They’re just doing what they’re told to do by the shareholders. They’re the largest cogs sure, but cogs nonetheless, and they’re replaced just as quickly when they fail to make choices that maximize profit above all else.
Should we appreciate art we were manipulated into liking, or pretending to appreciate?
Modern art has been the butt of many jokes over the years. A canvas painted white, a banana duck taped to a wall, literal piles of trash… the bar for what we call art is basically on the floor. A literal bar on the floor, given a good story, could itself be called art.
Art generally implies some level of skill. While some modern art can still fall into that category, other stuff requires no real skill, just the right connections and for enough people to buy into the grift.
I love art. Contemporary art is a bunch of ugly wankery by pretentious people who are bored of actual beauty and can only amuse themselves by creating ugly stuff, then claiming "you just don't get it". The world would be a significantly better place without contemporary art in it.
If there is any change I could make to the human personality at large it would be the reflexive opinion that anything you don't understand, or even dislike, is automatically morally wrong.
This might shock you, but most art is made by artists. Artists also tend to spend lots of their time consuming other art. This informs the type of art they create.
Here's an anecdote: Pablo Picasso learned to paint realistic scenes, including humans, the equivalent of Da Vinci or Michelangelo, around the age of 17 or 18.
Where do you go from there? In his case, he went towards cubism and other such "modern art". Why do you suppose that is?
Moral licensing probably. If they get social points for building a “fair” product they’re less incentivized to be “fair” in other areas. Also why the people who virtue signal the most tend to be the quite rude irl.
If you don't have regulatory hurdles to jump over then there are a whole raft of open-weight models that perform basically as well as Anthropic's offerings (and, in many cases, better because they're pulling fewer capacity-stretching shenanigans). Check out Kimi, DeepSeek, and GLM series models in particular with something like OpenCode or Pi as a coding harness. If you need a Claude-alike web UI then spin up Open WebUI in a container and expose it via cloudflared so you can use it on the go.
Kimi and DeepSeek are impressive for their size but still run slow on any hardware you or I would have. If you’re proposing we run those via a cloud service, I don’t see a reason to do that when it’s an inferior model and I still have to pay per token for it.
OpenCode Go provides a lot of usage for these models for the paltry sum of $10/month. Z.ai's coding plan provides a single-digit multiple of Claude Code's usage for a similar price and performance level. Kimi and DeepSeek models are hundreds of billions of parameters (or, in K3's case, >1T). Many of these models have Opus-level benchmarks and, as I pointed out previously, often practically outperform Anthropic models because they're more consistent.
It’s not, you drink too much Kool-Aid, and your employer has too much money. Or the things you do are simply so irrelevant that nobody cares you’re sending stuff to a company of dubious morality.
Well which is it? Because the first one also contradicts the claim i'm responding to. I think those thinking their codebases are so novel as to not dare to provide to the providers are quite full of themselves. 99.9% of code is mundane.
Nobody says novel. We just ”can’t send not our own intellectual property to a third party because contracts”. Then we roll out on premise. And we don’t miss anything, in fact we feel we gain more than we lose out. With proper tools we are not far off.
Personally I don't care about common benchmarks as I don't find actual coding agent performance correlates strongly with them. One reason to use open-weight models is that they don't hide the reasoning, so you can do very aggressive context management in your harness to use significantly less tokens. Smaller prompts are faster since KV is N^2 plus it can be dramatically cheaper (if you balance your aggressive context management with maintaining the prefix cache as much as possible). Even paying for API prices directly and using agents as much as I want I spend less per month than the $200 I spent on a Claude MAX sub when I had it.
Yeah, wish I’d ignored them sooner. “Spammers” don’t care if they bother people with offers that have no potential to benefit them. But actual entrepreneurs wanting to serve people with their products need to know that cold outreach is essential to get off the ground, and that they should fiercely ignore the people that call it spamming.
It's just the same thing viewed from two perspectives. For the sender, it's cold outreach - even when it's about some Viagra clone. For the receiver, it's spam - even when it's about AGI for $5.
reply