golergka

I’m not sure what else can you use to share the same business logic between website frontend, desktop gui and tui apps and backend

Marketing could of course be one of the main motivations, but it's not a "stunt". More like a marketing achievement, I guess? Stunt implies smoke and mirrors, and bun rewrite is quite real.

Needless to say, they also collect all the data and sell it to labs which want to distill the models they’re serving.

There’s a hugely widespread business model where criminals steal people’s API keys, then sell people (mainly in China) access to models through their proxies for lower prices, and then of course save all this data and sell it to Chinese labs for distillation.

Back in 2000s tech was about websites to share "I can has cheezeburger" photos and now we're literally in the middle of technological revolution that seemed like scifi less than 10 years ago. Not only AI, but biotech, space, fusion, robitics, even gussian splats. Solo software engineer can't create on of these now. But there's so much more invention of genuinely new, exciting new things going on that we can be involved with.

Yes, big corps suck. Always have, always existed, 20 and 40 years ago. Startups of 20 years ago have become big corps of today; now, new ones are in this "youthuful vigor" stage.

Compared to 2011, HN turned from optimistic techno-enthusiasts who couldn't wait to see new stuff around the corner into an aging millenial club, just like reddit. I used to come here every day and talk to people, and how I hardly do it once a month.

I think it's mostly demographics. Most people my age (I'm 38) have to come to terms with the fact that their most energetic age and most of their opportunities are already behind them. When faced with uncomfortable reality, many start looking for something or somebody else to blame for this. Not only billionaires, but tens of thousands of millionaires have played their cards better than I and others like me. They had the same opportunities, sometimes even less, but they have shown better judgement, strength of character and pure talent. Admitting this is not pleasant.

Except they’re paying $30b (the deal is signed for almost 3 years), there’s no reason to believe that SpaceX maintains revenue multiples and this deal creates a trillion in value, liquid cash is not the same as pre-IPO shares. And finally, the deal comes down to $11 per hour of h100 equivalent, which is pretty much within market which experiences a severe lack of supply.

After building some toy LLMs on my own I came to realise that architecture is not the hard part. Train is.

attempt != release to customer

when you're building a feature and have different ideas how to go about it, it's incredibly valuable to build them all, compare, and then build another, clean implementation based on all the insights

I used to do it before, but pretty rarely, only for the most important stuff. now I do it for basically everything. and while 2-4 agents are working on building these options, I have time to work on something else.

what's your plan to get all this data and maintain it?

it's not about the tokens. agents can script clis.

Imagine the incredible programs, websites, games, crafts, artworks, animations, performances, literature, journalism, hobby clubs, support groups, community organizations that would spring into existence if we all just had more bandwidth for them while having our baseline needs met.

If people find these things useful, they can actually pay for them. If you can't find people who value it enough to pay for it, then may be it's not as valuable as you think it is.

Most of the capital is owned (through pension funds and then different investment funds) by completely ordinary people who put their completely ordinary savings into it and get completely ordinary pensions out of it.

So you are effectively suggesting to forcibly take their money from them and then give it back, but through a corrupt and bureaucratic system of the state.

CEO profession is a magnet for male psychopaths, and social worker for female ones.

Anything you could automate you could do yourself. What’s the benefit?

As somebody who worked on two IDEs which didn't fork VSCode but still used Monaco for code editing views, I think forking VSCode is almost always the right solutions for a new IDE. You get extensions, familiarity and most importantly, don't waste valuable time on the boring stuff which VSCode has already implemented.

Nothing bad with using code other people made open. Our whole industry is built on this.

Obsidian.

ChatGPT/Claude can be absolutely brilliant in supportive, every day therapy, in my experience. BUT there are few caveats: I'm in therapy for a long time already (500+ hours), I don't trust it with important judgements or advice that goes counter to what I or my therapists think, and I also give Claude access to my diary with MCP, which makes it much better at figuring the context of what I'm talking about.

Also, please keep in mind "supportive, every day". It's talking through stuff that I already know about, not seeking some new insights and revelations. Just shooting the shit with an entity which is booted with well defined ideas from you, your real human therapist and can give you very predictable, just common sense reactions that can still help when it's 2am and you have nobody to talk to, and all of your friends have already heard this exact talk about these exact problems 10 times already.

Wha if you asked the blind man to play the role of helpful assistant

So effectively what you're saying is that institutions are willing to reward her with attention, but people are not willing to pay for her art after all?

Looks like this version finally works like an ultrawide external monitor.

Especially when you work with a language where an unfinished refactoring with give you the type error.

As a DJ, the endgame is building a set from a variety of different kinds of music which still sounds great together but doesn’t all follow the same boring formula. And it’s pretty great.

Ironically, LLMs are exactly what drives a lot of curiosity and learning without a purpose. I see it all the time on twitter — people getting chatbots into weird mental states, toying around with different systems on top of them, jailbreaking. More for the fun of the game than anything else.

You can't keep that curiosity and at the same time see one of the most wonderful and awe-inspiring technologies of the last decades as something threatening.

Just run them on every commit in develop branch after the PR is merged. Then you can always roll back or revert the merge and leave it to the PR author to fix it.

All E2E tests should run before deploy. Probably on every commit on develop branch, even. But there really is no need to run all E2E suite on every PR. In this case, the failure mode of this system, where PR automation failed to flag a breakage, is acceptable if it’s rare enough, so probabilistic solutions are OK.

OpenAI spent 50% of its revenue on inference compute costs alone

This means that they operate existing models with very healthy 50% profit margin, that’s excellent unit economics actually. Losing money by investing more into R&D that you make is not the same as burning it by selling a dollar for 90 cents.

if a restaurant abuses its staff

What exactly counts as "abuse"?

Here's what I've seen first-hand in a "labour-friendly" country. An employee doesn't show up at his workplace a few days a week, for several months, without doctor's notes or any real reason. Employer finally fires them. Employee goes to court and after a year gets a $20k compensation for "unlawful termination", even though his absence on the workplace was documented (but not properly processed, apparently).

I did a couple of quick searchers with help of ChatGPT, and it seems like in Norway, at least, a tenured professor would get ~$50k post-tax, a primary school teacher ~$35k, and a cleaner ~$20k. If anything, such low income inequality seems dystopian. I would expect talented and ambitious people rather move elsewhere.

So all engineers who have already been using better AI tools and had no reason to downgrade to copilot and cursor were fired?