Here's a good discussion from a group chat of mine.
It's mostly between Ilmu, the puristic mathematician, and Alex, the pragmatic research engineer.
The starting point was that Apple spies on everything you do on your Mac.
On modern versions of macOS, you simply can’t power on your computer, launch a text editor or eBook reader, and write or read, without a log of your activity being transmitted and stored. - Your Computer Isn't Yours
Alex: Is it time to start making open source software usable?
Ilmu: It already is. You just choose how close to the future you want to be; the closer, the more painful it'll be.
Debian stable is stable as fuck.
Alex: The open source software community is pathologically bad at UX
Adrien: I'm on Linux because of the UX
Ilmu: Yeah roughly what I was going to say. Vim has great UX. It's just catering to advanced users
Alex: Yeah, it does, for 1970 (I love vim and use it everywhere)
Ilmu: No, it's a timeless idea: modal key binds + DSL.
Alex: I just mean the idea that people are going to learn this shit as a rite of passage or something. Numbers are what matters. Lots of people working on OSS are working for large tech companies to pay their bills and keeping OSS borderline unusable for regular people is what prevents it from being profitable
I don't argue that the internals are often solid, but the community makes little to no effort to involve people with any understanding of UX
GIMP, Inkscape, and LibreOffice could be legit competition to Adobe and Microsoft, but they aren't
Ilmu: Look, the mac way is 'pay premium for a polished UX' — nothing stopping such a market from forming around configuring-reproducible-systems-AAS-consulting
Alex: I think that assumes a lot less centralization than actually exists in that space. Apple, Google and AWS don't want you configuring their systems and they are increasingly the internet. (plus Netflix, Facebook, and some others)
If you don't win the end user, it's a pyrrhic victory imo.
Ilmu: I mean the end user buying a preconfigured laptop from a 'corner store' run by an enthusiast. Idea being you come in and describe how you use a computer he lets you try a few configurations he has in store for interacting with a system and then you can buy one and have it installed.
Kind a like a clothing shop for 21. Century.
Alex: The great VS Code migration thing is a great example of how we can't even win the code editor space. Though I admit it has put pressure on Emacs to get better.
Ilmu: If you want something changed you can talk to a different third party
Alex: I personally don't see that being economically competitive [referring to the clothing shop for 21. Century idea], though I hope to be proved wrong.
I'm just saying it's a problem with the culture. People who already know how to use it (us) don't care and everyone else is like "of course they're going to spy on us, because they can" and they just use what works.
Ilmu: Yeah but all it takes is some competent people building a brand for systems with good UX, kinda like JetBrains but for normal people.
Alex: I think that could carry the day.
Ilmu: That creates the market and bam you have independent people showing up all over the place creating small incomes.
The infrastructure is almost at the point where this happens.
Alex: I'd argue that the infra was there in the 80s, no?
Ilmu: Why would you say so?
Or like what is the argument?
Alex: If anything, it's harder now because you have companies like M$ [Microsoft] weaponizing OSS
Alex: I guess I don't see what it is about today that makes that easier(?)
Ilmu: Yeah but now we have cryptocurrencies and reproducible builds of entire systems!
Ilmu: It's like night and day
Alex: my concern is that the issue has more to do with not knowing how to run projects and businesses than it does with particular technologies
I guess crypto makes it easier to make payments(?) but it was also pretty easy to send checks. I don't know that that's actually transformed much yet(?)
Ilmu: Ehh there were two arguments and the build system point is necessary and sufficient
Alex: As for reproducible builds, it's true that Nix can be a game changer, but I don't know that it had the adoption yet to make that difference. Maybe Rust or the scientific community or something will propel that along.
Ilmu: Yeah that's what I was going to say. The maturity of nix is the main blocker
Alex: I see the potential for sure. Could use smart contracts to organize distributed companies, could crowdsource development, etc. I just feel like the software is 80% of the way there and, if there isn't a conscious effort to improve UX, that last 20% is going to be the difference between outcompeting the giants or not
You could argue that we don't actually want to eat their lunch, lest that become the new establishment, but I think Stallman would disagree---the dynamics of [F]OSS are such that it's a different beast entirely
Ilmu: Plants can grow quickly or accurately, but not both - https://news.ycombinator.com/item?id=25092648
Ohm's law / CAP thrm
Alex: I don't disagree with that trade-off in the abstract, but I would argue that deciding on objective functions and planning in general are somewhat orthogonal.
One can move toward a suboptimal goal with whatever balance of quickness/accuracy