News for hackers. Stuff that matters.

sarchertech

Re:Last night an AI agent performed brain surgery on a mouse on Tuesday November 03, 2026 @07:45AM (#50351351)
I have interacted with you before and I’d like you to know that you are very unpleasant to talk and you have atrocious debate tactics.

The way you act here makes the world just a little bit worse.

I get that telling you this isn’t likely to be constructive, but if there’s only a 0.000001% chance you reconsider acting in such an unpleasant manner in your next interaction, it was worth it.

Re:Last night an AI agent performed brain surgery on a mouse on Tuesday November 03, 2026 @07:25AM (#50351158)
Let’s agree there’s a 0.000001% chance this study leads someone down a dead end and ends up causing harm.

I don’t agree that it had any measurable chance at all of helping anyone other than the author.

And if you’re being purely utilitarian, you need to weigh the opportunity cost of what else the money spent on this could have been used for. What else the researchers and peer reviewers could have done with their time, and what they else the mice could have been used for.

> is one of the least effective ways you have of influencing policy.

Sure but it’s also the easiest, so from a cost-benefit perspective, it’s an attractive option.

> That's not true at the local level! Public comment is a much bigger deal locally.

That’s a good thing. Giving people thw ability to meaningfully participate in the process beyond voting every 2-4 years is a feature. Even if it sometimes slows progress or reduces efficiency.

Re:Last night an AI agent performed brain surgery on a mouse on Tuesday November 03, 2026 @05:29AM (#50350222)
The problems with using an agent for surgery aren’t mechanical. There’s no gain by using a live mouse at this point other than getting attention for the researcher.
That’s not true. For federal comments you can google for a list of explicit changes to regulations that came out of public comments.

Even for emails and form submissions to members of Congress a 2021 study showed that congressional staff frequently review their correspondence records before writing speeches and drafting legislation.

To the extent that it is ignored, we should be encrusting legislators to listen not telling them to ignore comments because they skew negative.

Moral panics seem to be an inherent feature of democracy and they are problematic. But the solution isn’t to remove or curtail the average citizen’s ability to petition the government to stop something they feel will harm them.

As far as power plants go, to march the current estimates for data center demand over the next decade are astronomical. A large percentage of that power generation is going to be natural gas and biomass. You can’t really build these massive data centers on the timescale needed without them.

Is it stupid lump in a small data center with these massive projects? Yes it is.

Re:Last night an AI agent performed brain surgery on a mouse on Monday November 02, 2026 @09:53AM (#50337636)
This was an unnecessary experiment that was only done a live mouse (vs a simulation) for the attention grabbing effect.

Obviously an AI agent can perform a simple surgical procedure. The question is how to get it to perform reliably.

The reason AI agents work as well as they do with code is because they get feedback and retry in a loop. Surgery is inherently inflicting “harm” on a patient, so there’s no bright line guardrail to apply to the agent loop.

Agents performing surgery won’t be possible until models get much much more reliable or we develop a simulation that matches reality close enough that we can use that in realtime for their feedback loop. I think we’re likely very far off from either solution.

They’re trying to build one in the middle of nowhere about 30 miles outside my city.

But they also want to build an on premise 1000MW natural gas power plant and a 500MW biomass plant that will take millions of acres of forest to sustain.

I had a conversation with the OP a while back where he was posting about how people are much more likely to show up to speak against something at a town hall and how he was talking to elected officials about this bias.

I’m this very thread he is asserting that we should get rid of the public comment process which in many cases is just filling out an online form because it is biased towards naysayers.

The naysayers inherently feel more strongly, and so are more likely to make their feelings known regardless of the process. What other alternate is there than making a (maybe educated) guess.

You realize that the majority of “public comments” in the US are made by filling out a web form.
In many cases public comment doesn’t require showing up, it requires filling out an online form.

To the extent that it does require showing up, we should do things like hold multiple sessions at different times of day.

But none of those issues is what the OP is talking about.

“public comment process, which sharply biased towards negative comments structurally”

He’s saying public comments are inherently biased towards no even given equal opportunity to participate.

Telling people to ignore it when 100 people show up to your town meeting to say don’t do something because there’s secretly 500 people at home who want you to do that thing but aren’t doing anything to externally to indicate their wishes isn’t what I’m talking about.

I’m saying do things like hold town meetings multiple times at different times of day so people with different schedules can attend.

Not try to guess about people really believe. Give them easier opportunities to tell you what they believe.

In a comment a few weeks ago he literally said exactly that.

>local political participation

Voting skews older as well. If you don’t participate you don’t get a voice. We can’t try to divine the public’s true intent.

We should be making it easier to participate, not giving politicians an excuse to ignore those people who do participate.

No he is on record as telling elected officials to ignore people showing up to town meetings, writing letters etc… because he believes that negative comments are overrepresented.

Which may not may not be true, but you could say the same thing about votes on a referendum, or names on a petition. One side or the other is in most cases more likely to turn out. That is just how democracy works. You can’t divine the true will of the people, you can only act on their public participation.

Anything else is just a license to politicians to push through anything they please.

A company is trying to push through a very large wood burning power plant a few miles down the road that will impact air quality for half the state. If people convince politicians that dissenting voices are overrepresented, that’s one less tool we have to stop things like this.

Re:MicroZig to Adopt Strict No LLM / No AI Policy on Wednesday October 28, 2026 @09:46PM (#50283554)
What if the behavior you don’t want is reliance on the technology?
Re:CNC Stone Carving on Wednesday October 28, 2026 @06:46PM (#50282018)
You could probably get something that looks very close by 3d printing forms and casting it in tinted concrete.
No one in Congress, no TSA administrators, no President, and no judge has ever consciously had the thought that airport security was beneficial to them because it enhances state authority.

The TSA exists because it

1. Made some people look good

2. Made some people money

3. Was thought necessary by some true believers

The incentive structures aren’t aligned in a way that any individual cares about reinforcing state power like that. Mostly the people in charge care about getting reelected, keeping their jobs, making money, or in some cases doing good.

TSA lines suck because you have a bunch of hourly workers directing a a crowd of clueless people through a multi step process.

I just got back from Disney World with the family. We flew out of an airport with TSA, went through dozens of security lines at the parks, and flew out of an airport that doesn’t use TSA.

There was no meaningful difference interactions with any of the security personal. Some were nice some were annoyed, some were bordering on mean.

You clearly do have expectations if you are classifying people who don’t meet them as assholes.

People aren’t assholes just because they don’t accommodate your every preference.

> Why is it unreasonable to say, "I'd prefer something other than USian"?

Because all I’ve done is to fix my minor annoyance by making it the speaker’s minor annoyance.

And many people don’t like the term American, so now the speaker has to decide whether to please me or other people in the group, or come up with some ridiculous compromise like Citizen of the United States.

A big part of being an adult is dealing with minor annoyances.

Maybe you shouldn’t sort people into allies and assholes based on whether they are willing to change decades of language use to accommodate something that doesn’t even offend you (or as you say anyone else).

I try my best not to offend anyone at work, but policing my language to match the preferences of everyone around me would be exhausting. I grew up speaking a non-prestige dialect. I have a lifetime of experience policing my words. It can be exhausting. It’s easy to underestimate the mental cost if you speak a prestige dialect.

I spent a few decades being made fun of for saying y’all. I now find it a bit weird people who would have made fun of me for using it 20 years ago now use it themselves. I’m not offended by it, but my preference would be that non-southerners don’t use it.

However, I realize that it would be a bit much to ask that of everyone I interact with. And if I did it would be insane to label people who didn’t comply assholes.

I prefer that people that people don’t use learnings instead of lessons. I prefer not to be referred to as a USian. But I don’t expect people to follow my preferences.

Re:Dolly Parton's Imagination Library on Sunday October 25, 2026 @11:36PM (#50238800)
My wife wasn’t sad because he was no longer getting free books.
Re:Dolly Parton's Imagination Library on Sunday October 25, 2026 @04:17PM (#50235645)
My 5 year old just got his last imagination library book. It showed up right after Dolly passed away.

There was a letter from Dolly to him that ended “I will always love you”. Made my wife tear up.

Re:What shall I teach my son? on Wednesday October 21, 2026 @08:16AM (#50179632)
> It did control for SAT scores through, so the one thing this study is definitely not doing is examining the effectiveness of public vs private schools at effectively educating children

Run that statement through AI and ask it if it says that private education can’t impact SAT score.

> Look, I’m not going to rebut each of your points one by one. Ask AI

Why are you even here? I think you’d be happier just going back and forth with AI all day.

Re:What shall I teach my son? on Wednesday October 21, 2026 @07:53AM (#50179383)
Look at this guy over here with his math and his statistical literacy you guys. He doesn’t even spend all day talking to AI.
Re:Advanced coding in low-resource programming languages on Tuesday October 20, 2026 @10:32PM (#50175353)
They don’t have to produce an optimizing compiler to claim a win though.
Re:What shall I teach my son? on Tuesday October 20, 2026 @10:30PM (#50175338)
The study you’re referencing didn’t examine elite schools just non-religious.

And if you’re only counting elite schools there are so many uncontrolled confounders that this is a patently stupid argument.

Elite schools are selective, a large part of what they are selecting for is kids who are likely to get into elite colleges to keep their success rates up.

Of course kids who are screened for being likely to get into an elite college are more likely to get into an elite college.

The only realistic way to control for this is to look at kids who got into elite schools but decided not to go.

>check with AI

Kick rocks. I’m not normalizing this shit. Why didn’t you check with AI. It would have told you that omitting key information to save a few keystrokes was a bad move.

Re:What shall I teach my son? on Tuesday October 20, 2026 @10:19PM (#50175276)
It didn’t control for wealth though just income. It also didn’t control for parental education or familiarity with elite colleges. It didn’t control for parental expectations (seems likely that a parent who spends money on private school will be more likely to expect their child to attend an elite college).

It didn’t control for the value parents put on education in general.

It didn’t control for children’s expectations (seems likely that a high school student who wants to attend an elite college will be more likely to push to attend private school).

It didn’t control for extracurricular activities (it seems likely that private school parents will be more likely to encourage their kids to participate in extracurriculars that look good on an elite college application).

It did control for SAT scores through, so the one thing this study is definitely not doing is examining the effectiveness of public vs private schools at effectively educating children.

Re:What shall I teach my son? on Tuesday October 20, 2026 @09:58PM (#50175134)
>P(C | private elementary) = 5P(C | affluent public schools) = 35 P(C)

That can’t be true. A group’s probability cannot exceed the overall probability by more than the reciprocal of its population share. A 35× ratio therefore requires the group to represent at most 2.86% of the relevant population. For comparison, NCES reported that 9.3% of students in grades 1–5 attended private schools in 2016.

Re:Advanced coding in low-resource programming languages on Tuesday October 20, 2026 @07:22PM (#50174060)
SQLite at its core is a compiler, and a higher level one than C, so I don’t think that’s likely.
Re:Advanced coding in low-resource programming languages on Tuesday October 20, 2026 @03:14PM (#50171267)
Have they released this publicly? This seems about on the same order of difficulty to anthropic’s C compiler project, which ultimately failed. I’m surprised if they can do SQLite they wouldn’t show up Anthropic with a real working C compiler.
> a client that has been being developed for ~22 years, and I've rebuilt significant portions of it over the last 2 weeks.

Taking the end result of a 22 year old dev process and telling Claude to rewrite it, is in no way equivalent to doing 22 years worth of the work put in by the developers working on it.

That’s like saying replicating a paper that was the result of 22 years of research is the equivalent of 22 years of the author’s labor.

No Man’s Sky isn’t the equivalent of 10,000x the devs who made it. Because the procedurally generated content isn’t equivalent to handcrafted content. I’m also not aware of any procedural generation techniques in No Man’s Sky that weren’t possible 30 years ago.

As far as Tamriel rebuilt goes, if you’re talking just about just reimplementing Daggerfall in the Morrowind engine, you might be able to do that in a month of vibecoding.

But that’s not hundreds of dev years worth of work.

Older »