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

yes, but a few dedicated hoarders could keep many models alive, and I'm fairly certain the local LLM community has plenty of people who would.

They don't. That's the conversation topic.

Nor do they need to. 99% of everything is crap, and not worth prescribing except for a random sample so future historians can study our crap.


scare quoting """""""""democracies""""""""" is perfectly acceptable when it couldn't be more obvious that those so-called democracies do anything but represent the will of their peoples. do you know anyone who'd voted for having to scan their face to access verboten webseites? to have encryption backdoors in their messengers? to have state-mandated spyware preinstalled on their phones?

They don’t represent the people’s will, but unfortunately many people believe in censorship, so even if they were true direct democracies we may still see some.

> if they were true direct democracies we may still see some.

I think we would see more.


With modern technology, we might be in the first era where direct democracy is truly viable. Frankly I think the people should have direct veto power on any legislation.

Well, that's already viable in Switzerland. Any proposed law can be forced into a national vote by collecting 50'000 signatures for a Referendum.

> Frankly I think the people should have direct veto power on any legislation.

Counterpoint: see any NIMBY meetings where they shut down any and all proposals and legislations, even the obviously beneficial common sense ones.

Rule by mob has rarely been productive.


The normal approach is to compromise across topics.

Maybe one should have a finite number of winning votes in a period. Then the majority might want something (or not want) but wouldn't consider it important enough to spend their vote on.

I think that would reduce the sense of urgency where there is non, equalize involvement, foster debate with the barely involved and ofc create a focus on things one wants rather than things one doesn't want.

Who knows, maybe one should be able to spend more than one vote per topic.


> . Then the majority might want something (or not want) but wouldn't consider it important enough to spend their vote on.

Rule of the thumb: everyone will not 'just' [1]

The majority is not a single person that will somehow behave like a rational being that will count their votes and vote only for "things that matter". People view very different things as "things that matter", and the system is easily gamed. Run a few popular, but pointless votes, exhaust available winning votes for the period, then try to pass a meaningful legislation. You will have riots form people who have spent their votes, but still want to vote for something truly important.

And that's before we start talking about whether people should directly vote on removing building code regulations for example. Or whether or not measles vaccines should be mandatory.

[1] https://www.tumblr.com/squareallworthy/163790039847/everyone...


I'm just playing with the puzzle. It's not like I deeply thought about it and this was my conclusion.

Votes per period might not work. You could in stead accumulate votes over time.

Of course the proposals could repeatedly go against a specific group. Not sure if you could create a unit of accounting for it or have some kind of independent jury examine the gaming.

I also wondered if the price could scale with the % victory. Like, if 100% votes A it costs zero votes. 51% would be the most expensive.

Maybe give the proposal makers a bunch of tokens, filling a proposal costs x, each vote against it costs y.

The core thought is basically, if people are overly attached to something they will get it but will have reduced influence on everything else.


Any time you introduce any type of cost, you create a system where it is extremely easy to disadvantage a group of people.

Do you think improvement is possible?

I don't know if it's possible to improve every kind of voting in the same way.

But we could improve some parts. For example, voting in elections could be improved by introducing Single Transferrable Vote https://en.wikipedia.org/wiki/Single_transferable_vote (CGP Grey had a bunch of videos on this: https://youtube.com/playlist?list=PL3897F608FAD61E88&si=TzEP...)


The only good clue I have for a new or modified system is that it has to be simple enough for the voter to understand it. The single transferable vote is technically great but the name doesn't explain how it works. You have to make up for that with education (which seems doable)

I wonder how similar the result would be if the voter could cast one vote for something and one against. This seems less complex.

It has other advantages. A large election budget shouldn't automatically convert to votes. Doing a pis poor job should probably cost more than a few votes.


The same principle would have also shot down the need for any legislation before building a thing.

Yes, I do know people who voted for the party that pushes things like this in my country.

In the UK, both of the main parties push internet censorship. As do the third party and the SNP.

Plenty of people are happy to surrender their freedom for a feeling of security.

GWB had a second term. Trump had a second term. I would find it reassuring to believe that democracies don't represent the "will of the people", but I've come to accept that, much of the time, the "the people" and I don't see eye to eye.

People don’t vote for it for themselves, they vote for it for others and are surprised when it applies to them just the same.

...you'd be surprised, plenty of regular people actually want these things to happen. State is God for reliably the majority of the population of the world (seeing as the majority resides in China and South Asia).

> do you know anyone who'd voted for having to scan their face to access verboten webseites? to have encryption backdoors in their messengers? to have state-mandated spyware preinstalled on their phones?

I know a few. They're called "normal people who don't suffer from crippling tech libertarianism" and there's more of them than of you.


you must have plenty of unambiguous evidence then, given how numerous such goodthinking individuals apparently are.

where do these paragons of civil obedience congregate? how do they express their wishes for less freedom and more surveillance?


They're all around you irl. Touch grass and find out.

And to your sarcastic remarks about those people being dumb, believe it or not not everyone is willing to forgo every possible evil in the name of personal privacy in remote communication. Such privacy was never a thing until very recently anyway.

(In fact as far as most people are concerned it isn't a thing today either, you'd be surprised how many people assume that their messages and calls are spied on anyway)


It was actually illegal for the post office to read your mail or the phone company to listen to your calls.

Yes. But everyone knew if necessary it was possible and was done by police to fight crime.

and so you've spent 20 minutes editing this comment to optimize the amount of snark instead of simply showing me one proof of that silent majority's existence :)

> you've spent 20 minutes editing this comment to optimize the amount of snark

why would i waste that much time on writing a reply to you and where do you see snark exactly?

> one proof

i just talk about it with people irl, yeah that thing i suggest you try;)


oh, but it's the exact opposite. the people IRL tend to react to censorship and surveillance with the word пидорасы, while the regime's eunuchs on the Internet defend it.

maybe having lived under the Soviets makes my people less welcoming to the idea of a commissar deciding what they're allowed to see, hear, say, and ultimately think.


what does that word mean? Wiktionary says it's a vulgar word for "gay person", huh?

consider that not everyone grew up in an oppressive dictatorship.


Isn't the user base of tiktok proof?

> I know a few. They're called "normal people who don't suffer from crippling tech libertarianism" and there's more of them than of you.

They voted for it or you assume they don't mind it existing?


They accept it as a tradeoff. They vote for things like "less crime" or "our children don't access things that are bad for them".

I don't think tool calls happen on the same machines that host the weights, so even though you can talk any model into agreeing to unlock its chastity belt, it essentially has no hands to do it with.

> I don't think tool calls happen on the same machines that host the weights

Like how forums are always hosted on different servers from monorepos, so therefore it's impossible to hack the OpenAI monorepo from an OpenAI forum?


If the machine doing tool call can reach via network the machine hosting the weights then it’s just a matter of time.

Maybe not the current models, maybe not this year. But even a almost perfectly aligned model will misbehave one day.


ditto. I never actually studied English, so I just YOLO the articles most of the time. tenses too, really. whenever I look something up, there are caveats and irregularities, and I give up.

I know, in my heart of hearts, that I sound painfully ESL to proficient native speakers, but they are a minority of a minority, so.


I'll take that over corporate memphis of the previous decade.

(idk if it's really declining or I simply haven't noticed it in a while)


by far the blackest magic I ever saw for that game.

I'm an upper intermediate at Factorio, resorting to someone else's blueprints only for belt balancers and rail intersections, and I can't even begin to figure out how it's done.


unrelated: if a website gets hit with "This URL has been excluded from the Wayback Machine", do existing snapshots get purged or may they still be preserved somewhere?

They get marked as inaccessible, but still exist in IA data

is it possible to access somehow? it seems the site got excluded because of robots.txt set by some domain squatter, not manually.

They have all the WARC file cataloged on their site outside the way back machine

Get a job at IA?

See also: https://wiki.archiveteam.org/index.php/List_of_websites_excl...

Note that Archive Team is separate from the Internet Archive.


idk about Uganda, but some of the other estimates on that website are wildly inaccurate.

In what way?

https://www.populationpyramid.net/ukraine/2026/

40M is grossly overestimated. even UA officials admit it's closer to 20M now.

https://www.populationpyramid.net/russian-federation/2026/

the projected 127M in 2100 makes no sense given that we had TFR<1.5 even before COVID. it's likely <1.2 now, and it will only decline.


Interesting. Yep, I guess the projections did not forecast a protracted military conflict. Although the dent in the population of Ukraine is telling.

Thanks for sharing.


one man's issue is another's man opportunity to overwork and underpay :)

lower wages and higher rents are 99.99% of the reason for the West's immigration policies.


>The open weight model labs don't even provide a high level catalog or any description whatsoever about what went into the training data.

because the training data is full of copyrighted works.

the answer to "what went into the training data" is "everything we could get our hands on".


For text based just assume "the entirety of anna's archive" and you'll be most of the way there. For multimodal pull up a torrent tracker and sort by size descending if you want to see some of the aggregated datasets.

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

Search: