Hacker Newsnew | past | comments | ask | show | jobs | submit | pickledish's commentslogin

clicks

> When a fire truck fails to deploy in a burning building and four people die, the cause isn’t just mechanical failure. It’s a business model.

leaves


When a user clicks the 'close tab' button, it's not just sending a command to their browser. It's sending a signal of disapproval.

This is also the message I got from "the wind rises"! Though from talking with other people that takeaway doesn't seem universal -- which IMO is one of the ways to tell it's a great film :)

Don't know why you were downvoted -- I scoffed at that part too. The least we in tech can do is have some self-awareness


Agreed! I'm not sure why the GP comment has a somewhat negative attitude about it, I think it's great for people to realize this and talk about it, every year a whole new year's worth of young adults turn up not knowing it! Insert XKCD lucky 10,000 comic here


Not meant to be negative.


Ah -- IME people usually start a sentence with "in the nicest possible way" when they're couching something mean is all. Often it's even sarcastic, and the meanness isn't even meant to be couched, so I wasn't sure how to read it


Ironically, I went back and added it because I thought the comment sounded a little bit unfriendly on a first pass. Mission failed.


Agreed, the default connotation / reading of this phrase implies a snide attitude.


They're talking about this I'm pretty sure!

https://www.telotrucks.com/


Yes, those. Thank you!


Hah yeah, I'd never really thought about it, but newspaper headlines must be a late-game boss for non-native speakers, they are super convoluted


Sadly I don't think it works this way, at least IIUC -- the state can't withhold taxes from the federal government, because those taxes (from biweekly paychecks anyway) don't go through the states -- they go directly to the federal government. Some states are trying to pass laws to still make headway in this area, for reasons like you suggest, for example NY:

https://www.cityandstateny.com/policy/2025/10/state-lawmaker...

(it's a really interesting situation since I think I read somewhere that the reason federal income taxes are directly remitted to the federal government today, is specifically to disallow this kind of state retaliation)


Although, Republicans would ostensibly like to shrink the size of the government and Democrats would probably at this point prefer their money to go to an entity that will actually provide services. So I don’t really see why there isn’t a broad consensus for implementing this idea.


>Although, Republicans would ostensibly like to shrink the size of the government

Where did you get that idea? That's never been true. You may just have been hoodwinked by their "Two Santas" strategy.

Republicans have been explicitly playing that game as promoted by Jude Wanninski[1] since the late 1970s, and it's been loudly touted as quite successful by Republicans.

The idea is to cut taxes and spend like drunken sailors when Republicans are in power, then cry poverty and austerity when Democrats are in power, loudly calling for incredibly popular Democratic programs to be slashed.

I'm not making this up. See the links below. It's not like this has been a secret for the past fifty years or anything.

[0] https://www.milwaukeeindependent.com/featured/two-santas-str...

[1] https://en.wikipedia.org/wiki/Jude_Wanniski


Because Republicans don't want to shrink the size of government as a whole, rather they want most funding to go to a massive domestic paramilitary terror squad to keep citizens in line.


Aside from the rest of the (interesting!) nuanced discussion going on the comments here -- I really like his idea towards the bottom of combining the colors for numbers and strings into one.

They're both constants, so yeah, they should be!


Agreed!

And, maybe I'm missing something, but to me it seems obvious that flat top part of the S curve is going to be somewhere below human ability... because, as you say, of the training data. How on earth could we train an LLM to be smarter than us, when 100% of the material we use to teach it how to think, is human-style thinking?

Maybe if we do a good job, only a little bit below human ability -- and what an accomplishment that would still be!

But still -- that's a far cry from the ideas espoused in articles like this, where AI is just one or two years away from overtaking us.


Author here.

The standard way to do this is Reinforcement Learning: we do not teach the model how to do the task, we let it discover the _how_ for itself and only grade it based on how well it did, then reinforce the attempts where it did well. This way the model can learn wildly superhuman performance, e.g. it's what we used to train AlphaGo and AlphaZero.


Yeah agreed!

As you suggest, I've had a moderately successful time trying to get AI to write its own Sublime Text plugins so our favorite editor doesn't get left behind, so might be cool to try with this too?

https://github.com/pickledish/llm-completion


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: