Cohost September 2024 Financial Debate Retrospective: Making Sense of The End

[osteophage]

If there is anything about Cohost that you think is worth replicating, then it's worth understanding why it folded in the end. In the wake of the shutdown announcement, I've seen several competing narratives and explanations for what happened, and sorting this stuff out matters for figuring out what kind of lesson to take from it. So in light of that, this retrospective is my attempt to log and detangle some of those narratives, starting with a brief recap to bring you up to speed.

Crossposted to Pillowfort. Note, feel free to link to this post. Linking is the way that I prefer for people to share my stuff on other sites, if at all.

A Recap of the Cohost Financial Timeline

The Anti-Software Software Club (ASSC) launched the site known as Cohost in the year 2022, as bankrolled by an individual private lender. Over the next two years, the company would twice announce limited financial runway amounting to a month or less, each time inciting debate among the userbase about the site's financial future.

The first instance occurred in the summer of 2023. Mid-June the company announced that "at our current burn rate and bank balance, we will not make July 15 payroll," but not to worry—they had a plan. The plan was to borrow more money from their original funder, which they did. Still, for some, this announcement raised concerns; among others, those concerns were met with emphatic optimism. For example, a friend of the devs posted his confident speculation that "[the funder] will continue funding Cohost as long as there are positive signs of growth." And then there wasn't, and they didn't.

In mid-March 2024 the company announced they had lost contact with their original funder and that they expected to run out of money in early April, i.e. in about three weeks. This announcement alarmed the userbase enough to motivate the Cohost Plus push, in which users rallied to persuade other users to donate more money via subscriptions to Cohost Plus. This effort, while well-intentioned, was not itself enough to close the gap.

Instead the company got a second wind when a new, second funder stepped in to bankroll them on the condition of consistently posting financial updates. Subsequently the financial update for July was nearly forgotten. In the ensuing debate over the site's precarity, optimists objected to others figuring that the site was in danger or that it might shut down. Less than two months later, on September 9th, 2024, the company announced the site would shut down.

For further reading on the company's financials leading up to March 2024, see this much more extensive Cohost Financial Retrospective.

 

Cohost Shutdown Narratives and Mythologizing

This post will cover the following non-exhaustive list of narratives about the Cohost shutdown:

• Stripe Did Them In
• Centralization Did Them In
• Lack of User Support Did Them In
• Capitalism Did Them In
• Lack of Funds Did Them In
• They Successfully Broke New Ground

 

Narrative #1: Stripe Did Them In

Blaming Stripe for killing Cohost is a narrative that draws on the May 2024 Financial Update, where the company announced they had to abandon one of their monetization plans after finding out it was against Stripe's terms of service. The announcement characterizes the rule as something that Stripe suddenly added, as though ASSC were all clear to use Stripe for a subscription and tipping service before Stripe pulled the rug out from under them them. This notion is misleading in two different ways, one specific to Stripe and one not—but let's start with Stripe.

The changes this year to Stripe's list of restrictions and prohibitions reflect an effort to clarify which types of business were already restricted in the first place. Back in February 2022, the Restricted Businesses list began with financial products and services, such as "money transmitters," and the Prohibited Business list already included "Pornography and other mature audience content (including literature, imagery and other media) depicting nudity or explicit sexual acts." In January 2024, a new section was added to the Restricted Businesses list for "content creation related transactions," including "content-related tips and gifts," and the list of Prohibited Businesses was updated to include "Transactions that provide compensation to creators without an underlying piece of digital content associated," i.e. blanket tipping and subscriptions. These explicit rules are something that ASSC had the opportunity to notice or be advised of well before May 2024.

However, the problem here is not one that could be solved by just switching to a different payment processor. What ASSC was proposing was an entry into the industry of financial services, which presents a higher barrier to entry than other kinds of business on account of all the laws about money laundering and fraud. Doing your due diligence about money laundering and fraud is a serious undertaking and calls for a level of legal and accounting expertise which... suffice to say, did not appear to be among the strengths of the ASSC team.

 

Narrative #2: Centralization Did Them In

A portion of the shutdown commentary has asserted that decentralization is the only viable way forward and that the big mistake was in not federating Cohost with ActivityPub or AT Protocol. I'm gleaning this narrative from remarks about how federation "could have helped their revenue problems indirectly," perhaps by "delegating the expensive aspects of the services to microservices within the protocol." This read of Cohost's closure seems to figure that 1) decentralization would offload some of the server costs, and/or that 2) decentralization would offload some of the labor of moderation. I think this is the wrong read because server costs and the quantity of moderation reports were never the core problem, financially speaking, and so decentralization would not be enough to fix it.

Trying to offload hosting costs to other servers would not have made the difference in anything because hosting costs were not a proportionally significant share of Cohost's expenses. This was stated outright in the March 2024 Financial Update where they wrote "our largest expense remains payroll by a long shot." Payroll in this context means the site owners paying themselves $94,616/year each*, which in terms of payroll expenses for the company came out to, we're told, approximately $36k per month. Meanwhile the hosting costs were already being outpaced by revenue, and if you actually look at the numbers you'll see that no amount of cutting back on hosting could have made them break even.

(*One of the owners took a 40% pay cut announced in June, but even so, payroll expenses remained the largest share of expenses.)

Trying to offload the moderation burden to other servers would not have made the difference in anything either because moderation was a fixed cost. There was one moderator, paid a salary of $94,616 per year, on a fixed basis—not an as-needed hourly wage. This means that no quantitative change in the moderation burden could have possibly solved the shortfall.

 

Narrative #3: Lack of User Support Did Them In

This narrative attributes Cohost's downfall to other people failing to get behind their vision. Specifically what I'm referring to here is this idea:

instead of being an inspiring story about how things could be different, cohost will forever be a cautionary tale to never try anything that isn't corporate sludge, because the public would rather piss and moan and let perfection be the enemy of good than rally behind someone who's actually trying to improve.

To me this narrative does not align with the fact that Cohost's conversion rates were very high. The exact number fluctuated from report to report, but across most reports it was often in the double digits, which by all accounts is exceptional. If you want a subscription rate any higher than that, then I figure you might as well take the site paid-only.

 

Narrative #4: Capitalism Did Them In

Among those who had high hopes for Cohost, a common narrative seems to be that social media is doomed by the economic order to live or die by venture capital and ad-driven exploitation. I am inferring this narrative from remarks about how Cohost was "killed by capitalism" and that Cohost was doomed to fail due to the fundamental way our society is organized, or that a company running a good website is at odds with being monetarily successful enough to survive, or referencing capitalism as the reason why "you can't run a site like cohost on recurring donations+artist alley alone," as the business was "designed basically not to make any [money] by virtue of having virtues."

You may recall this is the same narrative predicted back on July 4th, 2023:

When cohost shuts down the blame is going to be put on a lot of things. They're going to be described as too good for this world. A plucky little queer co-op that wanted to be ethical and thus capitalism smashed them down.

This narrative can be contradicted by pointing to other comparable sites that have managed to stay afloat, and the strongest case study on that front is Dreamwidth. Dreamwidth has been in operation since 2009, has never taken funding from outside investors, and has never run ads. Its source of revenue is, in broad strokes, the same as Cohost's: selling premium features directly to users. This exclusive dependence on user goodwill has kept Dreamwidth going for over a decade and counting, which already shows that it's possible to make it work.

 

Narrative #5: Lack of Funds Did Them In

Lack of funding is one of the causes explicitly named in the September 2024 announcement, so in a way, this is (a part of) the official narrative. It's also true, in a barebones technical sense. With that said, "lack of funding" is a relative concept—relative to expenses—and how you assess those can go one of two ways.

By the standards of "the tech industry" (here meaning venture capital and monopolies), ASSC's costs were very low. Within this version of the narrative, their largest share of expenses—the site owner salaries—gets described as "well under market rate." Treating those salaries as a fixed cost then raises the question of how else the company could make ends meet, and for people who expect six figure salaries, the conventional answer in this line of work is data collection and ads. Since that kind of business model is what Cohost had already expressly positioned itself against, that leaves the other point of reference: niche alternative social media sites that refuse venture capital.

By the standards of niche alternative social media sites that refuse venture capital, ASSC was spending beyond its means. For a benchmark, again, the strongest representative is Dreamwidth, and Dreamwidth's cofounder has confirmed that ASSC's numbers are not realistic for making ends meet. Making ends meet in this business, the business of user-focused social media, is possible if you make the corresponding sacrifices, which is not what ASSC chose to do.

What sealed Cohost's fate in the end was ASSC's determination to try and have it both ways. They wanted near-six figure salaries and they wanted to eschew the kind of revenue sources that make those salaries possible. Trying to combine both can work for only as long as you can get outside parties to keep lending you money—so in a way, that's what killed Cohost in the end: not just a lack of funds, but the finite supply of lenders.

 

Narrative #6: They Successfully Broke New Ground

Setting aside the financial aspect, this narrative celebrates Cohost as a bold new experiment that proved something to the world. Salvaging some form of meaning from the project is an understandable impulse, but what I'm talking about here is not just people talking about what the site meant to them personally. What I'm talking about is the characterization of Cohost as "the first and only social media site in decades to be exclusively for its users," a little slice of history which "taught us yes, people still want a human, functional internet," which "started the path to proving [a less-scummy web is] possible," and which "pioneered" the "avoidance of dark patterns."

These are the kind of statements that would be true if Cohost had no predecessors, which it did. Cohost was already predated by Dreamwidth, Pillowfort, and yeah, even Waterfall (not that anyone should be emulating Waterfall). All of these sites chose strategies focused on the user experience rather than spying on users on for the benefit of advertisers. None of them have featured dark patterns, i.e. deceptive patterns, i.e. elements designed to manipulate users into subscriptions, ad clicks, or data collection permissions. Refraining from that kind of user exploitation is a trail long-already blazed by others, and another site following in their footsteps is great, but let's not treat something as something brand new when it's not.

Recognizing Cohost's predecessors and financial peers is consequential for figuring out what kind of lessons to take from it, and figuring out what lessons to take from it is, in a way, the company's dying wish. Here I'm referring to this sendoff post by Jae, one of the four members of ASSC:

cohost was built off of the lessons learned from a life spent on social media, looking at the mistakes others had made, looking at what they did right, and trying to build a platform that combined the best and accounted for the worst.

i hope that at some point in the future someone will look at the mistakes we made and the things we got right and do the same to us.

If there are lessons to be learned from Cohost, then it's worth sorting out what it was from what it was not. Cohost was not the first ethically-oriented user-centered website. Cohost was a website that some portion of its users enjoyed. Those users could have enjoyed it for longer if it had achieved sustainability, and sustainability has been proven possible by others who can be looked to as an example for how to do it again.