supafastcoder

A daily update on the best posts of the last 24 hours would be amazing: https://news.ycombinator.com/best

It's fun the first time but it quickly gets boring.

I mean, there’s only so many ways to optimize a black box

YC went downhill during Sam Altman and then Garry Tan pushed it off a cliff. There’s no point in joining YC anymore.

> When I read a self-help book, but it's laced with stories, I lose interest. Just state the point.

The reason why these books are filled with stories that repeat the same point over and over again is because then the idea will typically stick in your head. But some people have better imagination then others and come up with stories themselves when they read about a novel idea.

I strongly believe the potential valuation and financial performance was not the primary motivation for Musk to buy Twitter, it’s simply a toy for a man worth $200+ billion.

$32K is cheap, we got charged ~$110K in Southern California (two kids - two different hospitals - both over $110K). Luckily insurance paid 95% of it.

Is there a solution out there that manages all this paperwork and custom requirements/contracts?

Curious: what do you do when it's not a good enterprise lead but you still don't want to burn them as a customer? Do you point them to the self service solution?

imagine the hubris and arrogance of trying to control a “superintelligence” when you can’t even control human intelligence

I remember burning CD's full of 1 minute songs that never completed because someone was always calling our house line, disrupting the internet connection (and there were no resumable downloads). Good times ;)

I switched to using Instructor/Marvin which works really nicely with native pydantic models and gets out the way for everything else.

It’s as open as OpenAI

> it covered some generally profound ideas that still are relevant as ever and not widely understood

I've tried to read this book over and over again to understand what everyone is talking about but never found the insights that useful in practice. Like, what have you been able to apply these insights too? What good is it to know that we have a slow mode of thinking and a fast way? Genuine question.

I don’t use Claude 3 for anything else than coding, it works phenomenal for that use case. Maybe we’re seeing the emergence of super specialized LLMs and some LLMs will be better at some things than others.

> Google had far more to lose from a "copyright? lol" approach than OpenAI did.

The company that scrapes trillions of web pages has an issue with copyright?

To be fair, a lot of those Upwork job requests are now being written with AI…

I think it’s a cultural difference. I’m also from a non-dubbing country (Netherlands) and I can’t stand dubbed content either. On the other hand people tell me they can’t stand subtitles because it “reveals” what they’re going to say before they say it.

There’s no such thing as a broken system, every system is perfectly aligned to achieve the results it currently gets.

That’s why I suspect he’ll be joining Meta next.

> A worker in country X or country Y are very different for companies' balance sheets

Yes, but quite often, workers are in the same country (or even same state!) and still get paid differently based on CoL.

It would probably be a 5 euro wrench, they don't use dollars in Germany

Back in 2018, when Tesla was still worth $55B and Apple had more than $100B in the bank, they could've acquired Tesla in an all cash deal, but for some reason didn't. I think they would've made a great match.

> shitty browser

Well, well, even the iOS Safari that shipped with the first iPhone was miles ahead of anything else out there.

> No rules or laws that require it

It will just be FTC knocking on your door…

These demos are always the same: “<Product> find me the closest pizza”

Who actually interacts with a device like that? Is it an important problem enough to be paying $199 while I can do the same thing (often even better) with my phone?

Yes, but an LLM can’t distinguish out of the box what needs to be cited or not. So you need another processing layer (and probably a search engine layer on top of that) to do that.

Yep, that’s what I meant with the reverse lookup strategy.

You’re right, but I think it’s fundamentally impossible to do with the current state of technology. Imagine the word “democracy”, would you be able to tell me where you’ve first learned the definition of it and be able to provide a direct citation of it? What about the thousands of other instances where you’ve seen that word defined? (In OpenAI’s case probably hundreds of millions)

Current LLMs work in a similar way, the information is synthesized by correlation, there is no way that it can directly relate where it learned it’s output from. The only viable way would be to do a reverse lookup of the output to find out if there’s a similar worded content on the internet. (Or what Perplexity does, wrap an LLM around search results, but you lose a lot of flexibility in this case)