Farcaster: Hard to build, easy to break, and the path to resilience

Farcaster got a lot of heat in the last few months.

From Defiant saying that Farcaster has only 10% of their reported DAUs:

Through Peter Szilagyi saying that FC users are primarily bots:

Up to the usual suspect crypto hater Liron saying that FC valuation is overblown:

So is Farcaster dead or overhyped? 

I’ve been active on Farcaster since 2022 when there were maybe two hundred active users. The last few months, were intense, to say the least. But what actually happened?

Here are my 2 cents.

Tens of billions of individual steps have to go right in the correct order to create a human. But only one thing has to happen to cause its demise. 

After just five weeks a human embryo has a brain, a beating heart, a pancreas, a liver, and a gallbladder. By birth, a baby has 100 billion neurons, 250 trillion synapses, 11 cooperating organ systems, and a personality. It’s staggeringly complex. Death, on the other hand, is simple. Most deaths—trauma, heart disease, stroke, some cancers, infections, drug overdoses—are caused by blood and oxygen deficiencies. That’s it. A disease itself might be complex, but the fatal strike is not enough blood and oxygen getting to where it’s needed. 

Making a human: incomprehensibly complex. The death of a human: really simple.

- Morgan Housel,"Same as Ever”

The human body is hard to build and easy to destroy and so are our relationships.

• An amazing house party can be ruined by one asshole.

• A lifelong group of friends can break with one big argument.

• And a supportive community can be destroyed by a few bad actors.

And Farcaster, as a social network, is not immune to these dynamics. But before destruction, there was creation. So let's start with building.

Farcaster took a long time to build. The company was founded in 2020, and since then - just like in the creation of a human being - many things have gone well.

The team has chosen the right niche (Ethereum builders). They have picked the right people (Dan spoke with the first 10,000+ users). They created the right culture and enforced it (friendly vibes contrary to Twitter dunking culture). They also iterated quickly and were brave enough to drop the ideas that didn't click. And it paid off.

In early 2023 Farcaster felt like a cozy small-town community, where you recognize all faces pfps around you. It was a perfect mix of founders, devs, investors, artists, and writers. And when I say mix, I mean it in the literal sense - people really engaged and mixed with each other. Some founders started building projects together, some artists co-created art with their audience, and I know even one couple that met on the Purple app.

It felt less like a social app, and more like a scene:

“Before there is a subculture, there is a scene. A scene is a small group of creators who invent an exciting New Thing—a musical genre, a religious sect, a film animation technique, a political theory. Riffing off each other, they produce examples and variants, and share them for mutual enjoyment, generating positive energy.

The new scene draws fanatics. Fanatics don’t create, but they contribute energy (time, money, adulation, organization, analysis) to support the creators.

Creators and fanatics are both geeks. They totally love the New Thing, they’re fascinated with all its esoteric ins and outs, and they spend all available time either doing it or talking about it”.

- David Chapman, "Geeks, MOPs, and sociopaths in subculture evolution"

Even a few months later Farcaster was so niche that when I mentioned it to cobie, he thought it was merely a blip on the radar:

So just like in a human body, tens of thousands of individual steps went right to create the Farcaster scene. But only a few things had to happen to cause problems. 

The scene was great but Farcaster didn’t want to just create a scene. They wanted to build a network for 1B users. And they had to grow.

So in October 2023, Farcaster went permissionless. This meant that everyone could create an account. That meant that the Eternal September began and we are going to see more MOPs

If the scene is unusually exciting, and the New Thing can be appreciated without having to get utterly geeky about details, it draws mops. Mops are fans, but not rabid fans like the fanatics. They show up to have a good time, and contribute as little as they reasonably can in exchange.

Geeks welcome mops, at first at least. It’s the mass of mops who turn a scene into a subculture. Creation is always at least partly an act of generosity; creators want as many people to use and enjoy their creations as possible. It’s also good for the ego; it confirms that the New Thing really is exciting, and not just a geek obsession. Further, some money can usually be extracted from mops—just enough, at this stage, that some creators can quit their day jobs and go pro. (Fanatics contribute much more per head than mops, but there are few enough that it’s rarely possible for creatives to go full time with support only from fanatics.) Full-time creators produce more and better of the New Thing.

- David Chapman, "Geeks, MOPs, and sociopaths in subculture evolution"

Many people were concerned about how going permissionless would impact our vibes. Would Farcaster lose its magic? Wouldn't we get flooded with low-quality posts? The Farcaster team was worried about it, too.

Some of us discussed the potential impact of the Evaporative Cooling effect:

Openness is a major driver of Evaporative Cooling. If anyone can join your community, then the people most likely to join are those who are below the average quality of your community because they have the most to gain. Once they’re in, unless contained, they end up harming the health of the community over the long term.

Communities that are allowed to select their members in some way are much more immune to Evaporative Cooling. Unfortunately, most viable internet businesses have no choice but to set their business model to open. The nature of most Web 2.0 businesses is that they depend on extracting a tiny bit of value from a large number of users and are betting on their fuck you exit from massively exploding in scale. Building a thriving community that tops out at 10,000 members over the course of 10 years isn’t going to pay the bills.”

- Hang, "The Evaporative Cooling Effect"

And the beginnings were indeed challenging.

Some people didn’t get the Farcaster culture. Some people were airdrop farmers posting "wowow" under each post, which were labeled "low effort replies" (which later became a meme). Some people weren't even people as they were bots.

So we did what all small towns do, to keep their identity - we taught immigrants about our culture. And it worked. Some people put more effort into their casts. Many low-quality posts just moved to channels. And bots weren't seen that much in the main feed.

After maybe 2-4 weeks, the situation felt under control.

And then Farcaster introduced Frames.

Many devs were excited about the new playground, and a handful of blog posts covered this primitive, including the one by Antonio Garcia Martinez. Many apps used the Frame frenzy to grow their user base. Frames have been the first in what has been aptly dubbed as programmable social.

Unfortunately, many Frames went viral by asking people to repost them. This meant that our feeds got flooded with a lot of content we were not looking for.

Coincidentally with the Frames launch, came the people who wanted to make money.

And when I say "making money", I don't mean builders who charge a reasonable fee for using their apps. These people were different.

A subculture at this stage is ripe for exploitation. The creators generate cultural capital, i.e. cool. The fanatics generate social capital: a network of relationships—strong ones among the geeks, and weaker but numerous ones with mops. The mops, when properly squeezed, produce liquid capital, i.e. money. None of those groups have any clue about how to extract and manipulate any of those forms of capital.

The sociopaths quickly become best friends with selected creators. They dress just like the creators—only better. They talk just like the creators—only smoother. They may even do some creating—competently, if not creatively. Geeks may not be completely fooled, but they also are clueless about what the sociopaths are up to.

Mops are fooled. They don’t care so much about details, and the sociopaths look to them like creators, only better. Sociopaths become the coolest kids in the room, demoting the creators."

- David Chapman, "Geeks, MOPs, and sociopaths in subculture evolution"

Many of these new people focused on memecoins.

The late 2023 Farcaster's memecoins were primarily pretty innocent jokes. $POINTS was just a random experiment that went out of hand, $SPAM was Borodutch testing the boundaries of protocols, and $FART was about helping PurpleDAO raise funds. Even Jacek treated $DEGEN as a fun side project and didn't predict it'd reach $1B+ FDV.

But then came people who tried to exploit the Farcaster network. Bot makers who decided to farm $DEGEN and other coins. Veteran rug pullers. And other typical crypto shenanigans. It felt like the whole Crypto Twitter was crashing our party.

And although many people made money on memecoins (including me), the app quality got worse. Some new users saw a spam shitshow in their feeds and coined an opinion that "Farcaster is overrated". Some old feed conversations moved to group chats because our responses were full of airdrop farming replies. Some regular users started using the app less.

Although this increased activity was net negative for most users (Geeks and MOPs), the DAU numbers went up.

And it was something that made people even more negative about Farcaster. When Dan shared DAU numbers, without the caveat that a big % of this traffic might come from bots, it generated a lot of pushback from the crypto community. And why wouldn't it, when we got PTSD after so many protocols artificially inflated their numbers to generate hype?

In a one-two-punch fashion, at the same time, Farcaster made a $150M raise. This turned Farcaster into the main character for a few days, which also meant that some people got angry. Some because they were jealous, some because they didn't understand Farcaster's business opportunity, and some because they had fair questions about the numbers.

In the meantime, slokh, who ran a popular alt client for Farcaster, decided to quit.

So all these things together - spam, inflated DAUs, and pushback after the raise, resulted in a big hit for Farcaster. The network that was being built for 4 years, got almost crushed in 4 months. After all, it wasn't immune to the hard-to-build/easy-to-destroy dynamics.

But it’s not the end of the story. 

Farcaster was experiencing growing pains and trying to find a way to scale. Thankfully, it could check how other social networks dealt with that.

"I joined Facebook when it was less than a million members. Since then, it’s managed to grow by a factor of 500 but the quality of my experience has dropped by only maybe 50%.

The reason why is because when some random person is participating in Facebook from Brazil, it has an absolutely negligible effect on my experience. Because every user only ever see their tiny corner of Facebook, every user is in direct control of their own experience.”

- Hang, "The Evaporative Cooling Effect"

And this was a big bet that Farcaster made.

They created channels that divide the user base and do not let bad actors destroy things for everyone. The channels existed before the network went permissionless, but since then the FC team doubled down on them.

In other words, Farcaster created "warrens":

“There are two fundamental patterns of social organization which I term “plaza” and “warrens”. In the plaza design, there is a central plaza which is one contiguous space and every person’s interaction is seen by every other person. In the warren design, the space is broken up into a series of smaller warrens and you can only see the warren you are currently in. There is the possibility of moving into adjacent warrens but it’s difficult to explore far outside of your zone. Plazas grow by becoming larger, warrens grow by adding more warrens.

These are the two fundamental patterns of social spaces. Every social space can be decomposed down to a collection of plazas and warrens. In Facebook, your profile, friends and newsfeeds are warrens but fan pages, groups & events are plazas. Twitter is mostly a warren with the exception of trending topics which is the one plaza. On forums, the front page and topic listings are plazas but each forum thread is a warren."

- Hang, "The Evaporative Cooling Effect"

And - in a way - it worked.

Even at the peak of airdrop bots mania, channels helped to keep some spaces clean. Each channel's moderator took care of their small warren, which was more scalable than the Farcaster team analyzing each post on the network. Moderators could set up their posting guidelines, use tools like automod, and ask users to pay a small sum before they post.

What about the plaza? On Farcaster, it was the Trending Tab. Since it aggregated the most popular casts from the network, it was my go-to place to check what was going on.

And for a long time, it was terrible.

Because the Trending Tab included posts that were popular, it was very susceptible to engagement farming by bots and users. So for a long time, it was full of Frames where you had "recast to use the frame", $DEGEN airdrop farmers, and just bots.

Since the Trending Tab was often used by new users, who had empty-ish feeds as they hadn't followed the "right people" yet, it had far-reaching consequences. As we said before, many new users thought that the spammy Trending Tab represents Farcaster and just left the network without exploring it too much.

Since then, the Trending Tab has been fixed and now it surfaces relevant content. But what about the responses from bots that swarmed our notifications?

The team introduced Priority Mode which means that - by default - you only see the most relevant notifications:

It definitely helped to reduce spam but it came with a price. And this price was the new user experience. My friends who used FC last year received quite a lot of engagement. Now most of their posts aren't too visible unless they have a Power Badge. Apparently these are not isolated cases.

The good thing is that Power Badges are not only distributed to the "old guard". Many new users (with FIDs 250,000+) already have one. So hopefully after some time the valuable users are going to be rewarded with more distribution.

And what about sociopaths trying to extract value from the network? Since the memecoin mania ended, most of them went elsewhere. Or maybe they stayed but I don't see their posts because of the channels and priority mode? Either way, they don't seem to be a problem anymore.

And although the engagement went down by a lot, it now feels more organic:

We could say that these hard few months were Farcaster's path to adulthood.

Frames, memecoins, channels, bots, power badges... The last few months were very intense for Farcaster.

The team made some mistakes on the way (like not fixing the Trending Tab earlier), but all in all, I think this period made the network more resilient. It's also a great lesson on how to deal with bots, airdrop farmers, and low-quality posts. Everyone working on an app that uses social features should take notes. We at Kiwi definitely do.

For sure it's not the last time the network is being tested. If they want to grow to 1B users, there are many challenges ahead. But this story made me confident that the team will be able to handle these problems. And whatever comes through those gates, will be taken care of.

PS: if you haven’t tried Farcaster in the last 3-4 weeks, give it a try. It should be much better than what you remember.

PS2: If you read this post and are an FC user, we can get in touch - my handle is @macbudkowski.