November 27, 2022 by Jan Martin Borgersen
As Twitter burns, I've spent the last few weeks taking inventory about how I use social media, playing with my own Mastodon instance, and rebuilding my feeds on Post. These are my quick hot-takes:
Mastodon has built the "easy" part of Twitter, probably the first 10%.
Mastodon is a better replacement for Facebook Groups than Twitter. Partly because of #1, but mostly because Mastodon was built for communities, not for planets.
Post.news is trying to solve the most important primary use case of Twitter in a better way. When most of your users are readers who use the platform to follow news and influencers, how do you build a business model that:
I'm bullish on Post.
Let me elaborate, by first being critical of the open-source solution that's been around for eight years. What's the 90% of Twitter that Mastodon hasn't built?
It's simple to describe, and hard to build: moderation and scale.
Moderation is REALLY HARD, and it gets harder as communities grow and become more diverse. Mastodon's approach is like performing surgery with a sledgehammer, without a legal department. In three weeks of paying attention, I have seen entire new instances be de-federated by the largest instances because new mods, who have never done this before, made some mistakes. I have seen people of color get immediately harassed by trolls, or suspended for speaking truth to power. I've seen opinion bloggers suspended for spreading mis-information before they posted anything. I witnessed one member of an under-represented group complain that a "toot" was referenced by a blogger/journalist without "asking for consent." (I'm glad we expect consent nowadays, but toots are public posts on the Internet.) Mastodon may not have the same bot/troll/game-the-system problem that Twitter has (yet?) but it certainly carries its own set of issues.
What is the proper line between a journalist and a blogger? A reporter and an influencer? Does it matter? How do you validate identities, how do you score their reputations? How do you detect bots? Who calls the shots, and how do you hold them accountable without punishing the innocent users on their instance?
Meanwhile, there's no automation for discovering copyright violations, no established path for DMCA notices, no inherent liability protections for the volunteers running their servers out of their basements.
Moderation is REALLY HARD, it has to be done carefully, and it has to be backed up by lawyers on-staff. This is not something I trust to open-source engineers to get right by themselves.
Scale, in the sense of supporting exponential numbers of new users, is also hard, but thankfully we've figured this out over the last 20 years. Mastodon's version of scale is to spin up new instances, but each instance effectively has an upper bound on how many users it can support, thanks to both architecture and human resources needed for moderation. Neither is built to handle DDOS attacks, an influx of bots, or communication across lots of nodes in the Fediverse. Sure, a human can flip a switch and turn off new sign-ups, or de-federate a problematic node, but that's still the sledghammer approach.
Spin up the Docker file for a new Mastodon instance and you get a nicely packaged stack of Nginx, Ruby on Rails, Postgres, Redis, and hooks for static asset storage and mail processing. This is a few important steps forward from a typical web application circa 2005, but not that many.
To scale for anything over 100K users, you first need a load balancer in front of a pod of Nginx/Ruby servers with a shared Redis, and a distributed database. These aren't hard to set up nowadays (and you can accomplish this with Mastodon), but you need someone with some cloud architectural experience to make it happen right.
To scale for anything over 1M users, I would jettison the Rails App in favor of an event bus and a set of microservices, plus a flushed out CI/CD pipeline for everything, including serving your compiled, static browser-code from a CDN. This is a re-write of the Mastodon code.
Somewhere along the way you will reach a point where the chronological feed of posts is too much of a firehose to be useful. You need a machine learning algorithm.
Sooner or later you will want to be truly global. Now you have to start being careful about regional laws, especially in Europe. You need to translate/localize your strings, and you probably want automated translation plugged-in for user content. This, of course, opens you up to all sorts of NEW content moderation issues, since you're now dealing with people who don't speak your language, and governments who have their own ideas about what's legal or not.
We are well past Mastodon ... but perhaps that's also the point of Mastodon.
I don't mean to argue against Mastodon, just to explain why it is not the Twitter replacement folks are looking for. In fairness, I have great respect for what has been built, and the folks who have built it. The Fediverse is a thriving place built on the backs of volunteers. ActivityPub, in particular, I think is an important protocol for the future of mass communication. Mastodon itself just has issues of moderation and scale, and I've yet to see an open source project really take off without corporate sponsorship.
Mastodon was built for communities, not planets, so I believe it is a better compared to Facebook Groups or Pages than Twitter. It fits in an interesting niche. It is easy enough to set up, and it is flexible enough for moderators to create the environment they wish. There's a reason so many other social start-ups (CounterSocial, TruthSocial, etc) have begun by forking the Mastodon codebase.
I don't want to dwell on this point, or defend it too much, mostly because I don't know that Mastodon is the best answer for community use cases. It lacks features needed by small groups, and the federation bit is a questionable "feature" for a real walled garden community. But it is open-source and pretty solid for what it is, which makes it an important character when talking about social media.
I like Noam Bardin's vision for Post.news.
While its easy to hear the 10% of Tweeters complain about Mastodon and Post, Noam hasn't forgotten the 90% of Twitter users who are there to follow, to listen to their influencers, and to stay up on breaking news stories. I am one of the 90% who almost never tweets. I use Twitter for news and opinion, and to follow things like natural disasters. I don't engage, and I ignore comments, because I have learned to be careful of the outrage cesspool that is Social Media today.
Noam also recognizes a problem with current media in that publications need to generate income to stay alive, but when they put up $10-20 paywalls, 98% of consumers don't pay. So, the media outlets cater to the 2% who do pay, and you get skewed reporting. Most importantly, this model is terrible for real people who want to read multiple sources to get varied coverage of news, yet cannot pay high subscription costs to all of them. Post is trying to address this flaw in today's business models by creating a micropayments solution where cusomters pay for what they read, and publishers increase their audience. (citation: Noam's 11/21/2022 interview on Pivot)
If Post can hit this sweet spot, it's going meet the most important reason people flocked to Twitter. A planet-scale public social media and news platform should: