@Hamiller Friendica

Nach diesem Muster ist er auch bei Friendica und Hubzilla vorgegangen.

Na ja, es war ähnlich.

2012 war Friendica längst stabil und im Grunde fertig. Er hat es an die Community abgegeben, Red abgeforkt und mit Zot experimentiert.

2018 war Hubzilla stabil und im Grunde fertig. Er hat es an die Community abgegeben, Osada und Zap abgeforkt und mit Zot6 experimentiert.

2020 war Zap stabil und im Grunde fertig. Er hat es an die Community abgegeben und das zweite Osada gleich mit. Nachdem die Community umgehend Osada eingestellt hat, weil es eh mit Zap beinahe identisch war, hat Mike ein drittes Osada, ein neues Mistpark und eine neue Redmatrix abgeforkt, um mit Zot8 zu experimentieren.

Aus den Experimenten ging nie etwas Stabiles hervor. Statt dessen hat er von einem von den dreien 2021 Roadhouse geforkt, um mit der nächsten Zot-Evolutionsstufe zu experimentieren, die dann in Nomad umbenannt wurde.

(streams) aus demselben Jahr sollte dann Roadhouse in stabil werden. Und Mike wollte (streams) nicht wieder forken. Dann kam Mike aber an einen Punkt, wo er sagte: Nomadische Identität geht auch mit ActivityPub. Ich brauche kein eigenes Protokoll mehr, ich muß nur dabei mithelfen, ActivityPub dahin zu bringen, daß es Nomad ersetzen kann.

Weil er aber (streams) nicht forken wollte, hat er das Ganze auf (streams) selbst versucht umzusetzen. Blöderweise läuft das in der Praxis nicht so geschmeidig, wie es in der Theorie angedacht war.

Statt jetzt aber seinen einzigen stabilen Release endgültig in eine Bastelbude zu verwandeln, hat er jetzt Forte abgeforkt und nimmt das zum Basteln, während (streams) wieder auf stabile Beine kommen soll. Auch das macht er selber, weil das keiner für ihn übernimmt. Und die (streams)-Community ist keine drei Jahre nach der Entstehung von (streams) noch zu klein, um so bald die Entwicklung von (streams) zu übernehmen. Kaum einer zieht von Hubzilla um, ganz neu nach (streams) kommt eh keiner, auf Mastodon weiß kaum einer, daß es (streams) gibt, und die, die davon wissen, trauen sich nicht hin.

Und so wird Mike beides weiterentwickeln. Forte wird wahrscheinlich zunächst ein Soft Fork bleiben, damit Mike sich nicht dieselbe Arbeit zweimal machen muß.

So gesehen ist das eher vergleichbar mit Zap und den ersten zwei Osadas, wo Mike schon mal zwei Projekte mit in Teilen unterschiedlicher Codebase am Laufen hatte.

CC: @Raphael

#Long #LongPost #CWLong #CWLongPost #LangerPost #CWLangerPost #FediMeta #FediverseMeta #CWFediMeta #CWFediverseMeta #Friendica #Red #RedMatrix #Hubzilla #Osada #Zap #Mistpark2020 #Misty #Redmatrix2020 #Roadhouse #Streams #(streams) #Forte

@Jens Ljungkvist :mastodon: @Jeff Sikes @Kainoa @Chris Trottier Something similar to "one account on all projects" is already in the works.

By and by, #Fediverse projects may adopt #OpenWebAuth, a #SingleSignOn implementation developed by @mike for #Hubzilla and currently implemented on Hubzilla, its direct predecessor #Friendica and its latest not-quite direct descendant, #Streams. An implementation is also in development on #Mastodon. It should not be confused with #OAuth and #OAuth2, these are something entirely different.

What OpenWebAuth is that it recognises logins elsewhere. When I'm logged into this Hubzilla account, and I visit another Hubzilla hub or maybe a Friendica node or a (streams) instance, it will automatically recognise me. And it will grant me some extra "guest permissions" like being able to post directly on the wall of another Hubzilla or (streams) channel.

What it does not do, however, is give me all the power on any Friendica node, Hubzilla hub or (streams) instance that a logged-in user with a user account has.

I can't go to another Hubzilla hub and create a clone of my channel or create a brand-new channel or post an article or start a wiki or upload files just with my OpenWebAuth login credentials. And when Mastodon introduces OpenWebAuth, I still won't be able to go to any one random Mastodon instance and start tooting. All this would still require a local user account on that one specific instance.

One account for the whole Fediverse is utopic. It's technologically impossible or just very very very unfeasible.

The Fediverse has 24,000+ instances of dozens of projects. If you want full local user power everywhere in the Fediverse, you'll need one registered account on each one of these 24,000+ instances.

Whenever someone joins mastodon.social, then RATATATATATATATATATATA, 24,000+ more accounts with the same login credentials will have to be created automatically.

Also, the Fediverse has 12,000,000+ users. If you want full local user power everywhere in the Fediverse, then everyone else must have it, too. So every single instance of each Fediverse project will have to have one account per Fediverse user. The only exceptions would be those very few projects which are designed for only one user account.

However, personal instances of projects that are designed for multiple user accounts will all be affected. The hapless Mastodon user who comes over to your personal Hubzilla hub to act like a registered user will neither know nor care if that hub is running on a root server in a data centre with two 36-core Xeon CPUs and enough RAM to make a 3-D CAD workstation cry or on a Raspberry Pi at your home.

Now, let's assume someone has set up a new Web server with some Fediverse project installed on it. It doesn't matter if that's Mastodon or #CalcKey or #Lemmy or #Mitra or (streams) or whatever as long as it has #ActivityPub. They start that thing up for the first time: sudo systemctl start nginx or so.

And RATATATATATATATATATATA TATATATATATATATATATATA TATATATATATATATATATATA TATATATATATATATATATATA TATATATATATATATATATATA TATATATATATATATATATATA, that poor thing will sit for WEEKS registering over twelve million user accounts.

Why? Because anyone in the Fediverse might come over anytime soon and want to use just this one specific instance as if they had registered their personal user account there. In order to be able to do that, they need a user account.

By the way, not even the notorious featherweight #Pleroma could handle 12,000,000+ user accounts on one instance. Mastodon can do that even less, not to mention the heavyweight Friendica or the super-heavyweight Hubzilla.

Speaking of Hubzilla, maybe a new Hubzilla hub might get away more easily when starting up for the first time. On Hubzilla, ActivityPub is optional per hub and then per channel. The hub admin can switch it on and off, and if it's on, the users can switch it on and off again for each one of their channels.

So if ActivityPub is off on the admin side by default, new Hubzilla hubs will only register one user account for each Hubzilla and (streams) user out there, maybe also for the users on the few remaining instances of the #Zotlabs projects that went EOL on New Year's Eve 2022, #Redmatrix, #Osada, #Zap, #Misty a.k.a. #Mistpark2020 and #Roadhouse. They all speak one native language, #Zot.

But once the admin activates the Pubcrawl app for their hub, that hub will immediately start registering user accounts for every user on every instance of every project that connects to Hubzilla via ActivityPub, each account with one channel with Pubcrawl on. And it will spend weeks or months doing so and not have any server resources left to do anything else in the meantime.

Speaking of Hubzilla, there's also #NomadicIdentity, the killer feature of the Zot protocol. Hubzilla has it, (streams) has it, and the (un)dead Zotlabs projects have it.

Ideally, each Fediverse user would not get one account on each Hubzilla hub and each (streams) instance with one separate, unique channel on it. They would first get the accounts. On one account on one Hubzilla hub, one channel would be created. This channel would then be cloned across all Hubzilla hubs and to (streams).

Advantage: Each Fediverse user would only have one channel for Hubzilla and (streams) together. They would have the exact same content on all Hubzilla hubs and, minus what Hubzilla can do that (streams) can't, all (streams) instances.

Obvious disadvantage: Whenever someone decides to do something on that channel, it would have to be synced to all its clones in near-real-time, causing a lot of network traffic.

And if you set up a new Hubzilla hub or (streams) instance, the creation of 12,000,000+ accounts would actually become a lesser problem. The bigger problem would be the 12,000,000+ channels that will be cloned onto your machine with everything on them. You'd better attach a few petabytes worth of HDD capacities to your personal little Raspberry Pi.

By the way, if everyone had full local user rights on each Fediverse instance, the Fediverse would have over 300 billion local accounts.

@Kristian @Chris Trottier Free, non-corporate, decentralised projects have different intents and purposes than non-free, commercial, corporate, centralised silos. They're created by different people for different people, for different target audiences. And even the huge corporate silos don't start with a shiny iPhone app and then develop the server backend around it.

If it's free (as in, for example, Affero GPL), decentralised and distributed, it's made by geeks for geeks first and foremost. #Friendica first became available in 2010, and unlike Facebook, it never had the intention of becoming the next Internet for everyone in the world. Also, behind Facebook stood a huge megacorporation. Behind Friendica stood only one man, @mike, all alone, with zero budget. And yet, he managed to release something that was more powerful than Diaspora*, where at the same time only the crowdfunding campaign was running, would ever become.

Friendica's target audience were geeks. The same people that also used Linux as their main OS. Friendica wasn't made for the same people as Facebook or the iPhone. In fact, your typical Friendica user wouldn't touch an iPhone or any other Apple product with a 10-foot barge pole. They'd rather have a Nokia N900, and that was a clunky QWERTY slider that ran a modified Debian GNU/Linux.

#Redmatrix, the direct successor of Friendica, was experimental. Its sole purpose was to work on the brand-new #Zot protocol and the concept of #NomadicIdentity. It still had a small number of users and an even smaller number of instances, but they were generally voluntary guinea pigs. At this time, Friendica was already maintained by its own community which is about as far away from a Silicon Valley gigacorp as you could possibly get.

Redmatrix wasn't declared ready for prime time until late 2015 when it was renamed #Hubzilla. And even then, it didn't come with the "vision" of rolling over the mass market and replacing Facebook, WordPress, MediaWiki and the various GAFAM cloud services in one fell swoop. Again, Hubzilla was developed pretty much only by Mike Macgirvin.

#Osada and #Zap were both largely experimental again. Mike had forked them off Hubzilla because he still wasn't satisfied with what Zot could do at the time. However, the development of the new version #Zot6 couldn't happen on that monster named Hubzilla that was in everyday use now. That's why these two new projects were launched.

There's a good reason why they were two projects. Zap was there first. Zap was the actual Zot6 testbed, and thus, Zap was Zot6-only. Osada retained compatibility with Friendica and Hubzilla to test how well Zot6 would interact with ActivityPub with had meanwhile appeared as a draft and, IIRC, adopted by both Friendica and Hubzilla. Eventually, Osada and Zap ended up having the exact same codebase, and the difference between the two was an admin switch: ActivityPub on made it Osada, ActivityPub off made it Zap. As this was non-sense, Osada was axed, and Zap got ActivityPub and was declared the next stable one.

First, Zap's main killer feature over Hubzilla was Zot6 which had introduced #OpenWebAuth. When Zot6 was finally backported to Hubzilla, the remaining advantage was that Zap wasn't nearly as bloated with a somewhat less overwhelming UI. By the way, Redmatrix continued to exist with one user until Mike Macgirvin upgraded his own instance to Zap.

Now, again, you can't tinker with something that's stable. And tinkering continued. #Mistpark, Friendica's early name, returned in 2020, as did Redmatrix and Osada, all as Zap forks at various stages of instability and being experimental, none intended for a wider audience. And all created by Mike Macgirvin again. You could happily switch back and forth between Redmatrix, Osada, Zap and #Misty by simply rebasing your server code. (Installing either usually involved "git clone".)

He actually had a very good reason for this maze of names: He is opposed to big mass products with big brand identities. He wants to offer people technical solutions, not cool stuff with a sleek brand on it.

Anyways, on top of all this came #Roadhouse, another fork from somewhere in this conglomerate which was created in 2021 and solely intended for the development of the next Zot version, originally named Zot8, now known as #Nomad. Roadhouse was so experimental that there has never even been an official text saying what it actually was.

Also in 2021 came #Streams, a Roadhouse derivative that started out just as mysterious but was eventually intended for the public. It's often also referred to as (streams) because it's different from its predecessors in one point: It's even less of a brand. It isn't a product to be used as-is. (streams) is not a "Fediverse platform" that's waiting for its own iPhone app. (streams) is a code repository on Codeberg. And its purpose is for others to take the code and make something out of it. It isn't meant to be run as-is, although you can do that, and some people do. And even then, it comes without a fixed brand and kind of asks you to "rebrand" it, even on a per-instance basis. Most (streams)-based instances don't identify as (streams). Mike who is still involved in the project has his own instance based on (streams) but, probably deliberately and intentionally, still has it identify as Zap.

Another interesting fact: (streams) uses a wild hodge-podge of free licenses. Most of it is in the public domain, but parts of it are under various free licenses which aren't compatible with each other. This is fully intentional, too. It makes using (streams) for commercial products pretty much impossible because no corporate legal department will be able to figure out how to legally comply with all these licenses at the same time. Free use stays basically unlimited, though.

By the way: As of January 1st, 2023, Redmatrix, Osada, Zap, Misty and Roadhouse are EOL and discontinued, and their code repositories were closed. Instances running them can and shall be upgraded to (streams). All that's left is Friendica (the old faithful one), Hubzilla (the nomadic monster) and (streams) (the one for the tinkerers).

Now there's still the question: Why do all these projects, in fact including #Mastodon, use this approach? Why do they start with a server platform plus Web frontend instead of doing as big corporations do and start with an iPhone app and develop a server backend around it? Why appeal to a small bunch of Linux nerds rather than to a mass-market of billions?

Because if you want to go free and decentralised and distributed and federated, you'll need those Linux nerds before everyone else.

First of all, you'll need someone to run instances. Thus, you'll need people who are willing and able to do that. This requires Linux knowledge. The ability to use the command line. The ability to set up and configure a Web server. Network knowledge to connect it to the internet. You can't set up a Web server from zero with three taps on a mobile app.

In fact, when Diaspora* was young, it only ran on Mac servers. All four creators were Apple fanbois who didn't care for anything without the Apple brand on it. The Diaspora* server application was built against macOS. The result was a dire lack of public pods (instances) and everyone piling on the official pod. Mac users don't run Web servers at home, and I guess there were no hosting companies that offered Web hosting on Macs. The devs eventually had to make the server app at least halfway Linux-compatible to get more people to run pods, and you still had to compile Ruby on Rails from sources on Debian stable because Diaspora* depended on a newer version.

Also, you'll need these tech geeks to spot and report bugs. Your typical Windows or Apple user doesn't report bugs; they only complain about them or switch to a competing product. In stark contrast, many Linux users even know how to file a good and informative bug report. Some are even capable of submitting pull requests with bug-fixing patches through git.

And at least in the case of Mike's projects, you'll need a community that's capable of taking over the project itself and continuing its development. You'll need people who know how to code. You'll need people who know how to use git. And so forth. You'll hardly find such people amongst the masses who have spent all their digital lives in the cosy world of corporate-designed GUIs.

If, for example, Mastodon had started out with iPhone and Android apps and gone from there, appealing to a rather tech-illiterate mass audience, it would probably never have become decentralised. At least not beyond the federation between mastodon.social, mastodon.online and whatever more instances Eugen Rochko would have had to launch because these two were full.

And why not? Because Mastodon wouldn't have appealed to people who know how to install and run Mastodon instances. Mastodon would have only had the Windows/Mac/iPhone/Android crowd as users. All the geeks who would have known how to set up and run a Web server would have stayed on Friendica and Hubzilla. Some may have used ActivityPub to connect to Mastodon, but hardly anyone would have switched to that actually inferior platform with a wholly different crowd on it.