36
votes
cohost.org to shut down by the end of 2024
Link information
This data is scraped automatically and may be incorrect.
- Authors
- cohost dot org, Siph, Lilac, Damien, Siërra R., "Botfly Mother "Inhaling The Sodium Vapors"", ISO standard fire lizard, Kotetsu, mattachine meow machine, Katja [💼⚖️🤵] Antares, a 🐈🐉., Otter Arriving on Platform 1, phat_sumo, Lily🌸, colette, Saturn🪐, 🌷 doodlemancy 🌷, Caique Assis, The Mysterious Choster Known Only As "VampireExpert", Omio (et. al), 𝕷𝖚𝖈𝖆𝖘, Noel Belmonte, trainboy, The Void Which Stares Back, But In A Cute Way, hecate, Luisf, Balina‼, Noxulous of Fyth, Phorm The Vixdjinn, JP, bat, Tamber
- Published
- Sep 9 2024
Looking at their financials, it seems like, at least architecturally, they got a bit too big for their britches. I'm not sure what stack they're running, but costs of -$40k just for infra seems... very high? A cursory google search seems to indicate they have around 20k active users, which would be almost $2 for each user!
It's hard to imagine that whatever they have isn't very much overkill for their amount of traffic. For these kind of experimental sites, I think priority #1 is costs, costs, costs - it needs to be as barebones and efficient as possible, since you don't have the financial backing a company would have.
Keep in mind that they're ignoring a few best practices - especially when it comes to user media compression, or rather, the lack thereof. If they're doing this with a background as web design and dev hobbyists and simply don't know how to scale a service, I can believe these numbers.
It does seem absurd given that (1) computers are very fast (provided you don't do anything really silly) and (2) if you consider that Tildes which provides a service that is not totally dissimilar runs on a single server (iirc).
Tildes doesn't host images, that does make a huge difference. In fact, for any website that allows user uploaded content (other than text) this will be the biggest cost as far as infrastructure goes.
So the comparison you are making isn't equal. Probably not even a little bit close to equal, if I am being honest.
Let's round the estimated total number of users on Tildes up to 25,000.
Horizon is an image hosting service I've been using and enjoying quite a bit. Its free tier gives each user 500MB of storage.
Let's pretend, hooray!, Tildes just enabled native image posting as a privilege for users, not entirely unlike the ability to edit tags. Now, users who ask for the privilege can post images directly into text topics they write (like they would on their own blog), and those images are hosted and stored on the Tildes server.
Let's assume 10% of Tildes' total userbase takes full advantage of this feature, and maxes out their 500MB storage limit.
So that's 2,500 users * 500MB of storage, for a total of 1,250,000 megabytes, or 1,250 gigabytes, or 1.25 terabytes of storage used.
That's a lot, but also, kind of not? If we were just talking about a regular desktop computer, I think this community could fundraise $500 to get two 4 terabyte NVME drives to stick in the Tildes server. One drive for image-posting, one drive as a back-up. That would be enough storage to last for years at Tildes' current rate of growth. Unfortunately, getting storage onto a server often isn't that simple or cheap.
However, I vaguely remember @Deimos or @Amarok saying something along the lines of "Tildes owns the hardware it's hosted on", but I might be wrong, and that still doesn't give us an actualy idea of what the costs would be. My point is though, those kinds of storage costs for Tildes might be cheaper than people expect.
Storage isn't the only cost to hosting images though. There's also costs for processing or compressing images, and bandwidth for transmitting them. Those might cost even more than pure storage over the years.
Then finally, there's just all the little extra bits of technical attention that images require. Should they be automatically resized or compressed? Should that compression be lossless? How high should the compression level be? What Exif data should be stripped out? Should users get to choose what Exif data to strip out? (Some users might post nature photos and explicitly want the location that photo was taken from to be available, for example.)
So yeah, there's a lot that goes into hosting images. Cohost basically said "fuck it we ball" and let anyone post images at their original quality however much they wanted, and I'm sure that was a significant cost. I doubt it was the reason behind their unsustainable financials, but still worth mentioning.
Tildes is on rented hardware - one fairly average, barely awake server, which is all you need when you aren't pumping images, video, and 30mb of tracking scripts to every user with every page refresh. :)
The storage and bandwidth and system complexity hosting images/video/etc are just not worth the hassle - let other websites handle that and just link to them. The text is all you need for the community to materialize, and it's probably better overall for one's mental health as well.
While one could certainly set it up to share other things, I've been down that road before myself. It doesn't take long until people are uploading everything from lord of the rings to clown porn. Then come the copyright claims and subpoenas. Ugh.
I think if I were to set something like that up, I'd build the entire back end on bittorrent or some other shared hosting tech. One can facilitate sharing without dealing with the hosting and legal costs.
This is something people overlook. Generally speaking, storage is much more expensive for servers. Even more so if you don't want to risk losing all data, so you need basically twice the hardware for redundancy reasons.
If you own the hardware and have relatively easy access to it, then it might be a bit cheaper.
Possibly, although I think that even if the hardware is owned it will be more expensive than most people do expect.
Generally speaking, though, if you don't own the hardware, storage will be the biggest cost factor.
Once you start factoring in bandwidth costs, images will be a lot more expensive. Certainly with a large enough user base or if they are linked elsewhere on the internet and go viral.
I wouldn't be surprised if it was a big reason. There is a reason that there are so few image hosting services, certainly free or cheap ones. The ones that are around are often ad riddled, compress images a lot, don't let you link to images directly, offer very limited free tiers (if it all), etc, etc. All of which is related to keeping costs down as low as possible because hosting images simply is not cheap. Or, to give a different example. Note how Horizon on the free tier as well as the paid tier has a little asterisk next to it. When you scroll down it says "Unlimited usage is subject to fair-use guidelines". Which is there for good reason and will allow them to limit the impact of images that blow up in popularity.
Kind of but also not really. There's nothing stopping you from slapping a few WD Green SSDs in there and calling it a day. Personally I wouldn't go that cheap but when I worked for an MSP we would absolutely buy consumer tier SSDs for the servers we managed for our clients. Keep them in RAID10 or 5 in case one goes out and you're set. Like, yeah, you could just buy the server from Dell or whoever preconfigured with your drives and then your paying 4x the cost, and you could buy server tier storage guaranteed to last until the heat death of the universe, but there's very little practical upside for the cost. As long as you have a backup that you can swap to if/when you need to take it offline then there's no real reason to dump cost into storage.
Sure, they're not the same thing (I imagine Cohost will need more resources), but spending $2/user/month is an incredibly huge amount of money for a social media platform. Although I think the $40k figure includes their payroll (of 4 people) so that makes a lot more sense.
Sure, I am just responding to the comparison you made to Tildes :)
A note, someone else said that the number of users the $2 is based on might not be correct. With then number they found, it comes down to roughly 0,20 per user.
I'm fairly certain it includes dev salaries (which would mean they are paid below market rate), the post states that they've only stopped taking a salary "as of today."
I would be really surprised if it did. Considering they had 4 devs, that's not "below market rate", that's like "below minimum wage in most developing countries" rate.
Is that really so ? $40.000 / 4 is $10.000 per month and $120.000 per year. That seems quite a lot higher than minimum wage even after tax. Or am I missing something completely obvious, I'm not sure how you got that.
It's not a monthly expense report - it's a year-to-date one. 40k for 4 people + server for 8 months of runtime would mean that whatever the 4 people were being paid would barely be considered income.
No, I can confirm it's a monthly expense report. One indication of this is the fact that some numbers wouldn't make sense if it was year-to-date. Income, for example, couldn't possibly go *down * from July to August.
How is it a year-to-date report ? If that was the case why would this month only have a $600 increase in expenses when it averages at 5k for the previous 8 months. And the previous July financial update had expenses at 58k for June so it wouldn't make sense for it to go down to 41k then.
Wikipedia says it had 203,805 total users in November 2023, but doesn't cite a source:
https://en.wikipedia.org/wiki/Cohost
That's about 70,000 more total users than they had when, in February 2023, Business Insider claimed the site had 20,000 active users: https://www.businessinsider.com/twitter-alternatives-social-media-discord-mastodon-hive-cohost-truth-2023-2#cohost-5
I agree though that the costs seem high, whatever the actual number of active users was.
Well, MAU and total users aren’t the same. Per Wikipedia
It seems like the amount of MAU didn’t go up all that much, even as the total users do. I think that makes sense, you’d probably see a similar pattern with tildes. A lot of people will bounce off these small indie social networks.
Oh, I totally skipped over that section of the Wikipedia article. My bad
Oh god, image compression is websites 101/as far as I'm concerned. It is enabled by default on Hugo, for instance.
Found this statement in the comments from one of the staff very interesting/strange (and maybe not a great idea?). Sad to see something like this couldn't be open-sourced. Almost seems counter to their entire mission/manifesto that complains about the "big companies" and such... well killing any future your code could have in other hands that aren't big corpos seems ignorant
Presumably, if they had taken off and paid back the loan they'd be free to do with it what they wanted I would think.
That's explicitly what "collateral" means. They used the company (its assets, really, by my understanding) to secure the loan, which the lender is going to insist on because otherwise they're just picking up an unmitigated liability. When the loan is paid off, the lien on the collateral is released and full control returns to the owner. But consequently, if the company can't service the loan, the lender has the right to claim that collateral in order to recoup some of their losses. And because of these conditions on the loan, the company can't just sell or give away the collateral. (Although I bet you could get a team of accountants into a knock-down drag-out brawl by asking what effect open-sourcing a codebase has on its actual monetary value.)
Now, what is a bank going to do with the shattered remnants of a failed social media platform? I dunno, though my guess is "sell it for pennies on the dollar at auction". (Code is remarkably illiquid for how expensive it is to produce and how extraordinarily valuable it can be.) Where their code will ultimately end up, I would not begin to speculate, though "split among multiple owners and thus functionally gone forever" and "eventually open-sourced and/or dumped into the public domain" are both plausible.
The indie social media microblogging site cohost.org is shutting down at the end of this year. Like Tildes, I thought it was an interesting experiment in social media design, and it's a shame.
Sad news.
I wasn’t on cohost myself, but I was cheering them on from the sidelines. They felt like an unofficial sibling site to Tildes — we are to reddit as they are to Tumblr.
Impressively, they seemed to overcome the first major hurdle that novel online communities have in that they managed to gather an active user base and avoid becoming a ghost town. The few times I checked in on it, the site seemed like it had healthy activity.
A sustainable funding model is a whole other hurdle though. It’s a bummer they weren’t able to make it work.
Turning the focus inward: this is a bleak but worthwhile reminder that setting up a recurring donation is a great thing to do if you enjoy spending time here on Tildes.
Archive link: https://archive.is/1Oonr
I really wanted to like cohost but the culture never jived with me, and their lack of updates and financial transparency despite their continued communicated insistence on both, made it hard to trust the project was in capable hands. I’m curious to see how they handle data export, another long-promised and never delivered feature, now that the site has a scheduled death-day; I don’t have high hopes considering they’re already walking back promises about what the site would be “if” they ever shut down.
Was it another attempt at a Twitter-like?
It never felt that way to me. The UI has a more Tumblr feel to it I think. Here's an interesting post, if you want to check the site out yourself: https://cohost.org/mcc/post/178201-the-baseline-scene
I wonder how squabblr.co stays afloat.