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

The whole theme of llm dev to date has been "theres more common than not" in llm applications

RAG is nowhere near obselete. Model performance on enormous sequences degrades hugely as they are not well represented in training and non quadratic attention approximations are not amazing

Think that's the first time i've seen someone write about checkpointing, definitely worth doing for similar projects.

Only 92 antiprotons but still an exciting feat


You (briefly) have an antiproton in your possession around once a day, assuming you get an average amount of sunlight. Some days, you might even have two!


This just in: seasonal affective disorder confirmed to be caused by antiproton deficiency


Assume geometric scaling for production as all MBAs do and we'll have enough energy to reverse global warming by lunchtime.

YMMV. I heard that they weren't able to count them correctly. And they might not even be there.

No way to approach this other than gross simplification, but we have seen generally that as technology improves, standard of living goes up. Autonomous machines breaking that trend, maybe, the only argument I find persuasive is that it makes the centralisation of power easier.


Do you think people are happier now than they were in the 80s/90s/00s?

I don’t think so, at least not in the us. Granted I was younger during those times .


Yes, dramatically so. The 1980s and 1990s sucked, we live in the future.


I was very much there during the 1990s, people were median happier than in 2026.

Unless you mean globally, in which case it's near impossible to tell. But GP likely meant the US or Europe.


> But GP likely meant the US or Europe.

Reminds me of movies "Aliens attack Earth. By Earth by obviously mean USA. By USA we obviously mean New York City. By New York City we obviously mean Manhattan.".

Half of Europe was literally undergoing complete economic collapse in 90's.


Your supposed connection of country-level economic situation and happiness reminds me of this great comment:

"I can't stand the conflation of "satisfied" and "happy." It's insane. There is more happiness in one Zimbabwean (country "happiness" rank: 143) than in one hundred Icelanders (country "happiness" rank: 2, worldwide antidepressant consumption rank: 1.) Go stand in a crowd of people and count the fucking smiles and the fucking laughter.

"It is all part of this broader wave of newspeak. If you can quite literally redefine happiness, you can redefine anything. Nothing has meaning anymore. You will live alone, you will consume antidepressants, you will be protected from the sunlight, you will not smile, you will not laugh, and you will be happy." [0]

[0] https://news.ycombinator.com/item?id=46295573


Hmmmm maybe economic success removes discomforts which removes unhappiness but this means that moments of happiness don't feel as good compared to the baseline. Like, people in rich countries are more stably content, while people in poor countries have more emotional variance over time.


Which half?


Look at home prices and then tell me it sucked.


What home prices? There are many places (second-rate cities or rural) where home prices are just fine. Not everyone has to live in SF.


We periodically see posts on HN about the 'loneliness epidemic' and simultaneously the solution for home prices is to live in a rural area where you have maybe two neighbors and it's a 30-minute drive to see your friends or colleagues.

As someone who grew up in a rural area I don't miss it and I think people who call it a good standard of living have interesting criteria. That cheaper rural home also comes with secondary costs, like a longer commute.


Housing prices are not binary: they have skyrocket in every major city and city-adjacent suburb. The middle class has consistently declined since the 70s.


> , standard of living goes up

For some of humanity, perhaps. For the rest of the planet being destroyed, warmed up, bleached, demolished, turned into data centers, all this technology is destructive.


The API/online model inference definitely runs through some kind of edge safeguarding models which could do this.


Because writing is really hard.


Interested if these TUI agent systems have any unique features. They all seem to be shipping the standard agent swarm/bg agent approach.


"This guy is coding everything in the terminal, he must be really good!"


Not terrible if they proactively depricate slop features


PhD students are levy infantry at best with Postdocs being the armoured levies.


Is this Gondor or Mordor?


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

Search: