scientists narrowed down a “sweet spot” of between 6.4 and 7.8 hours of sleep per night

If you sleep less than that on weekdays and more on weekends, does it average out?

Even if it worked perfectly… isn’t clerking the way future attorneys and judges are supposed to get experience?

So maintain chip volume, but switch production from memory chips to corn chips.

If one is opposed by 53% and the other is opposed by 71%, a title suggesting that either is “popular” is a bit misleading.

I think the group of people spending the majority of their lives communicating online would be the first to insist that people who spend their lives online shouldn’t be put in charge of anything in the real world.

The killer feature is the lack of connection to Meta.

What if the risk of self-harm is a result of using ChatGPT?

the whole reason it was possible to experiment cheaply and come across this serendipity was because 9 months ago, faced with the choice to either do the bad easy thing or the good nothing, I chose to do the bad easy thing. The SQLite database worked! I understood how it worked, behind the scenes with its B-trees and its Full Text Search extension.

I always wonder what the tradeoffs are for using SQLite for personal projects… being a stepping-stone to rolling your own non-sql alternatives is one I hadn’t thought of.

RSS feeds seem like the obvious replacement—how is your vibe-coded solution better?

Did he try offering them FIFA Peace Prizes?

It doesn’t take much to boost the price of a stock by 400% if the stock is already practically worthless.

Or 5/9 of the Supreme Court.

The title suggests that the government is pushing OpenClaw specifically, but the text says the opposite:

Yet as more ordinary Chinese get hooked, the government is pulling back. Chinese authorities have stepped up warnings of security and data risks and instructed government agencies and companies in sensitive sectors such as banking to curb OpenClaw’s use.

Yeah—but in theory you only need to train once, while inference costs are ongoing and scale up with usage.

I guess it’s ultimately a business decision by AI companies to weigh how often retraining is worth the cost.

TurboQuant, meanwhile, could lead to efficiency gains and systems that require less memory during inference. But it wouldn’t necessarily solve the wider RAM shortages driven by AI, given that it only targets inference memory, not training — the latter of which continues to require massive amounts of RAM.

I didn’t realize the RAM shortage was mostly due to training—I would have thought inference was at least a big a factor.

deleted by creator

First off, that would incentivize researchers to reduce their citations, making it harder for everyone to find relevant related work.

Secondly, it would be trivial to circumvent: just cite review articles from other countries outside this payment system, that reference the works you actually want to cite.

How long before the mice get sued for copyright infringement?

Mixing vermicompost into the regolith seems a bit like cheating.

MediaWiki’s probably overkill for basic wiki functionality, but I use it for the sake of Semantic MediaWiki and associated extensions. But SMW has more of a learning curve, so it might not be worth it for a casual-use wiki.