Social Media and Advertising


Instagram announced their new "Threads" app yesterday, with Mark Zuckerberg talking about having a Twitter-like platform with 1B+ users, and hailing the app's rapid adoption of 10M users within hours of launch. Meanwhile, Bluesky has 250k users and is working on...scalability. Every new platform is focused on scale, exactly like the previous platform, because they are all funded by advertising. To make this work, they'll always need:

  • Reach
  • Engagement

Reach will always be about ease of onboarding, low cost, and scale. It's inherently about including everyone, regardless of whether or not those folks want to talk with each other.

Engagement will always be about finding and surfacing the most outrageous material: violent, scary, controversial, weird, incorrect.

When you combine these two, it's sort of obvious why social media trends towards being a mess: it shoves everyone into a giant pot and then takes the worst of what people are doing and puts it in front of everyone. And every platform, whether it's Facebook, Instagram, Google+, Twitter, or Bluesky follows this same pattern. As long as you're using the same exact recipe, the result after baking will be pretty much the same.

It also leads to an unsolveable problem: when you stick everyone together and promote only the most outrageous posts, endless debates about what should be allowed are inevitable, since different people have different standards and tolerances. Nothing in history suggests that some platform is going to come along and find exactly the right balance here: different groups just have different standards. And so social media consistently tries to solve this by using the lowest common denominator, putting in safeguards and restrictions for every group that has some concern: Muslims want to ban pictures of Mohammed, Christians want to ban pornography, China wants to ban criticism of the government, Elon Musk wants to ban anyone who tracks his jet. Of course, banning all sorts of content requires judgment, and AI simply isn't there yet (heck, most people can't agree about what should be allowed!), so any successful platform needs to have moderators, either paid (usually the case) or volunteer (in cases like Reddit), that try to enforce bans correctly. This enforcement then leads to issues of control, and who owns what content, and what the content can be used for. This has become particularly relevant as companies are training AI on social media posts, since they are a ready supply of human-generated writing.

And so the endless debates around misinformation, banned content, moderation practices, privacy controls, and monetization churn on, making no progress. If you look at past social networks, there are a host of related issues that have been contenious: real name policies, the meaning of a verified accounts, fact-checking / labeling policies, the "algorithm" (how the feed is ordered), and rules governing suspension and banning of accounts.

One solution is to abandon the recipe. Consider twtxt, a regressive social media app that is nothing more than a text file sitting on a server. A person's identity is the URL to that file, and the file can be fetched by others' twtxt clients at-will, in the same way a browser would fetch a web page or a feed reader would fetch an RSS feed.

There is no promotion of outrageous content, since it's just text files with a post-per-line. This means no moderation is needed, since every user has complete control over what they see: they will only see messages from feeds they follow, and unfollowing a feed is easy. It's trivial to create a client that automatically filters out content with particular keywords. Rather than turning to some international corporation to tell you what you're allowed to read, you can just decide for yourself. This means that each user has control and can dial their own filters as desired.

The flip side of all these advantages is that it doesn't work like advertising-funded social media. It's smaller, slower, and less volatile. You have to discover new feeds by following links in posts from those you already follow, since the platform won't do that work for you. It's less convenient, so it will never gain mass adoption, but the filter it applies selects for folks that are willing to do a bit more work. This is probably a really good thing, and seems like a fair trade besides: for folks that think Twitter should do a better job of moderation, using twtxt gives you complete control!

Critics might point out that twtxt will never be as big as Twitter or Facebook or Instagram, and I agree. The difference is that critics see the size of those platforms as a feature, and I see it as a bug.

Which brings me to my thesis: I don't think we're pursuing the goal of social media incorrectly, I think we have the wrong goal. We're looking to have an authority feed us information, and I think we need to instead take more individual responsibility for how we gather information. The first step is to take control of what you read by choosing it, rather than having it fed to you, and the most effective way to do this is to stop using products from companies that are primarily funded by advertising.

In short: twtxt seems to be the "just publish a website" version of social media. It's extremely simple and effective at letting folks stay up-to-date on accounts they care about.