Let's start with the fact that payroll and subscription tier can't be "giving back". That's just running a business. Whisper was a fig leaf when it was an open-everything era. Where are frontier weights now? Anthropic's papers describe how they did it, not who they did it to. That methods section isn't a royalty check for millions of people whose work is baked into those weights.

"Too hard to compensate fairly" is demonstrably false. If ASCAP figured out fractional royalties with paper ledgers, so can high tech. It's not _hard_, it's expensive which is _the actual reason_, but it's dressed up as logistics.

"Lizardbrain jealousy" bit is reducing critique to tribal envy and that's a move in the debate when your on the positive side of the curve and you'd prefer the conversation to end there. It's stupid, come on.

What I wanted to contrast it with is the non-capitalist version you kind of waived off. Publicly funded research, open initiatives, public private partnerships.. that's literally how we got transistor, the internet, GPS, mRNA, hell even math underneath the transformer. Capitalist layer didn't invent any of that, it wrapped subscription around it. The actual productive parts of the stack came from exactly the model you're mocking. OTOH, credit where it's due for Google's research arm that actually contributed back at the foundational level. An exception to the rule, if anything.

China and EU are somewhat doing the right thing though. Not at the scope yet.

So, what's the deal?

As with ANY work in life, the quality of the result is a direct reflection of care and intention behind it. Simplified, it's a reflection of how much _you_ put effort in it. It always shows. Even in AI day and age. It's just that path to a result (without effort) is now way shorter so volume is showing up and diluting the overall impression. The latter kind of cheapens every field it touches, so even more effort will have to be put in to show up on the radar.

being key here. if you realize it's not all that sane when you start reviewing things, what happened in our case was it allowed us (there were other signals as well) to regroup on our practices and then it was painless.

I don't think you would be able to be compliant as a solo dude though, not easily. A bunch of protocols and practices revolve around governance, handovers, failovers, risk mitigation etc and if you're the only guy there's a hard path ahead. Are you reviewing and approving your own code that goes to production? If things go down and you're the first to call (let's say by automated alerting) and you're not available, who is the next one to call as in what's the documented succession plan or automated remediation.. etc.

Compensatory controls do not strictly require a human, they require mitigation of risk associated with a single human. You'd have to automate a lot of these governances "gates" then. So it would be possible, since evidence you would have to provide is work not org-chart, but it'd be a ton of work.

I went into it thinking I need to answer these 167 documents and provide evidence on an ongoing basis, but it actually also transformed the way we do things. I think for the better. At the end of the day, I also think this can be gamed as probably most certificates, but it's not worth it and transformation you go through makes sense.

Meanwhile: Error - The following features required to run Godot projects on the Web are missing: WebGL2 - Check web browser configuration and hardware support

This is on Chrome 148.0.7778.96 (Official Build) (64-bit) on Fedora 44, 14900K, 4090 RTX, 128GB RAM

35MB WASM which relies on browser, which relies on drivers and OS, AND it doesn't work as advertised. I get the point, but point was mistaken.

One that does seem to be an odd man out is Genera. What a concept.