ignore xkcd 927

just make stuff. make new standards

xkcd 927 has done a lot of harm, same with the term NIH (not invented here)

just because something exists, doesn’t mean you shouldn’t make your own version of it; your own twist on it

the term “fragmentation” is dumb, we need more stuff, not less. we need as many independent implementations and standards as possible. monocultures suck

all monocultures do is entrench the existing standards and implementations. it gives them power

look where relying on linux all these years got us. it’s just corposlop now

@lumi different approaches are better suited to different usecases. simples.

@izzy that, as well! and how can we come up with something better if we keep pointing at existing stuff?

@lumi @izzy polyproto solves thi-

@star @izzy polyproto is very cool :3 and so are you!

@lumi @izzy But seriously, p2 is mostly an amalgamation of various specs and protocols that have existed for longer than I have. And in many ways, I am convinced that that is one of it strongest upsides

@lumi it depends. standardization is good sometimes. please do not make yet another incompatible and legally questionable license just because you didn't write any of the existing ones, for example. also, standardization on USB-C is mostly a good thing and ISO 8601 is a good date format (please don't start inventing your own and using YDYD-MM-YY or something stupid like that)

for communication protocols standardization is also a good thing, the fediverse would not work if every implementation had created its own protocol instead of using activitypub

@noisytoot yeah it definitely depends, though i do feel there it is better for people to make lots of stuff and then see what ends up being used

@lumi i mean yes and no…

the comic is mocking “one universal standard that covers everyone’s use cases”. which yeah is usually a silly idea.

but some people take it to mean ~”you shouldn’t have more more use cases”. which is also a silly idea.

also the comic’s examples are a bit outdated. (before USB c; unicode adoption was lower)

@lumi @noisytoot I don't know. Sometimes? Sometimes not? I'd definitely agree that it should always be possible for people to do so, on many layers; it should be accessible to do this. And of course anyone can tinker on whatever they like. That's most definitely great not just for oneself, but for the larger community as well. But on a bigger, coarser level, I think we ought to work together more closely again

@lumi it's ironic because the comic can also be interpreted as taking potshots at monoculture too, like just reading the text it's literally about how you can't just make "one unified standard that covers everyone's use cases" and it's foolish to even try.

@lumi

Reminds me of James C Scott's Seeing like a State, and its cousin Seeing like a Software Company

https://www.seangoedecke.com/seeing-like-a-software-company/

tl;dr: a centralized, hierarchical power will want control over everything, and that is best achieved by making everything _legible_. Make it reportable to your hierarchy. It's worse, it's less performant, useful, cheap, a monoculture of cultivated sameness, not a problem: as long as the hierarchy can control it, it's good, _even though_ the real work that actually makes it not crumble apart and makes what the clients actually want is off-the-books

The anarchist view goes completely the opposite: favour the little things, favour the backchannels, the stuff that doesn't scale, to create a mesh of completely different but ultimately interdependent people working directly for each other

Some even go further and say this is in line with anarchist thinking:

https://eli.li/the-seed-beneath-the-snow

#anarchism

@lumi
I can't re-toot this hard enough

@lumi The phrase "Don't reinvent the wheel" predates both of those being used to stop people from making things.

So even the people saying not to create new versions of things aren't following their own advice.

@lumi @noisytoot

My hottest take is that the USB-C standardization has been a nightmare for usability because you literally cannot ever know if one device will work with another at all ever because they all use the same port. Of course I also just think that USB-C is a bad design but that’s another topic.

@lumi fragmentation is multiple implementations of the same loose standard being implemented in incompatible ways, making it so that even if the standard being followed is the same everywhere, they aren't actually compatible

posix can be seen as such a loose standard, but my go to example is definitely sql, as the incompatibilities are much more blatant, and you WILL run into them

@lumi Yeah, although for me standards ought to only be multi-implementations documentation/specifications, the de-facto thing leading to monocultures ought to be called something like hegemony instead.

Like good standards are ones that encourage different implementations to participate in a common ecosystem (which is part of why some standard bodies are so infamously bad, as they're not doing their job properly), and typically tends to work in an implementations-first manner.

btw reminds me that for quite few years I've described x86_64-*-linux-gnu as been my Windows, as it's what proprietary and badly interoperable software nearly exclusively targets in the Unix space, and so often leads to things like emulation being made for it.