>>=

this is like a complex way of admitting that what u want out of life is a trans woman with a codependency streak

I still have this nervous twitch from my Tumblr years expecting any reaction to be the start of a dogpile

I do love how on this stupid website you can reply saying essentially "capitalism bad" and get met with thunderous applause. Social media really is something.

Skeeter needs to not send me a distinct push notification for every single like and reply this must be horrendous for people who are actually popular.

Similar story with news media buyouts. You might take a loss on the purchase, but when it buys you control over the civil society, it's not looking so bad for the rest of your assets. Influence is a profit stream for a diverse portfolio. Especially now that they're just stealing from the treasury.

It's also about profit. Compare lobbying: you might spend hundreds of thousands, even millions buying influence. This is cheap compared to either the profits you immediately accrue from favourable legislation, or the long-term risk of politicians who might undercut your investments gaining power.

It may be that a joint financing model could do the trick, but inasmuch as free software is more interested in making good software and keeping the ecosystem healthy than getting large and capturing value, it might always be screwed, and inasmuch as it's not, it may sacrifice being free software.

The yucky and dysfunctional version of this was when everyone got into DAOs a few years ago. That was silly. More practically, the Linux foundation is fairly successful at funnelling money towards infrastructure code, but not in a way where working in free software is much of a career.

More cynically, someone needs to crack a profit model for free software where maintainers of the software have a better solution for accessing the funding necessary to keep alternatives to the tech monopolies running besides begging for donations like they're Jimmy Wales if he worked for We.

Hopefully renewed European state interest actually gets some money into free software. If there's one model that might actually work, it's a bunch of nonprofits sustained as a public good in the security interests of the states that use them. But that feels overoptimistic.

This all comes to mind because I wanted to use Matrix about 7 years ago. I learned that, at the time, it was janky and the foundation was cash-strapped and the protocol needed a lot of fixing to deal with its security issues. All of these are still the case, but now also the French military uses it.

Firms will fund free software if its development benefits them, but if it's feature stable and not hugely significant, or if it wouldn't make much difference, they'll just free ride it. This is not great; sure, anyone can contribute, but it takes a real sicko to maintain a project for free.

Not all free infrastructure code helps a profitable firm secure a competitive advantage. Google made Chromium to control web standards and dictate everyone's access point to search, but WINE was struggling until Valve figured out it could fund it to undercut their PC platform monopoly.

So there's no economics of free software because there's no economics of software. Software in general is gratis, but big-to-huge firms have monopoly pricing power over managed networked software services and charge rents. But that still doesn't explain why infrastructure maintainers are broke.

The ecosystem controls the software services the Apple user accesses, the software services make the money. All profitable tech firms more or less work either like this, or the version of this where they sell managed hosting and networking services under a similar model.

Free/nonfree code can be preferred for different reasons, depending on its effect on the competitive environment, which can be important, but if Apple open-sourced iOS tomorrow, a chasm would not open up beneath them. Apple being the only distributor is just part of how they control their ecosystem.

But that's all because software businesses are largely indifferent to the copyright model of the code they're distributing, if they run a SaaS business (or an enterprise consultancy, infrastructure, and support business, in RedHat's case).

So I should adjust my claim here, because there are businesses that make money by distributing free software, and all the big tech companies are among them. There are businesses that specialize in mostly or only maintaining the free software ecosystem that make money.

But staffing requirements are high to get a service off the ground at all, and relatively smaller and smaller as it scales, making it possible for a large provider to extract huge economic rents on a software service.

A networked software service is both useful and labour-intensive, so it's worth someone's while, particularly a business' while, to pay someone else to do it, rather than hire their own staff to run their own servers.

The past decade and change has been the generation of SaaS, because you can't charge for software, but networked software isn't easy to set up on your own and can have a lot of benefits, like files being available for access on any computer, anywhere on Earth.

It's not so much that "free doesn't mean free as in beer" is false in practice. That is the case, but it's true in general that copies of your binaries aren't the part of software that makes money, no one has had a business in selling code in years, because copying over the internet is free.

The really unfortunate thing about the economics of free software is that, broadly speaking, there aren't any. A lot of free code has come to be infrastructure and is maintained purely as a hobby by one person who's grown to hate their hobby.

What is it about being on Twitter-alikes that makes people so volatile and cruel? Perhaps such people are just the sorts drawn to Twitter clones. Maybe it's a platform thing. All I know is I wasted way too much time finding out why that fork happened and I am no happier for it.

>lab considers migrating to matrix
>time to do my due diligence
>dendrite is still kinda borked and synapse is bad
>but there's performant feature-complete Rust homeservers!
>more than one
>they forked
>there's so much drama
oh no

On that front, tech is also obscurantist and manipulative about things that aren't slapped with the "AI" brand. I'd rather be less worried about these cars not really being AI because they have a safety fallback, or because GOFAI isn't AI anymore, and more worried about the total lack of regulation.

Whatever the case, it has not seemed to me like Tesla or Waymo are transparent about what kind of "AI", if anything we'd recognize as AI, is responsible for making decisions on their cars. I'm not even fully convinced this is an AI issue, unless the issue is tech being obscurantist and manipulative.

I could be pretty off-base here, as self-driving cars aren't really my field, but it seems to an extent they're doing a pretty GOFAI-ish event-driven decision making thing, with a lot of handcrafted rules, rather than using RL heavily. That may have changed since the last time I read on them.

In particular, autonomous cars use computer vision (which is currently fading out of being "AI") and God knows what else (Musk at some point seems to have suggested there are TNNs in Teslas, but who knows what they're doing). They are branded AI, but we're not sure exactly what they're doing.

It's fairly well-known that "AI" is whatever we can do with computers right now that has yet to reach a mature level of understanding. In which case, it seems like AI is meaningless as a broad category for critique, unless it's taken to include excel formulas.