I really don't get this concern. Web hosting is cheap. Like, astonishingly cheap. Mastodon instances take up a couple GB of RAM, some storage, and relatively light bandwidth. I would be shocked if an instance cost more than $40/month to host, and I'd reckon the typical instance is less than $20/month.
That's the sort of money people happily pour into hobbies without a second thought. Collectively my Discord server of like 30 people pay more than that on their premium offering. All it takes is one person in your community to foot a relatively trivial bill. Or, orchestrate a Patreon or similar.
I would be shocked if an instance cost more than $40/month to host, and I'd reckon the typical instance is less than $20/month.
That information is public for at least one moderately popular instance, fosstodon.org. According to https://hub.fosstodon.org/about/#current-funding the current cost is $1895.50 per month, plus $200.00 for the CDN. (And yes, last time I looked, their Patreon is going to be more than enough to cover that.)
This is something I’ve been feeling “all along”, but clearly the mental image of a typical Twitter use cases varies too much between stakeholders e.g. Mastodon backend devs, frontend Twitter app devs, casual users and heavy users. And Twitter employees, too.
To an addict like I am, Twitter timeline is something measured in tweets per second. It is a constant flow, a time synchronized operation, not something that I expect to be event based.
Mastodon devs and ActivityPub spec devs must have a complete opposite of that, that users only post content when there is something to say, and the post will be pushed when a post was made.
The latter model is certainly doable with the ActivityPub protocol, $20/month budget, and by stretch can be handled by Mastodon as it is built. But I think there is this misunderstanding, and it just seems that none of it is built to cater for those with the former model.
Comments
I really don't get this concern. Web hosting is cheap. Like, astonishingly cheap. Mastodon instances take up a couple GB of RAM, some storage, and relatively light bandwidth. I would be shocked if an instance cost more than $40/month to host, and I'd reckon the typical instance is less than $20/month.
That's the sort of money people happily pour into hobbies without a second thought. Collectively my Discord server of like 30 people pay more than that on their premium offering. All it takes is one person in your community to foot a relatively trivial bill. Or, orchestrate a Patreon or similar.
That information is public for at least one moderately popular instance, fosstodon.org. According to https://hub.fosstodon.org/about/#current-funding the current cost is $1895.50 per month, plus $200.00 for the CDN. (And yes, last time I looked, their Patreon is going to be more than enough to cover that.)
They've got 50,000 users. That can't be typical, right?
This is something I’ve been feeling “all along”, but clearly the mental image of a typical Twitter use cases varies too much between stakeholders e.g. Mastodon backend devs, frontend Twitter app devs, casual users and heavy users. And Twitter employees, too.
To an addict like I am, Twitter timeline is something measured in tweets per second. It is a constant flow, a time synchronized operation, not something that I expect to be event based.
Mastodon devs and ActivityPub spec devs must have a complete opposite of that, that users only post content when there is something to say, and the post will be pushed when a post was made.
The latter model is certainly doable with the ActivityPub protocol, $20/month budget, and by stretch can be handled by Mastodon as it is built. But I think there is this misunderstanding, and it just seems that none of it is built to cater for those with the former model.