Adblockers and alternative frontend will only get you so far in this. It's literally just a matter of time until step it to yet another level. Want some real change? Stop giving them power. #Boycott these fuckers.
"Youtube has started to artificially slow down video load times if you use Firefox" -> https://archive.is/yAdak (link to avoid direct #reddit)
Bring down closed data silos like #YouTube. Empower #foss federated platforms like #peertube.
@kzimmermann Are you willing to financially support big #peertube instances?
They will need big data storage and a moderation team.
@Revertron @kzimmermann IMHO This is why large instances shouldn't be a thing anywhere on the #fediverse .
The sustainable and maintainable approach is smaller instances, and more of them. Individual creators should try to either join small instances or host their own like a website.
@lps the problem is that smaller instances and more of them consume exponentially more resources, making the whole system exponentially more expensive to operate.
That’s just how the protocol was designed, unfortunately.
People complain about Bitcoin being resource-intensive, but Fediverse faces a similar problem.
@volkris @lps @Revertron @kzimmermann
yeah, perhaps if you completely ignore the p2p and p2n baked into PeerTube. anyone who has used decentralized media distribution IRL can see through this obvious FUD.
Centralization is only better at gatekeeping, user tracking, and ddos vulnerability.
@errhead I’m referring to the design of ActivityPub at the core of Fediverse.
If you’re talking about something else that’s fine, but it’s not what I’m referring to.
@silverpill if you read through the ActivityPub standard, the protocol is designed to scale with number of instances, and it is a scaling factor larger than simply linear.
You say AP is very easy on resources, but that flies in the face of so many instance operators experiencing meltdowns and surprise hosting costs when the resources exploded far higher than they expected.
Which makes sense because, again, the protocol is designed in a way that involves exponential increases.
@errhead I’m referring specifically to the ActivityPub standard, not Mastodon.
AP lays out the requirements that instances communicate with each other, manage inboxes and out boxes, and all of this other stuff.
It is not a lightweight protocol.
@volkris @lps @errhead @Revertron @kzimmermann
Yes, AP is very easy on resources. I've built a full-featured AP server from scratch so I know how the protocol works and how it behaves under different conditions.
In AP world you only receive stuff you care about, so most of the time server is just waiting for something to land into inbox, or for outgoing request to complete. There's nothing else to do
>so many instance operators experiencing meltdowns and surprise hosting costs when the resources exploded far higher than they expected
Tell them to use better software.
so many instance operators experiencing meltdowns and surprise hosting costs when the resources exploded far higher than they expected Tell them to use better software.
Commending and adding to this: I’ve also written an ActivityPub server project that’s running in production now, which was originally intended for a niche purpose (but may likely be applicable in much broader scope) where: majority of server activity is essentially just static file hosting. It’s just legitimately Apache/nginx serving up flat HTML and JSON files, as requested, while the fedi software generates those files as new posts are made.
Outward delivery could be moved to a different server entirely that doesn’t even have to be always-online, inward delivery can also be on a separate server and subdomain entirely (and the inbox is the only public/federation-facing that even needs to run any dynamic code). ActivityPub was much intentionally meant to allow simple minimalistic implementations, per IndieWeb influence/inspirations, it wasn’t required that you’d have to install some +1GB filesize suite of software or anything.
The reason there’s extraordinary hosting costs is because of people using grossly-overarchitected or grossly bloated software projects, and are greatly overcomplicating the problem. In my dev lab, apparently a production Mastodon installation and it’s dependencies takes 2.5GB for Ruby, it’s libraries/dependencies, and Mastodon and it’s dependencies. That’s far worse than even majority of Electron-packaged software. Meanwhile I have something in just ~3.5k lines of PHP, that also just lets a webserver be a webserver, and it’s federating just fine and handling massive storms of queries just fine on a $10/month VPS. It’s also of course worth mentioning: all the other options out there that aren’t Mastodon, that also scale far better, but everyone acts so anemic to them, as if Mastodon “is the only way” as some weird ego thing.
@volkris @silverpill @lps @Revertron @kzimmermann
You're conflating the horrible federation model of Mastodon with all of ActivityPub