not_a_bot_4sho

(I came to delete but was too late. So edits are in.)

Wow, that's a stark take. I suppose I'm biased towards a scientific viewpoint. All the best.

If you're not doing *at least* say 100 iterations (thousands are preferred!!), you do not have enough data to draw any stable conclusions.

Interestingly enough, using an LLM-as-judge is a great way to approach things like this at scale but you do need to invest in some Cohen's Kappa or Fleiss' Kappa understanding which means putting a human in the driver seat to evaluate the effectiveness of your non-human judge. Absent of that, it's just another case of human-centipede but with LLMs.

I also think the classifier is way too crude

Developing a classifier can be a weird ordeal when you leave the ML world (of making it work) and enter the realm of policy makers applying it.

My guess is they're being ham fisted with little tolerance for false negatives. We can all guess why...

Yay, John!

I think we're mving towards humans no longer needing to understand a codebase, and letting AI drive it.

I can see this being true for non-critical software like entertainment, media, and so on.

Definitely not true for systems where security stakes are high. Like banking, aviation, defense, etc.. AI will surely contribute but not independent of human engineering understanding.

The Pirate Bay is considered one of the worst pirate sites now

I'm curious. What are the best sites nowadays? Still torrents, right?

I used to be in the know but switched to legit things a while back so I wouldn't have to explain how dad is getting Bluey episodes.

Living there with local wages and taxes?

I wouldn't call MacOS 'free' given it has very strict requirements for proprietary hardware.

Nice, this comment broke my composure.

Sounds like the viewers were highly unlikely to have clicked through. Cost the advertiser a view but lost the conversion.

Useful ad for Facebook. They made money on it. The advertiser didn't.

overpaid manager class

Ugh. Tell me you're early in career without telling me. Sophomoric take.

Thanks!!

I don't know the particulars, but if what's going on in WA is like what's going on in my California town, I don't think it's reasonable

I was driving on a WA highway going ~47 mph (I made note of it for reasons soon to be clear!) when a group of 3 tweens on ebikes, none wearing helmets, passed me up in the bike lane going at least 10mph faster, and I watched them run through a red light up ahead. They got lucky.

I don't have a problem with kids being kids but this is over the top dangerous. To be clear, it's a few small groups of kids. There have been crashes, and they sent a Bellevue woman to the hospital in a hit and run, but as far as I know there's been no fatalities yet.

We don't need to drop the hammer on kids. We just need some common sense.

Same.

But with the callout that SpaceX (via xAI) is trying really hard to be a giant.

Waiting for an AI startup to create a phononym of this, in the same vein as Google did...

Far too much of this bombardment is self inflicted and very much related to smartphone use.

I'm not sure how many times this has to be restated.

This strike me more as a repeated internet myth more than anything else. There is near endless opportunity for purpose-specific robot forms.

This isn't about children, and never was.

This is 100% about children.

At the moment, there is world wide pushback on children using social media. Ex countries pursuing this right now: Australia, United Kingdom, Canada, Indonesia, Malaysia, Brazil, France, Denmark, Norway, Spain, Italy, Greece, Austria, Poland, Slovenia, South Korea, and Thailand. I'm sure I missed a few too.

And in the US, no federal efforts (yay dysfunctional legislature!) but at the state level; Florida, California, Tennessee, Utah, New York, Nebraska, Texas, Arkansas, Louisiana, Georgia, Virginia, Mississippi, Colorado, Connecticut, Maryland, Minnesota, Montana, Vermont, and Ohio are all pursuing different forms of age restriction.

The science is pretty clear at this point: social media is terrible for mental health, attention, and addictive behaviors.

Now how to effectively restrict it without trampling on privacy rights? That's a very difficult question requiring some compromise. (I don't have the answers...)

"The Department of Defense is the country’s largest employer, with more than 2.1 million Military Service members and over 811 thousand civilian employees."

From https://comptroller.war.gov/Portals/45/Documents/afr/fy2024/...

I was going to say, isn't React something to hold against Meta? Being intimately familiar with it, I don't consider it a positive contribution to the world.

No one is stealing anything here.

Cursor users are willfully providing it by using their product. Not unlike uploading a personal photo or video to social media -- that's not yours anymore. You gave all rights away when you put it on their servers.

One of my grandparents had AD and became strangely lucid one day, a couple weeks before her death. It was like she finally woke from a years long fog.

I remember it vividly because I got to tell her I love her and she smiled and said it back to me. She seemed to understand me for the first time in years.

It was short lived but I'm very grateful for that single exchange it gave me.

I'm curious. What did you build? Sounds like you might have an interesting "Show HN" post on your hands

It's not impossible to comply with. It will just take additional effort to ensure compliance.

For example, they can put this burden on enterprise customers to verify and attest citizenship. This is commonly done today for some types of cleared work.

For consumers, I'm sure it can be done if the monetary incentive is there. People will hate it, but it can be done.

Assuming it was cleverly designed to be impossible to comply with is giving far too much credit.

Add a Trump son to the Anthropic board and all friction is gone

I added some brief commentary here: https://news.ycombinator.com/item?id=48507278#48511524 (or just refresh parent comment replies to see it)

It scratches the surface really but hopefully provides a helpful starting point.

Back with some papers. (Apologies in advance; I typically don't edit/format comments much here, please bear with me.)

Notable papers describing performance improvements with prescribed roles and personas:

- ExpertPrompting: Instructing Large Language Models to be Distinguished Experts (2023) https://arxiv.org/abs/2305.14688 (if you're going to only read one paper here, maybe read this one but know there has been a lot of follow up with more modern models.)

- Expert Personas Improve LLM Alignment but Damage Accuracy (2026) https://arxiv.org/abs/2603.18507

- When Does Persona Prompting Actually Help? (2026) https://arxiv.org/abs/2605.29420

- Unveiling Power on Combining Prompt Engineering Techniques: An Experimental Evaluation on Code Generation (2025) https://doi.org/10.5753/sbbd.2025.247251

- A Pattern Language for Persona-based Interactions with LLMs (2025) https://www.dre.vanderbilt.edu/~schmidt/PDF/Persona-Pattern-...

A TLDR of my *admittedly heavily biased* mental model (so take it with a grain of salt): personas do improve task alignment and precision to measurable effect but with observed negative impact to accuracy and knowledge grounding. Overall, this makes it quite suitable and preferred for code generation scenarios. (Don't over-index on 'accuracy' here as meaning "bad code", it's more about verbosity/jargon reducing clarity of higher order goals like business objectives and system architecture.)

Outside of code generation, personas have the interesting effect of increasing implicit biases and stereotypes. It's not hard to imagine something like "you are a left|right wing politician ..." or "you are a senior-citizen|teenager ..." influencing token space construction considerably.

Was that ever actually shown to be effective? Is it still?

Yes! Personas demonstrated measurable improvement in a few different ways, with caveats of course. The common intuition is that personas influence token space in beneficial ways.

I'll come back here later on desktop and link a few (still) relevant papers on this topic.

Interesting. My pixel experience has been the opposite; very polished and pleasant. 10A. Wonder why the variance?

I've seen plenty of crashes over the years. Primary in one private repo where gc and pruning triggered unexpected exits for a fixed duration of time.

That said, stability overall has been nothing short of fantastic. And I can't answer the question of "why?" for this particular rewrite.