dr_analog
top 1% of underdog fetishists
No bio...
User ID: 583
I was unaware of this article.
My attitude towards AI tools for the last 18 months had been "yeah they're useful but if you try to get too ambitious with them they waste more time than they save" and I was like AI 2027? Ha, try AI 2035.
But something fundamentally changed with the models in the last month. I'm low key freaking out at how goddamn useful they are now through Claude Code and OpenAI Codex.
Forget METR evals. My personal real world evals are that they're 6/6 on doing 2-4 week long tasks in 1-2 hours.
I'm just doing this on the $20 OpenAI Plus monthly plan. They haven't throttled me yet. Feels like year 2020 Uber rides across town for $4 stuff.
Originally, I had kind of given up on claude code a few months ago because it was wasting more time than it was saving me. It would lose the plot pretty quickly even though my instructions and the goal were still well within the context window. I considered this an architectural limit of LLMs.
But as of the last two weeks, holy shit. claude code (and codex), they just grind away at problems. They don't lose the plot. They back out and try different approaches. They run micro experiment to test assumptions. They'll run the tools with --help and --version and check the man page. They'll step through the code of the installed version of meson to see why the config file is not behaving the way it expects. I just give them like one simple prompt and it'll chug away for 15-30 minutes just trying shit like an overly caffeinated engineer. They'll run builds and look at errors and fix them until it's clean.
And again, they don't lose sight of the goal. It's amazing.
I'm not really convinced I need to be more than "experienced computer user with good taste" in the end. Probably not much more than a product manager.
I was mentioning io_uring or JITs etc as examples of extreme customizations the agents can do. But if you just tell it what you want and then complain "it's not fast enough, make it faster" it will, on its own, come up with and implement optimizations. The standard ones like better algorithms or pre-computing stuff, but they will get increasingly sophisticated and extreme if you keep saying "make it faster".
My specialized knowledge helps us not get painted into a corner as much, but does it matter when the paint dries instantly and it can repaint the entire house in minutes?
My biggest advantage for now may be that I can approximate in my head the theoretical limit for how fast something could be physically driven on the hardware, so I'll know when to stop saying "make it faster". I'll also know that when the coding agent says "I'm going to bake in hardware assumptions and weaken consistency models" that it might be worth stopping as well.
But this edge won't last for too long.
dr_analog
top 1% of underdog fetishists
1d ago
I normally use Claude Code but this week I've been using GPT5.2 with Codex on the $20 plus plan. They haven't throttled me yet, probably because they're blowing VC money like ZIRP is back in style.
It's so fast and IMO better than Claude and needs less direction. I'm going to be devastated when and if the party ends and I need to pay actual costs.
Well. I don't know what the rest of you do but every elite dev I know that had been skeptical and unimpressed by AI agents has been converted in the last month.
Android apps. Firmware for wearables. UNIX TUI clients. Code analysis tools. Web apps. Flight simulators. Gaussian splatters. One guy writing a functional formula language for a network message bus. 3D games for PCs. A Signal clone that doesn't require phone numbers. Bots to run trading strategies.
If you're in a niche where this hasn't happened to you yet, bless. It's probably better for your mental health to not cross this threshold.
I'm not vouching for Cursor or Copilot or the general chat experience. But Claude Code and OpenAI Codex, within the last two weeks, running the latest best models, are what are scaring the shit out of me. Before this I was mostly in "meh, loses coherence too fast, maybe in 5 years they'll figure it out" mode. No. It's here now.
dr_analog
top 1% of underdog fetishists
2d ago
.I had previously gathered the impression that AI coders were still at the level of an enthusiastic but sloppy apprentice.
Not anymore. They're basically as good as a senior software engineer now, except they finish 100x faster. And never need to rest.
At this point anyone not using them is resisting out of inertia. Or fear. Well placed fear, perhaps.
I am beginning to low key worry about how good the latest coding agents are, claude code and codex, within the last two weeks. I am routinely building apps in 2 hours now that might have taken me a month if I had to use my own brain.
They'll tailor make stuff to any level of customization or weirdness. Use io_uring? Do this part in x86-64 assembly? Want a JIT for you app? Want to see if we can make this algorithm run on your GPU? Sure it's all good.
If I spot a bug it will take less than five minutes to fix. They never give up.
The slowest part of the loop here is me. I can't test changes and describe features fast enough.
This latest app I've put down about 3500 lines of code and haven't looked at any of it. It may as well have been written by a total stranger in a language I don't understand, it just slows things down too much if I try to read the code.
I am guiding it using my skills and experience but it doesn't really matter. When it can rewrite the entire app in an hour there are not that many bad architectural decisions that can't be undone.
Obviously in a mature product with users and deployed infrastructure, radically changing your approach is harder. But even already it tries to steer me away from crazy stuff.
I'm sorry but software engineer grunts are cooked. If you can't design or product manage yourself, you're going to be unemployed. What does it mean to be a SWE in a world where software is built as fast as you can describe it?
dr_analog
top 1% of underdog fetishists
2d ago
OP is reasonably expecting survivor testimony and other physical evidence the FBI had access to substantiate, which seems in short supply.
dr_analog
top 1% of underdog fetishists
4d ago
But I would fault a person who had prolonged, extended, beneficial relationship with Epstein while fully knowing who he is.
That's the key here. How do you prove someone fully knew who he was? My contention is that the people who were curious about it were probably satisfied pretty easily. From my original comment:
A man with that much social approval could easily say, if anyone ever confronted him, "oh, that. yeah, it was a thing with an escort. it was consensual. she said she was over 18. it got blown up into something. I paid my dues. trying to move on" and be happily believed. Due diligence: done. Very few people with the liberal morality to [okay with people hiring escorts] wouldn't have bought that excuse
Absent seeing him mess with underage girls, or noticing a lot of underage girls in his company, it's probably not that legible.
dr_analog
top 1% of underdog fetishists
5d ago
Well, what's the actual difference between someone who marries you that thinks you're perfect versus someone that marries you because you check a lot of boxes but not all of them? The person who thinks you're perfect could still discover some other person that's better than you in some way and leave you for them?
dr_analog
top 1% of underdog fetishists
6d ago
I still have a hard time believing the average person would not accept an invite to a party at a rich guy's mansion if it had a few household names in attendance. And feel chuffed to bits if Epstein took a liking to them and wanted to introduce them to more people, and would easily look past his minor legal trouble.
97% is hyperbole but it would be high.
The other day, quantum computing expert(?) Scott Aaronson wrote about how he didn't meet Epstein and summed up in a comment something I had been thinking as well.
I had a further thought. Back in 2019, when Epstein became a central topic of conversation following his arrest and then death, and lots of my scientific colleagues were telling stories about their contacts or near-contacts with him, it struck me that there were zero stories about any scientist—liberal or conservative, male or female, morally naive or morally astute—saying, “no, of course I want nothing to do with you, because you’re friggin’ Jeffrey Epstein, the infamous mass rapist!”
So I concluded that, if anyone now imagines that they would’ve responded that way, it’s almost certainly pure hindsight bias. Indeed, even after Epstein’s first conviction, a short jail stint in one’s past for “soliciting prostitution” simply doesn’t sound disqualifying, according to the secular liberal morality that most academics hold, unless you researched the details, which most didn’t.
All of the pearl clutching about how powerful men (and women) who associated with Epstein must have clearly known what he was about and what he was up to as a convicted pedophile ephebophile, when it's almost certain that 97% of the population would have gleefully accepted an invitation to one of his parties filled with leading scientists from MIT and Harvard, heads of state, CEOs, inventors, billionaires, and the rest of the somebodys.
A man with that much social approval could easily say, if anyone ever confronted him, "oh, that. yeah, it was a thing with an escort. it was consensual. she said she was over 18. it got blown up into something. I paid my dues. trying to move on" and be happily believed. Due diligence: done. Very few people with the liberal morality to be condemning him for hiring an escort wouldn't have bought that excuse and gone back to dreaming of rubbing shoulders with the who's who and maybe getting a sweet private jet ride. "Can he really be such a bad guy if all of these other great people are hanging out with him?", thought all of the other great people hanging out with him.
dr_analog
top 1% of underdog fetishists
6d ago
By "sex was hot af" I don't mean "the person was more attractive", but rather "sex with crazy/sociopathic people is exhilerating".
dr_analog
top 1% of underdog fetishists
6d ago
Do men not have the same thing?
Men also have that borderline girlfriend, you know, the one that's an artist, where the sex was hot af but the relationship was otherwise tumultuous and they had to stop seeing them because they kept getting fall down intoxicated in public while out by themselves. Eventually they "settle" for the girl who doesn't have the mental illness and drug abuse issues but it's true they don't like to dirtytalk as much.
One of my great fears is being caught in a struggle session like this. I'm not really kidding. I used to have nightmares about it as a kid, when I didn't even know what a struggle session was. I would dream that there was a huge crowd around me and they were angry at me, and someone that was my friend would be interrogating me with inane questions that seemed to have no right or wrong answer, but no matter what I answered the friend would announce it to the crowd with disgust and I'd be booed and jeered. Just complete lunacy.
I put myself in the yoga studio manager's position and wonder what I would say. The troll in me would lean into it: "do you stupid motherfuckers think we took the sign down because we're trying to welcome ICE agents to do yoga? do you think they do yoga?" and "so, what I'm hearing is you're mad that we had the anti-ICE sign up but had to take it down? how mad are you at yoga studios that never had the sign up to begin with?"
But I realize the right move would be to say "actually, " lowering voice " one of our dark skinned owners is anti-ICE, but she's afraid her green card will be revoked if ICE finds out she had that sign up. Please be her voice out there, because she cannot speak for herself, here" and then we all :resist-fist:
dr_analog
top 1% of underdog fetishists
7d ago
I think all those facts might be related. If you're old enough and wealthy enough that you don't have to work anymore, it can be a shock to your sense of identity. What do you do with those extra 2,000 hours a year?
I can just imagine society finally becoming wealthy enough by 2035 that we could institute a sweet UBI for all, and then by 2037 we have the most brutal civil war because we believe we're the most oppressed people ever.
Prediction markets probably have no good things. The author of The Laws of Trading wrote about why prediction markets can't deliver what they promise. A bit technical on the trading side but should be navigable by laypersons.
In 1964 there were 458,000 measles cases, and 421 deaths, over a smaller population, no lockdowns.
How barbaric. Our ancestors were truly uncivilized.
Isn't it, compared to influenza, 10x as infectious, with 10x the hospitalization rate and 5-10x as deadly?
If we had a vaccine that reliably stopped influenza (instead of the bullshit yearly one people try taking which misses 75% of the time) I can't imagine why we wouldn't all be on it? But the measles vaccine is a lot more reliable than the influenza vaccine? And you don't have to take it annually?
It seems like a tragedy that our society is rejecting the measles vaccine. What am I missing?
dr_analog
top 1% of underdog fetishists
8d ago
I mean, 99.9% chance none of that happened so I'm okay with saying if your state is that competent and brazen you deserve to dominate us.
dr_analog
top 1% of underdog fetishists
9d ago
I'm still skeptical, as it ascribes to netanyahu/idf intelligence a level of power that seems unreasonable on multiple layers, but it would be the very, very funniest writing of this season if the thing that gave away his cover was a public (now-private) fortnite profile, of all things.
If the Israelis busted Epstein out of jail and left a body double and he's just been hiding out in Israel this whole time they deserve to rule the world.
dr_analog
top 1% of underdog fetishists
13d ago
It's quite a mindfuck to have AI produce like 700 lines of code that work on the first try, which would cost like $500 to have a human write, but if you ask it to write a 1000 word essay it goes by in a blur that seems okay at first but if you read it with your brain on it sounds bad and wrong.
In truth if you look at the code more closely you could make it much prettier and more sensible but it all kind of works well enough that you don't care. Whereas it's really hard to make a written essay acceptable.
dr_analog
top 1% of underdog fetishists
13d ago
Someone recently claimed that people here would greatly outperform the market given their higher-than-average intelligence.
IMO the way to beat the market with your higher than average intelligence is to either bust your ass and build something great that people need or join a quant fund and work in a mentorship with other traders and use their capital and proprietary technology stack. Everything else is just spinning a roulette wheel.
dr_analog
top 1% of underdog fetishists
13d ago
ie “if you’re so smart, why aren’t you rich?” was a favorite discussion topic both on the SSC open thread and on this forum in the early days, and still comes up occasionally.
The problem with trying to profit off of being smart is that
(a) Prices are set by the marginal trader, not by the average. There's a lot of stupid people pushing prices every which-way. It's true that dumb people create what could be opportunity, but other dumb people also ruin the opportunity
(b) Related but different, is there often isn't enough liquidity to exploit your information.
Sort of tangentially, something that comes up a lot of the time in like, any real business I've looked into, is that local competition is crazy and desperate enough to disregard regulation. They get away with it most of the time, but I'm just not comfortable risking six figure fines (or worse) just to make a go at house flipping.
I can't really imagine all of those men thought they were knowingly raping this guy's wife against her will. I have to believe he told them it's her kink to be drugged up and raped and then she wants to watch the videos later. Or something. This is a lame excuse but I can see at least a few pathetic horn dogs falling for it.
Now let me see if any of the men offered any defense.
Update: I'm back. A few of the men said they thought it was a role play situation and that they were led to believe this by her husband. But most of them sound utterly indefensible. So, I underappreciated the horror by quite a bit.
There's a full table here with notes about each man and their defenses, if available.
https://en.wikipedia.org/wiki/Pelicot_rape_case#Convicted
They kind of sound like the dregs of society to me.
But yes, what is there to say? You can find dirt bags to do sex crime if you look long enough. The were all publicly shamed and convicted. The right thing happened, at least. My update is France takes these crimes more seriously than I would have expected. Also the right thing.
I'm slightly reassured it was only 72 men over 9 years and not, like, a thousand. Suggests the candidate pool is not a vast ocean.
More options
Context Copy link