology Everest.

***

Well, it’s 2023, ChatGPT is out there and Google’s founders got back to its HQ to discuss the threat posed by Open AI.

Recently John Gruber asked on Twitter if anyone is aware of a good native outliner app for Apple platforms. There are two extremely tragic aspects of the reactions:

• There aren’t any.
• Most people don’t know what outliners actually are.

What Are Outliners

Outliners are a very curious category of software products that have been mostly used by a small number of geeks but recently captured more attention with the launch of Roam Research. Yes, it’s also an outliner.

An outliner is a text editor that organizes information in a hierarchy, allowing users to control the level of detail and to reorganize according to the structure (source). In outliners the bullet item is the atomic unit of information. Your notes can be very detailed, yet organized for a casual reader to get a quick overview. Certain apps add their own flare to this workflow but similar at their core. Examples include apps like emacs org-mode, Workflowy, Dynalist, OmniOutliner, Little Outliner, and Roam Research.

Why People Love Them

Outliners impose a very strict hierarchical structure that shapes everything you write. Paradoxically, they’re also extremely flexible.

• Outliners are great for note-taking because they incentivize you to structure your notes.
• Outliners are great for task management because they let you build what you need at any particular moment. Ordinary task managers impose a specific workflow: some have projects, some have areas, some have priority tags. And if you were to look at people’s outlines – they are dramatically different.
• Real life is often too messy to accurately separate quick notes from tasks. Outliners can digest both notes and tasks at the same time.

I noticed that any meeting notes I write in them immediately become better as I a) ensure that any bullet item covers one thought or an action item b) related notes are attached together.

But outlining isn’t just about writing in bullet points, you could do that in a Google Doc. You can also expand and collapse certain nodes along with their children to focus on what you need at the moment. Modern outliners provide additional features, such as allowing you to zoom on specific items and add tagging and filtering to enable a second dynamic hierarchy in addition to the core static one.

That makes outliners perfect for task lists. Despite all kinds of productivity and task management apps out there, I noticed that most of the executives I know end up just writing everything as a list in some note. Or maybe something slightly more complicated.

I’m still not sure if that is a sign of failure of all these apps or if this particular approach is truly the best and we just haven’t seen the app built to capture that. Outliners are close, Stewart Butterfield, CEO of Slack, has been a fan of Workflowy.

emacs org-mode

emacs has a document editing, formatting, and organizing mode called quite literally org-mode. Org-mode was created by Carsten Dominik in 2003, originally to organize his own life and work.

It’s very geeky and just like many DIY solutions popular among the UNIX fans is known for its quirks, steep learning curve, and issues with specific implementations. Yet people still use it and add new feature through plugins.

Omni Outliner

OmniOutliner is an outlining app for macOS and iOS created by the famous productivity masters at The Omni Group. OmniOutliner has most of the features of a conventional outliner and comes in two versions: Essentials and Pro. Essentials is heavily stripped down and available for $19.99. The Pro version is priced at $99.99 and has extensive customization options, section navigation, automation, and other features.

It’s more focused on actual lists keeping and formatting. Compared to many other apps in this list it’s what Pages is to TextEdit. And this is also probably the only outliner that has a good native iOS app. Unfortunately, it’s pretty classic and doesn’t have any tags or filters – even though OmniFocus, their well-known project management app, excels at these.

Workflowy

Workflowy is one of the most famous apps in the outliners space. It was created by Mike Turitzin and Jesse Patel. Workflowy redefined the concept of the outliner in multiple ways.

1. Workflowy threw away the old-fashioned file-based structure. You don’t create new files or folders in Workflowy, instead you have a giant infinitely-nested list. But you can zoom in on any node and focus on it and put the nodes you often need in favorites.
2. Workflowy added tags and saved searches that allow you to filter nodes based and only surface what you need at the moment across the entire graph.
3. More recently Workflowy added dates providing even more project management capabilities.

For a couple of years, the app wasn’t seeing any updates and that took a toll as many users tried to switch to something else. In 2018 the co-founder promised to make it better. Since then they launched new apps and added some long-requested features like the dates support.

Workflowy is my personal outliner of choice now. My project management workflow is based on a long tree of clients with all the stuff I need to do and check. I mark specific tasks with tags: @do for what I’ll be doing today, @later for non-urgent things, tags for individual team members to remember what I delegated, and dates when we have specific deadlines. Workflowy allows me to quickly switch context from observing everything we have on our plate to focusing on a single project or task. It’s also just helpful for organizing my thoughts – I wrote this post in Workflowy first.

Still, the development pace isn’t exactly where I’d love it to be, it’s tedious to add notes and tasks to Workflowy from other sources (no options to add via email or via share sheet) and the iOS app is a pretty obvious wrapper on top of the web version that doesn’t feel native. Workflowy also has suboptimal official manuals: you’re supposed to read the blog and for a lot of stuff the most recent records 4-7 years old are often outdated.

Workflowy’s free tier is pretty robust yet you can only create a fixed number of new nodes each month. Early users have it fixed at 250, new users only get 100 now. The paid version is $4.99 per month and $49 per year.

Dynalist

Dynalist was created in 2015 and was more of a response to the stalled development of Workflowy.

Compared to Workflowy, Dynalist has more advanced dates, attachments, version history, calendar integration, and a ton of other stuff. It also just gets so many things right and the pace of development is much faster. The app itself doesn’t look that polished and mobile apps are especially bad. They’re wrappers of the web version and aren’t good ones.

It gives more for free yet costs $10 a month ($96 a year) if you want advanced ones (and believe me, you want the ability to quickly jump between nodes from the keyboard).

The good thing is that both apps can easily import the stuff from the other one and you can simply test both side-by-side to understand which one you prefer.

Roam Research

Roam Research is a “note-taking tool for networked thought” that has extended the feature set of a basic outliner by miles. It’s still an outliner and everything you type is a bullet-point. But the app doesn’t want you to create a defined hierarchy – instead everything is supposed to be dynamic. Roam allows creating new pages by just writing something inside square brackets (wiki-style), see backlinks (all the nodes and pages linking to the one you’re seeing right now), and block references (linking and repeating a specific node to provide context). The app wants you to build your own private wikipedia for your knowledge.

Roam Research has one of the steepest learning curves I ever saw in software. It allows you to do almost anything and that’s overwhelming. Thankfully, it has a very opinionated part of the workflow – “Daily Notes”. The idea is, instead of trying to find the perfect place in your hierarchy for all kinds of notes, you just type everything in the note titled with today’s date and add links and tags to connect those items. You can log your morning run, mention that you talked to a long-time friend, and type favorite excerpts from the book you read – and everything would magically fall into place. Roam also has some more unexpected features, such as Kanban boards, Pomodoro timers, and programmable graphs.

The app gathered a massive following – check out the #roamcult hashtag on Twitter. Roam doesn’t have a mobile app yet, its web version is relatively slow – it might take 20 seconds to open it for the first time but even then jumping between notes causes a noticeable delay. It’s also the most expensive at $15 per month.

What Follows

Maybe at this point Roam Research doesn’t have as many users as other outliners but it definitely has the most vibrant community expanding its ideas and coming up with usecases.

I believe it’s going to be beneficial for the whole space:

• More users get the concept while not everyone might need a powerhouse that Roam is.
• Other apps see the outlining approach validated and new features proven useful.

The team behind Dynalist has already added backlinks to their core product and even launched Obsidian, their take on a knowledge management app that’s based on local Markdown files. Since it’s not an outliner I won’t be talking about it much but you can read some of my thoughts here.

In my mind, the perfect outliner should nail the core workflow, have stellar tagging and filtering views, but also enable quick entry from other apps and services, moving the nodes around through keyboard and native mobile apps. I imagine it to be a mix of Workflowy and something like Things 3 taking the best of both worlds. If you ever stumble on something like that please send it to me.

Superhuman is an app that has been reviewed and discussed from top to bottom, we’ve seen countless mentions of “Superhuman for X”, discussions of “consumerization of enterprise” and I saw Rahul Vohra’s Product-Market Fit score estimations in dozens of startups’ decks at this point.

But I always felt that one topic has remained largely in shadow: how Superhuman moved the Overton window of apps pricing.

A Bit of History

Back in the days software was quite expensive. The list price of Photoshop 1.0 for Macintosh in 1990 was $895 (that’s $1,767 in today’s dollars). In 2011 it was priced at $699. Not everyone paid those prices of course. I even had a theory that Adobe specifically wasn’t fighting the piracy of their products (their protection was very basic) for individual users because it acted as a top of the funnel. Teens would learn Photostop and Illustrator to draw and publish something on DevianArt. Some of them would become designers and come to a workspace expecting to see Adobe products there. All good! In comparison, local Adobe offices were pretty adamant about piracy at workspaces.

Generally, software was distributed via the sharewave model through various catalogs and websites. Even smaller utilies could cost $10/$20/$50 because not a lot of people would discover them in the first place and even lesss people would pay. “Professional” software was expensive but individual users weren’t the buying personas for it and the prices didn’t matter that much.

Then came the mobile. Apple AppStore redefined the pricing of mobile apps and games. Every month millions of people were buying new smartphones and becoming new users for the apps in the store. Trying to compete for them, the developers have driven the costs way down. Now people expected apps to be either free or a couple of dollars – $5 is already quite expensive!

But then somethign changed. The market has become saturated. People are still buying new phones, but less often – and they already have previous iPhones (or Android phones) and had the apps they needed. One-time small purchases were no longer able to drive the revenue.

Apple initially offered subscription pricing on AppStore in 2011 but only for content. Subscriptions for apps appeared in 2016. Meanwhile, Adobe switched to SaaS-only model in 2013.

Since then multiple old apps, including Fantastical, 1password, Ulyssess and many others have switched. Subscriptions incentivize developers to care about their existing userbase and provide recurring revenue to fund the development.

In 2016 Recode wrote that “The app boom is over”, an editorial in TechCrunch stated that “The air of hopelessness that surrounds the mobile app ecosystem is obvious and demoralizing”, and The Verge wrote that “the original App Store model of selling apps for a buck or two looks antiquated”.

Originally, the subscription pricing wasn’t that scary. Things 3 for iOS/iPadOS/MacOS costs a few years of Todoist Premium costs, which makes them relatively comparable. The developers weren’t ready to charge much more for consumer-grade software.

The New Age

Superhuman shocked a lot of people with its $30/month pricing for an email client (that’s $360 per year). I believe that this greatly extended the “Overton window” for consumer software pricing and enabled other companies to charge 2-3x more of what they initially did.

There have been a few email clients before that and most of their either:

• Were acquired by large tech companies for reasons unrelated to their core business [1], [2]
• Failed drastically [1]
• Were noticed selling insights based on users’ communications [1]

Turns out it’s very hard to make a good email client. Even while they were alive, most of those apps weren’t polished enough.

What if you can do a better job but need to charge more money? You probably need to ask if there’s a market for this. Designers would spend their entire days in Photoshop or Illustrator. There are different pricing plans, but a monthly single-app subscription is about $30. What if the user spends an entire day in email? Does it worth it? I guess so.

What’s even more interesting, such pricing allows the products that earlier could only be small utilities supported by indie teams turn into venture-backed companies. Regular pricing, especially one-time purchases, can’t really enable the margins and growth rate needed. You can think of Evernote that wasn’t able to recover to its early expectations.

Now we’re seeing dozens of old and new apps embracing higher pricing. One of the common aspect is that you can technically find more primitive ancestors for the newcomers that were simpler and charged less.

• Roam Research, a “a note-taking tool for networked thought” that became quite mainstream on Twitter recently, is touting a price of $15 a month as they come out of their current “early-access” status.
• Sunsama, a better task management and calendar app, is $20 a month ($16 with an annual discount).
• Fanstastical, a great email app for Apple’s platforms, used to cost $49.99 for a Mac app as a one-time purchase, now it costs $5 a month ($40 a year with an annual discount).
• Linear, a better issue tracker, is expected to be aroud $10/month when they come out of beta.
• Command-E, a less hacky Alfred for Mac, is expected to be aroud $10/month when they come out of beta.

If you have any other examples, please share, I’d be happy to add them.

What’s Next?

The tools we use in our digital lives are mostly subpar. They rely on the decades-old principles. Consumer software has been optimized for the mass user resulting in the lowest denomiator effects – Gmail is used by 1 billion people so it has to work for 1 billion people. Enterprise software instead has been optimized for entire organizations and the buying persona. The CIO might love the product but it’s not necessarily the best for individual contributors.

There are now two combining effects: the widely discussed consumerization of enterprise, meaning that software is adopted bottom-up and optimized for the experience of individual users, but also the expanding realization that companies can charge way more for personal productivity tools – and use that revenue and expectation to build better tools.