Actively pushing against others to adopt your protocol is how you ensure that your protocol won't "win." I'm a firm believer that a decentralized protocol will be the future of social networks/feeds as we know them. I don't know if it is going to be #ActivityPub or #ATProto or something else. But I believe a decentralized protocol will be the future. But the winner will be the one that centralized-acting services adopt. That's a good thing for everyone. It means users have data portability. 🧵
But I'll tell you right now who will not win: the protocol and community that wants to make it as difficult as possible for people to adopt/build-off of and interoperate with. Ogg Vorbis, Opus, and Theora tried to do the whole "open" standard audio and video codec thing. It was bad quality and hard to implement, but “open.” Everyone smart used MP3 anyway. And eventually, MP3 became patent free. Even smarter people started working on things like VP8 and VP9 and HEVC and eventually we have AV1
And AV1 is still in-progress, but it has major adoption from almost everyone and is royalty free. It has succeed in all the areas Theora has failed. It's really easy to be so insular and dogmatic that you become Theora. Meanwhile, actual good innovation that can benefit everyone (and yes, that does mean for-profit companies) is what changes the world
@film_girl Truly open web standards were what allowed us to loosen Microsoft’s grip on web browsers. It’s remarkable how IE adopted standards, other browsers were able to break in, and now IE is a distant memory. If folks had tried to reject MS’s adoption of web standards, this would look very different.
@film_girl Still very new to the fediverse, and I always avoided any details of technical details of protocols at work but I think you are exactly on point.
Protocols flow to the path of least resistance. I wish I could explain the benefits of open protocols to my normal friends. Portability is a good example here.
@film_girl Great thread, just based on what you mention: it will be decisive for the overlapping of any protocol, vital for it to dominate, otherwise it will be reduced to just a handful of geeks, just like an elite.
@film_girl yep, anything successful means nearly everyone (the exceptions being the ones that don't follow/moderate to established community standards). That has to include corps and governments (like the EU is moving towards) and even reporters. But it doesn't mean including, as Gab found out, Gab.
Tbh I think so far as I understand AP the biggest challenge is going to be server resources on existing instances, which will mean donations from those that are able.
@film_girl Exactly. I ran FreeBSD for years. It’s very macOS like (for obvious reasons to long timers). But Linux took the jobs it was well suited to do because it was unencumbered by legal questions. Arguably it then went on to never gain a huge footprint on the desktop because of intramural open source petty conflict and a cultural disdain for the casual user. If the point of the protocol is to be open and distributed then support that. Defederate later if needed.
@Gte DING DING DING again! I also preferred FreeBSD for the same reasons (FreeNAS was the last thing I had with it and even that, which I think is TrueNAS now, has gone Linux, butI went Synology, which is the macOS of Linux NAS boxes). Defederate later, indeed! But force people to use/depend on you. Samba is included in WSL2 FFS! That's proof that it won over SMB! Ubuntu came closest to winning on the desktop and the toejam eaters even managed to scare them back into the enterprise data center
@film_girl@Gte I was so surprised after not paying attention to Linux closely for a few years to see all of the Ubuntu hate a while back. They had the audacity to do something pretty minor from what I gathered, how dare they.
@moirearty@Gte they had the audacity to invest tens of millions of dollars into something and try to make product-led decisions instead of arguing by committee. How dare they!
@film_girl@Gte TrueNAS still based on FreeBSD; but they started TrueNAS Scale, which is Linux based (still alpha-like software, which is buggy as hell).
@lapcatsoftware@film_girl Haha! Sold! And, yeah, I do feel good about a sort of more sane take on Unix getting a broad reach. Arguably macOS has changed so much it has moved away again but they’re siblings at least.
@Gte@lapcatsoftware yes absolutely! for years I used to always joke that the year of the Linux desktop was 2002 when Mac OS X finally got good/reliable. And I still sometimes quip to the "btw I use Arch" crowd that "btw, I use actual Unix” - but I agree with Guy that macOS has changed so much it isn't the same anymore, but is at least still siblings!
@film_girl@Gte@lapcatsoftware The real ironic part is that the latest and greatest macOS Ventura (Intel + Apple Silicon) is still officially certified as all-caps UNIX, and Oracle doesn’t even maintain it for Solaris anymore!
@Gte@film_girl I also ran FreeBSD on my hand-built home computer for years. I loved it so much. It saddens me to remember that this was 20-25 years ago though. It feels like last week!
@film_girl@Gte If I could frame this, I would. EXACTLY THIS. Can a developer come along and build something interesting inside a day without asking for permission or getting pushback for existing? Great. That's what we need.
@film_girl@Gte This phrase keeps coming back to me. Especially considering the ethically dubious origins of IBM compatible PCs and the queer hippies who took DARPA money to build the internet.
@ethanschoonover@film_girl I don’t think it’s dogma per se in this particular scenario. i think it’s an unqittingly narrow view of both “open-ness” and “easiness”
@film_girl Opus & Vorbis "were" "bad quality?" Opus has completely taken over for anything real time - Discord & many communication apps have adopted Opus. Vorbis is used by Spotify. I don't know who "everyone smart" is, maybe I'm just an idiot
AV1 is winning largely because it is Google's baby. Development of the reference encoder isn't "open" at all, and their monopolistic influence has caused problems for competing codecs (like JXL for images, which is superior to AVIF) because of Chromium
@film_girl AV1 is technically VP10 (even on Windows SDK there's some mention of it). I wouldn't say AV1 has won though. Apple still doesn't provide hardware decoding and without hardware encoding it's not practical for most people either.
@robUx4 no, you’re correct. Obviously AVC and HEVC are the clear winners, even beyond VP9. But as a royalty free solution, it’s going to be AV1, not Daala or Theora or anything else that doesn’t have the backing of the people that actually do the work and have the users.
@film_girl@robUx4 How are you defining “winner?” HEVC has had zero penetration outside the Apple ecosystem, & essentially doesn’t exist on the Web. And AV1 is very much not VP10 - it is built off Daala as well as technologies proposed for VP10. That’s like saying “Opus is essentially CELT, SILK sucks.” I get the point you’re trying to make, but these codec comparisons are wholly uninformed & off base.
@film_girl Did not see that comparison coming! I don’t think the Theora people were nearly as dogmatic and stupid as the people pushing against Big Tech on the Fediverse.
@nmn You’re probably right and I regret the analogy now b/c my mentions are full of very angry Ogg fans (who knew!) but I still stand by the overall point!
@film_girl@nmn Yeah, it was unfortunately a very weak analogy because it was almost entirely incorrect. I don’t even hate your initial point, but as someone who actually understands multimedia codec technology, you ruined your message.
@film_girl All three protocols you mentioned were developed after mp3 existed as alternatives that people could use without the licensing cartel - and I don’t think implementation difficulty was the problem at all (they are OSS), or even quality.
Mp3 remained used by pure market inertia dynamics.
@chucker@Migueldeicaza@film_girl This. Given the patent history related to mp3, the chances of an independent audio code not infringing on any patents is slim. And which big company is going to take the risk of losing a multimillion dollar patent lawsuit.
It's safer to stick to formats covered by the mp3 patent pool
@chucker@Pwnallthethings@Migueldeicaza@larryosterman@film_girl Not just for the Mac. Project K (which eventually became .NET Core) started with bits from Silverlight on Windows. In fact the netcore TFM meant "Silverlight" (specifically for Win8/WP8), which is why .NET Core ended up using netcoreapp.
@Migueldeicaza@larryosterman@chucker oh, Moonlight! I loved that. Silverlight/WPF had some interesting things from what I remember (which isn’t much). I was invited to a two day Microsoft training thing on Silverlight/WPF in 2009 and I showed up with a MacBook and was using Parallels to use it and the Microsoft trainer like made fun of me for being a Mac user. Meanwhile, I got Visual Studio to work and half the class didn’t so, apple hardware: 1, Dell: 0
@Migueldeicaza@larryosterman@chucker in a pre-HTML5 world, codec bundles stuff was so much more complex that Hulu and Netflix had to use Silverlight for DRM
@film_girl@Migueldeicaza@larryosterman “yeah we wanted to play streaming video in a browser”
— “sure”
— “…so we used a portable .NET runtime packaged as a Netscape plug-in and used that to play Windows Media Vi—“
— “what”
A colleague also bought a Dell and then needed a Mac to publish Flutter apps, and ended up preferring the Mac so much that he now uses that for his .NET stuff
@film_girl I worked in an all-Windows coding shop. I brought my personal MacBook Pro to a meeting once, so I could refer to some PDFs, and my anti-Apple manager really got onto me. "You've got a lot of guts bringing that in here!"
fun fact 1: those of us who traded in lossless (deadhead tapers) refused to use mp3 and any proprietary format in favor of old fashioned wav or flac. There's an entire sleeve of disc's somewhere in my garage full of them.
Also fun fact 2 from my time at rollingstone dot com in 96-98: all their content was ripped onto .wma files coz it was faster than RealPlayer, higher bit rate so better sound, smaller files than mp3, and universally available on every windows machine running Netscape at the time... Also was how we traced what files were being shared on Napster, Coz wma files had a unique signature tied to it's source.
@eljefedsecurit@larryosterman@Migueldeicaza@film_girl that reminds me of QuickTime 4 with its bizarre UI that included a volume dial nobody understood (it looks like you’d operate it in a circular way, but you were supposed to drag vertically) and a channels list which made teenage me wonder why The Rolling Stones would produce content for Apple in the 1990s
@film_girl That isn't quite right. Vorbis and Opus weren't bad quality (though the packet interlace stuff in Ogg was mistaken), but what scuppered them was bigco's with submarine patents. I tried to get QuickTime to ship Vorbis support and was told by our VP that Dolby and Fraunhofer would start patent litigation if we did.
Similarly Apple fucked up by patenting part of the QuickTime Movie file format and threatened lawsuits, so other players dropped support.
@sandofsky oof. They were so busy being mad about TiVo they completely missed the cloud revolution coming until they could try the even more unpopular AGPL.
@sandofsky@film_girl I'm not sure open source that’s a one way street is better. We end up with stuff like heartbleed. We simply expects a tiny team or one person to basically keep all our infra going. It's not the worst idea in the world to expect something in return. I personally think dual licensing & those who can pay should pay is better than the GPL strategy. However, the idea that the world's largest software company should get something fundamental to their OS for free feels wrong.
@fds@sandofsky to be clear, most large software companies don’t get fundamental parts of their OS for free. They contribute to the core of many projects, they hire engineers and designers and security people to work on stuff. Large companies can do better, always, but Heartbleed and Log4j didn’t happen because of what license they used. It happened for a host of a reasons but a license change to GPLv3 wouldn’t have prevented it.
@film_girl@sandofsky A change of license would mean they wouldn't have used it, may do it themselves & take ownership of it.
I've never worked for Apple but all companies I've worked for that claim to contribute to open source do the bare minimum & there's usually a fight involved to spend any time on that.
I think it would be healthier overall if devs were better at asking for compensation for their work. Then hire people to help maintain the code vs hoping companies are charitable.
@zippy1981@sandofsky and Apple probably could have found a way to bundle bash or make it a shim that would then download at a certain prompt or something, but their lawyers clearly advised them it wasn’t worth the effort. And most users who needed a newer shell were already using homebrew or macports but yeah, it’s what finally made them go to zsh.
@zippy1981@sandofsky I will say that for a lot of people that choose GPL v3, the friction it creates for other entities is a feature, not a bug. But it also leads to lower adoption of things with that license by people who aren’t ideologically tied to the FSF’s philosophy.
@film_girl there is nothing ensuring data portability in either of these protocols. Even if they were welcome to the network and not yeeted to the nethersphere like Gab and their associates, when (WHEN, not IF) the large centralized instances decide to defederate, the users on them will lose their outbound social graph and won't be able to do anything about it, remaining locked into those services just as if the Fediverse has never been.
@film_girl that was the delusion with which people enthusiastically welcomed Facebook and Google adopting XMPP for their instant messaging system. We have seen how that went. What makes you think it will be different this time?
@oblomov I mean, a) I think that’s a strawman and a misunderstanding of history. It was smartphones that killed XMPP, full stop. If the protocol had been updated and iterated on to adapt to the changing nature of mobile devices more quickly, services like WhatsApp wouldn’t have had to create their own protocols (originally based on XMPP) and FB and Google might have used them longer. b) letting a service interoperate doesn’t mean you let them take over.
@oblomov like, there are LOTS of reasons XMPP didn’t survive/thrive, and I’m angry about that too. But let’s not rewrite history and pretend as if Facebook and Google had continued to support it that the world would have been any different. Modern smartphones and the challenges between switching across data connections and sync and push and latency around XMPP on mobile connections were not good enough until it was too late. Short of a Time Machine, I think we always end up where we are.
@film_girl I'm sorry but that's hogwash. None of those were issues that couldn't be solved within XMPP, had there been any actual intent to be interoperable. But interoperability was never the point: it was just a capture + rug pull to vendor-lock people into their respective platforms. Heck, Google kept XMPP access to their chat until just a few months ago, years after it had defederated. And Facebook is going to do the same to AP, like they did to XMPP and RSS, if we let them.
@oblomov sure it could have been solved, but it wasn’t. By anyone. And the moment from desktop messengers to mobile ones happened quickly and that became the focus. I think overlooking the very real technical challenges that led to everyone essentially dropping XMPP is a mistake. If Google did a rug pull as a way to try to do audience capture, it sure backfired. They don’t have a coherent messaging strategy today, let alone a real competitor for iMessage/WA/Signal/Telegram.
@film_girl FLOSS developers did work on solving those problems. The corps never even bothered to look into the solution because those issues were just the excuse for the rug pull, as was the mobile growth. Also, the fact that for some of them the rug pull wasn't successfully as they might have hoped is irrelevant. It served the purpose of killing IM interoperability just fine, siphoning out the momentum from its development. And that's what we're looking forward to for AP?
@oblomov they didn’t tho. They really didn’t. I still remember this post from a decade ago b/c it really solidified to me the challenges that XMPP faced that it just wasn’t designed to handle. https://op-co.de/blog/posts/mobile_xmpp_in_2014/. Matrix is the true successor but we’re in an app silo world now that I sure hope will break apart at some point. If a protocol is good, people are going to use it. Again, Google gained nothing if the plan when not using XMPP for hangouts was to dominate the market. They LOST market
@film_girl oh please, as if any of those were showstoppers. Unreliable online indicators are basically a staple of the proprietary protocols. Cross-device logs? Store everything in the server forever and send it to clients when they connect. E2EE? As if FB or Google offered that. Not even WhatsApp pretense of supporting the Whisper protocol can be trusted, with their app. And again, Google's rug pull being a failure doesn't excuse them, and most definitely doesn't excuse FB.
@film_girl Well put. There’s also an 900 lb gorilla no one wants to talk about. Engineers who know very large scale systems seem quite skeptical that decentralized will ever survive a surge of massive user accounts with the software we have now. I still want to know if any AP server yet can handle one account with a million followers. Anyone know of one doing that?
@film_girl curious about your perspective on why centralized-acting services are what it takes for a protocol to win. (i agree with you, for the record, but look forward to your take)
@austinha because to paraphrase Jim Barksdale, there are only two business models, centralizing and decentralizing. But honestly, b/c you need to cut the friction to get people to use these thing and it is going to be natural for one or two servers to become defacto dominant places. It's similar to email, right? It is open standards but most people use a hosted option from a small number of webhosts. But that doesn' t mean you can't still benefit from IMAP and SMTP if you roll your own
@austinha Mastodon, no. Now, you could have Mastodon instances that theoretically become “centralized-acting" and then adopt their own features or flags. Mastodon.social really isn't that as much as its detractors try to pretend otherwise (it is much more of a vanilla or reference-implementation IMHO), but something like Medium or Mozilla's instances could be. If/when Tumblr gains ActivityPub support, I think that will be centralized-acting -- even tho it won't be on Mastodon necessarily
@film_girl ok, this helps a lot 🙏 maybe it’s more about being centralized-feeling, then? like, the current refrain is that UX is the biggest problem in the federated social.
@austinha right, centralized by default. So you don't have to know/care that it is decentralized. That's a nice bonus, but it isn't necessary to know or care.
Add comment