@[email protected] , I’m sorry to bother but is it really true? Are you paying almost $5000/month out of your own pocket?
If true, why? This is not sustainable. Don’t you think that by letting so many people free ride on your generosity, you end up hurting yourself and the possibility of cottage-industry of professional hosting providers?
Ok, so you are not taking anything out of pocket at all? That’s better than most, I suppose.
Still, during the interview you touch on the subject of how the donation model is not sustainable and it can only works at the scale that Fedi is right now. Wouldn’t you consider then switching to a different model?
@rglullis I think the donation model is working ok at this scale, but I don’t believe it will scale up to the hypothetical future we were discussing on the show where the fediverse became the social media platform for the masses. There are somewhere around 1 to 2 million active fediverse users, depending on how you count. If that were 100x or 1000x larger, we would simply crumble - I don’t think the general architecture scales that well (think of all the duplicate storage that we end up paying for across various instance) and generally, people who use social media are far less concerned with the core value propositions of the fediverse, like privacy and whatnot. I know that’s hard to accept, but we’re here because that’s how we think. So no, I don’t think we will have a future where a 500,000,000 active user fediverse can be operated off of donations from members. I also very much doubt that people would pay a fee to be here when corporate social media alternatives are “free” to them
I agree with pretty much everything you are saying, but I disagree on the solution. I think that us insisting on the donation model is putting an artificial limit on further growth. It “works” for this 1M-2M MAU, but these numbers are not enough to attract other players and who might be willing to try different approaches.
I think we need to change the general mindset that we “need” the donation model to keep the people around, and flip to a system where every user is expected to pay a little bit. And yeah, you might argue that not everyone is able to afford it, but it would easier to come with systems where not-paying is the exception instead of the rule. We can have a system where every N paying subscribers guarantee one free spot, with N=2, 3, 5, 10, up to the admin. We can have a system (like I have in Communick) where customers can buy “multiple seats” and invite whoever they want. Alternatively, we can set up a Caffe sospeso system where donations are still accepted, but accounted directly for someone who wants to claim it.
You are misunderstanding the main idea behind the whole system. It is fork-able. So people can always change things they personally find they don’t like about it. You can not have anything where everybody has to do. Because those who don’t agree have all the technological and legal right to ignore you and do what they want instead. And this is the point with libre platforms ( or libre software in general ).
Whatever solution we find needs to take this fundamental thing into consideration.
I don’t think the general architecture scales that well (think of all the duplicate storage …
That’s my hunch too, although haven’t studied in detail - so I wonder how we can fix it ?
Is there an forum that discusses this scaling issue (in general, across fediverse) ?
Storage Duplication is I think not necessarily an issue of ActivityPub, it’s an issue of implementation of it. Because all posts can technically live on their respective servers. And rendered directly or almost directly. Like it can be copied over for the time it is relevant, and then discarded to be available only from the original server.
That makes sense, to store only popular stuff, or temporarily - especially for ‘heavier’ images (although as we see with lemm.ee, that leads to issues when an instance dies). Yet I also wonder about the scalability of just the minimum meta-info, whose size does depend on the protocol design.
For example with Lemmy every upvote click propagates across the network (if i understand correctly, mastodon doesn’t propagate ‘likes’ so consistently, presumably for efficiency, but this can make it seem ‘empty’). Maybe such meta-info could be batched, or gathered by a smaller set of ‘node’ instances, from which others pick up periodically - some tree to disperse information rather than directly each instance to each other instance ?
As the fediverse grows, gathering past meta-info might also become a barrier to new entrant instances ?
I suppose this community is as good as any. But it’s difficult to talk in general about this as each fediverse app has different performance needs/characteristics, so I’m not sure if you can extrapolate anything in general. But perhaps?
I think “hundreds of thousands and even millions” is a bit of a stretch. Wikimedia’s annual report mentions donors at a level of “$50,000+”, and I’m guessing most of those are probably closer to 50,000 than to 100,000. Tbf I suppose that’s over just one year, so perhaps your statement isn’t entirely inaccurate.
Maybe the problem is that they are using ridiculously overpriced enterprise services like AWS or Azure, which provide their own solutions for a lot of common things like backups, replicas, logging, etc, but cost 100x more than what you can get with DIY on some cheap VPS if you’re fine with spending 1.25x more time.
Also, given that the instance is called “infosec.exchange”, you can be sure that he is not running this on some cheap VPS.
because cheap VPS will not give you enough bandwidth, or they oversubscribe their datacenters and their advertised speeds are far from real, or they have terrible support and if something goes down you are going to have a hard time to bring things up while having to explain to 10-15k people why things stopped working, or because the reason they manage to get such low prices is because they are selling user data on the side…
I’m not saying that the only correct alternative is to go to the big cloud providers, but there is a reason why “cheap” is not the sole criteria to choose a service provider.
Yes, the cheapest ones might have some risks, I mostly presented it as an example of what the opposite extremity looks like. There is a lot in-between, something a bit more expensive is even more guaranteed win. For example last time I used Hetzner, I had a server with 64gb RAM, 2TB SSD, and 16 cores Ryzen for something like €34/month. Hetzner support is very decent and they’re very well known, have decent reputation and been providing their services for a long time.
I’m currently watching the interview (quite interesting, to be honest, funny to watch an interview about Fediverse services). Could you please share the timestamp when Jerry talks about finances and costs?
@[email protected] , I’m sorry to bother but is it really true? Are you paying almost $5000/month out of your own pocket?
If true, why? This is not sustainable. Don’t you think that by letting so many people free ride on your generosity, you end up hurting yourself and the possibility of cottage-industry of professional hosting providers?
@rglullis @blenderdumbass I have donations from members that cover the costs.
Thank you for chiming in, Jerry!
Great interview, I only watched a part of it, but it was very interesting and refreshing to see your perspective on things. Thank you!
Ok, so you are not taking anything out of pocket at all? That’s better than most, I suppose.
Still, during the interview you touch on the subject of how the donation model is not sustainable and it can only works at the scale that Fedi is right now. Wouldn’t you consider then switching to a different model?
@rglullis I think the donation model is working ok at this scale, but I don’t believe it will scale up to the hypothetical future we were discussing on the show where the fediverse became the social media platform for the masses. There are somewhere around 1 to 2 million active fediverse users, depending on how you count. If that were 100x or 1000x larger, we would simply crumble - I don’t think the general architecture scales that well (think of all the duplicate storage that we end up paying for across various instance) and generally, people who use social media are far less concerned with the core value propositions of the fediverse, like privacy and whatnot. I know that’s hard to accept, but we’re here because that’s how we think. So no, I don’t think we will have a future where a 500,000,000 active user fediverse can be operated off of donations from members. I also very much doubt that people would pay a fee to be here when corporate social media alternatives are “free” to them
I agree with pretty much everything you are saying, but I disagree on the solution. I think that us insisting on the donation model is putting an artificial limit on further growth. It “works” for this 1M-2M MAU, but these numbers are not enough to attract other players and who might be willing to try different approaches.
I think we need to change the general mindset that we “need” the donation model to keep the people around, and flip to a system where every user is expected to pay a little bit. And yeah, you might argue that not everyone is able to afford it, but it would easier to come with systems where not-paying is the exception instead of the rule. We can have a system where every N paying subscribers guarantee one free spot, with N=2, 3, 5, 10, up to the admin. We can have a system (like I have in Communick) where customers can buy “multiple seats” and invite whoever they want. Alternatively, we can set up a Caffe sospeso system where donations are still accepted, but accounted directly for someone who wants to claim it.
You are misunderstanding the main idea behind the whole system. It is fork-able. So people can always change things they personally find they don’t like about it. You can not have anything where everybody has to do. Because those who don’t agree have all the technological and legal right to ignore you and do what they want instead. And this is the point with libre platforms ( or libre software in general ).
Whatever solution we find needs to take this fundamental thing into consideration.
That’s my hunch too, although haven’t studied in detail - so I wonder how we can fix it ?
Is there an forum that discusses this scaling issue (in general, across fediverse) ?
Storage Duplication is I think not necessarily an issue of ActivityPub, it’s an issue of implementation of it. Because all posts can technically live on their respective servers. And rendered directly or almost directly. Like it can be copied over for the time it is relevant, and then discarded to be available only from the original server.
That makes sense, to store only popular stuff, or temporarily - especially for ‘heavier’ images (although as we see with lemm.ee, that leads to issues when an instance dies). Yet I also wonder about the scalability of just the minimum meta-info, whose size does depend on the protocol design.
For example with Lemmy every upvote click propagates across the network (if i understand correctly, mastodon doesn’t propagate ‘likes’ so consistently, presumably for efficiency, but this can make it seem ‘empty’). Maybe such meta-info could be batched, or gathered by a smaller set of ‘node’ instances, from which others pick up periodically - some tree to disperse information rather than directly each instance to each other instance ?
As the fediverse grows, gathering past meta-info might also become a barrier to new entrant instances ?
I suppose this community is as good as any. But it’s difficult to talk in general about this as each fediverse app has different performance needs/characteristics, so I’m not sure if you can extrapolate anything in general. But perhaps?
Thank you!
Why shouldn’t the donation model keep working? Wikipedia works on donations, why can’t the fediverse?
Wikipedia had big donors who can donate hundred thousands of dollars and even millions
I think “hundreds of thousands and even millions” is a bit of a stretch. Wikimedia’s annual report mentions donors at a level of “$50,000+”, and I’m guessing most of those are probably closer to 50,000 than to 100,000. Tbf I suppose that’s over just one year, so perhaps your statement isn’t entirely inaccurate.
I wonder why it needs so much money for infra? Last time I rented a VPS it was €7/month for 8 Core Xeon E5 V4, 12 GB DDR4 RAM, 150 GB SSD/NVME, Unlimited Traffic, 1 Gbps Port.
Storage. In the video he says that backups alone costs $500/month.
Also, given that the instance is called “infosec.exchange”, you can be sure that he is not running this on some cheap VPS.
Maybe the problem is that they are using ridiculously overpriced enterprise services like AWS or Azure, which provide their own solutions for a lot of common things like backups, replicas, logging, etc, but cost 100x more than what you can get with DIY on some cheap VPS if you’re fine with spending 1.25x more time.
Why not, though.
because cheap VPS will not give you enough bandwidth, or they oversubscribe their datacenters and their advertised speeds are far from real, or they have terrible support and if something goes down you are going to have a hard time to bring things up while having to explain to 10-15k people why things stopped working, or because the reason they manage to get such low prices is because they are selling user data on the side…
I’m not saying that the only correct alternative is to go to the big cloud providers, but there is a reason why “cheap” is not the sole criteria to choose a service provider.
Yes, the cheapest ones might have some risks, I mostly presented it as an example of what the opposite extremity looks like. There is a lot in-between, something a bit more expensive is even more guaranteed win. For example last time I used Hetzner, I had a server with 64gb RAM, 2TB SSD, and 16 cores Ryzen for something like €34/month. Hetzner support is very decent and they’re very well known, have decent reputation and been providing their services for a long time.
I’m currently watching the interview (quite interesting, to be honest, funny to watch an interview about Fediverse services). Could you please share the timestamp when Jerry talks about finances and costs?
34 minute mark. It says in the blog post from OP.
Thanks