>voting for policy positions that obviously work against those voters' own interests
Delusions of grandeur / superiority complex much? Have you considered a posibility, that these voters just dont agree with your idea of what their best interests are?
>fictional terrorist organization called "antifa."
An organization so fictional, that just a few months back multiple members were convicted in a terrorist attack
“Organization” is a stretch. Criminals? Sure. Violent thugs? Maybe.
The “terrorist” word, however, has been stretched and distorted for political ends. It seems like any crime motivated by political or religious ideology is now “terrorism”, and thus subject to much harsher punishment than those same crimes would normally receive.
By this definition, the Jan 6 rioters were terrorists - yet they received pardons.
The UK's position on Palestine Action is the poster child for this. Completely harmless old people holding placards have been prosecuted under terrorism legislation passed after 9/11. "Terrorist" has become a label applied to anyone opposing any government policy.
We need to roll back all those counter-terrorist laws in every Anglosphere country.
If people could imagine it being used against them, maybe they might understand. Like labeling the Republican party a terrorist organization because they sponsored Jan 6 insurrectionists.
> Antifa defenders just hide behind the fact that the movement is decentralized and leaderless by design
Calling a movement an organization is a fiction, unless you live in an alternate reality. Should i consider you part of an organization because you tweeted a certain hashtag once? Are you the member of a band because you whistled a song on the subway? Are you in the military because you watched a Rambo movie when you were a teen?
For a group who's so bloody-minded about "words have meanings" about certain things, this whole "movement and affiliation and organization are all obviously exactly the same thing" claim is blatantly ridiculous.
When Republicans in the US communicate and sometimes coordinate with totalitarian international leaders, does it mean we get to label the whole thing as a single "international totalitarian organization"?
> Have you considered a posibility, that these voters just dont agree with your idea of what their best interests are?
I mean, there's not really room for debate on it. Voting to take your money away from you and give it to people who already have more money than they could spend in a lifetime is obviously against your interests. But that's what Republican policy does. That's why they have to make increasingly outlandish fictions and fund whole media empires to sell those fictions: to convince people to keep voting against themselves. Between either a billionaire, or the guy down the street making burritos for $8 an hour, it's obvious which one took all of your money. It's the guy with the money. The increasingly elaborate alternate reality exists to hide this from their voters, so they can keep raiding us.
- "artificial intelligence has three quarters to prove itself before the apocalypse comes" — Mar 2024
- "If OpenAI doesn’t either reduce their $8.5bn operating costs to $1bn or less and raise at least $5bn in the next year, they will die." — Jul 2024 2025 costs: $34B
- Generative AI “isn’t getting much more efficient” (Jul 2024 ). OpenAI’s frontier-model API price fell from $30/$60 per million input/output tokens for GPT-4 to $4/$20 for GPT-5.6 Sol, alongside major capability gains.
- “I would be shocked if [Musk’s] wealth doesn’t return to something more like he had in 2019 or 2020” (Dec 2022 ). Musk is now worth approximately $873 billion , several times his wealth when Zitron wrote this.
The actual quote and topic is far different than you’ve portrayed it. Here’s the actual quote: “ yet the company says that it expects to make $11.6 billion in 2025 and *$100 billion by 2029*, a statement so egregious that I am surprised it's not some kind of financial crime to say it out loud.” from https://www.wheresyoured.at/oai-business/
He goes on to specifically discuss how unrealistic $100B by 2029 is given that OAI is structurally unprofitable.
The fact that you’re skewing your misinterpretation of what he said so much shows your own bias I’m afraid.
This is another case of missing the forest for the trees. In the scope of what Zitron is talking about, the difference of >$2 billion of revenue doesn't make a difference of his overall point of it not being close to enough to cover their spending.
I follow Zitron but the way he talks has been grating and it is even more obvious how biased he is when he has guests on. As if he's trying to lead them into agreeing with his more extreme claims.
That being said, why does this blogger get a pass at this criticism of just being blanket "wrong" when the point of the quote is still very much valid in context? It seems to be the same in the next article, quoted from Ed's blog with the same point. Is this more evidence of being "wrong" or did he correct a previously wrong value and came to the same conclusion (which seems reasonable to me in context)?
It seems like this blogger is guilty of the same criticisms he has against Zitron. They seem eager to find where Zitron is wrong, exaggerating the value of when he misses the mark and without looking at big picture. Then they make sensationalist claims based on those findings.
I think it's fair to call his specific predictions "early." I'm saying his timelines are too short. Like almost all bearish market commentators, he doesn't understand the scope of the government stimulus. This is a systemic problem in market commentary. But, don't kid yourself. If the US stops massive deficit spending, a lot of what Zitron's saying moves forward on the timeline.
A prediction definitionally includes an outcome, and it _can_ include a timeline. (if you’re serious about your forecasting, then you generally include a timeline; forecasting is non-actionable if it does not include a timeline).
You might choose a different timeline for Ed’s outcomes. Okay: those are now your predictions, not his.
Zitron has done us the favor of including timelines with his predictions, so that Dan can invalidate almost all of them.
It's not government stimulus driving this. It's that when you start with a "successful" company, the financial consequences of bad decisions or bad leadership can take a long time to have a negative effect, if at all, because financial success has a momentum to it.
I left Microsoft during the windows 8 cycle in large part because I could tell nobody knew what the hell they were doing, which is an objectively true statement about the time and place. My dad happened to buy Microsoft stock at that same time and did very well with it.
That's the paradoxical problem that I think all big tech has. Poor decisions by incompetent people, met with inexplicable financial success.
Look at GDP growth (it's not great). Where are the sales coming from? The modal revenue boost to Mag-7 companies since Covid has been a profitless tech firm wasting venture money whose source was debt enabled by low interest rates.
The US can safely reduce its deficit spending by $1 trillion and absolutely nothing big will happen.
The US typically adds that much annual GDP every 16-24 months at this point. The notion that somehow the gigantic ~$31 trillion economy will fall apart if the outsized deficits don't continue, is very absurd.
The exact same things were said of the Bush deficits. The US economy was supposedly dead in 2009-2010. Here in 2026 the economy is 50% larger inflation adjusted and it has left most of Europe in the dust. The housing bubble contagion was much worse than anything we're sitting on now with AI spend.
Why do I say $1 trillion instead of $2 trillion? There are plausible scenarios where increased taxes bring down the deficits (the Dems will take the House + Senate + Presidency, we'll see how much taxes go up), there's no plausible scenario where the deficits go away completely.
I disagree. The massive divergence between US household income and European household income since 2009 tracks almost perfectly Fed interest-rate policy and government deficit spending.
The federal government has spent the time since 2009 lighting the furniture and the house on fire to support unsustainable increases in domestic standard of living by effectively mortgaging the future. The problem we have dwarfs anything that was happening during the Bush era.
Likely the difference between you and me is this: I'm 50 years old, already wealthy from tech, and planning to leave the US. You're probably still trying to earn your way. Sadly, I'm pretty sure we've pulled the ladder up, and folks like you are going to reap the whirlwind, as they say.
Would you say the 2012 apocalypse doomsayers were correct but just "early"? After all the sun exploding in a billion years will ultimately prove them correct.
> OpenAI’s frontier-model API price fell from $30/$60 per million input/output tokens for GPT-4 to $4/$20 for GPT-5.6 Sol, alongside major capability gains.
on this specific point: tokens aren't fungible between models right? Like the argument Zitron makes is that improvements in output are due to, glibly, using more tokens to get there. We see people turn on new models and instantly use up all their tokens.
Like the actual measure is more something like "for this specific task, did it cost less now to do it than it did 3 years ago with these AI pipelines" right? The token pricing isn't actually relevant in that discussion.
If you're saying that OpenAI forecast something and you're saying it's validation that the numbers match the forecast, I don't think you're really paying attention to the problems (with annualised run rate bullshit when these are proper companies who could be publishing proper numbers), with unclear financials that are designed to look good, etc.
> There will always be someone willing to build that thing
… with sufficient time and money. As opposed to the minimal time and money required if the baddies are just pulling an existing solution off the shelf.
I hope you are joking, or haven't thought this through! If you sincerely believe and act on what you're saying here, you should be shunned from polite society.
Well, there are quite a few frames to think about. What about the unintended consequences of building something for a worldview at the time? Then the next generation of stewards take over, do not have the same understanding of the problem the system solved and different ideas on how to use the system. That's definitely something worth writing about. Stewardship in software is a real and common aspect of the greater system we all live in, and has unknown consequences with each generation that takes over. When you step in as a gen-1 builder, you should probably consider what it is you are about to build.
Just because someone is willing to build it doesn't mean it has to be you.
We may all play some part in societies that have some degree of monstrous outcomes, but I feel like staying on the "tragic civilian casualty" side versus the "legitimate target" side of a freedom fighter's bombing campaign is probably the way to go.
It's not about someone willing or not, it's about you, an ostensibly grown adult who should think in terms of your own responsibilities instead of others'.
Terrible argument. You can say ”why not work as a concentration camp guard? There will always be someone willing to do it so you might as well cash in”
CLAUDE.md is mostly powerless against the reinforcement learned crap. I'm up to three separate instructions telling it to cut out the hyper verbose, retelling history comments and it still writes them every time.
The best trick I have after asking it nicely in all sort of ways is:
1. Have it build a scoring script that penalizes words outside a simple English list and approved jargon. Penalize sentences over 15 words as well. Add whatever else.
2. Run it in a loop to reduce the score while preserving intention
This works much better than other ways I’ve tried. Of course it costs more. And I would apply it only to the output to the user, not the thinking process (I think the AI thinks better with their crazy English)
Of course, sometimes nuance is lost by this process. That’s just the nature of making things simpler.
I haven't tried this with a score but I have a simple skill with some examples of PR description changes and good PR descriptions I'd previously wrote and I just run it on the description.
It does cost more but I haven't tried cheaper models to see if they can get the same results. Curious if anyone else has.
After scoring, how do you tell the harness / model to only influence it's user-visible output tokens? Is there a deterministic way to specify this or is it a plain-text instruction in a hook or skill?
> CLAUDE.md is mostly powerless against the reinforcement learned crap.
When you dont know the cause, you dont have a fix. Thats the biggest issue i have with all of AI is that we dont know how it works, and yet we think it will be great ! This is more like a religious belief than a scientific one. There is no causal model of how it works, there is no theory. And the temerity to call it intelligence is annoying.
Where is the causal model of how the human brain works (on the level you're requesting)? If a causal model is needed before calling it intelligence, then humans are not intelligent.
You could probably keep the Claude slop hidden and have a fresh model generate a paraphrased response for anything human visible and keep the Claude responses as thinking.
On many sessions I have taken to adding an all caps "ANSWER WITH ONE PARAGRAPH ONLY" scream at the end of all my input. It's the only thing that gets results.
I suspect the root problem is these issues aren't at the system prompt level, they're in the RHLF/fine-tune. And due to safety/jailbreaking fears, all prompt content and user-instructions are nerfed in priority.
You would hope? Really really hope? that they could observe this, and target it?
Like, Claude going off the rails isn't something that takes a lot of effort to demonstrate. Literally anybody with a CLAUDE.md has seen the behavior over and over and over.
Hey Ants, can you maybe just not release the next version, no matter how good it seems on benchmarks, if it can't follow the goddamn instructions? Please? This seems trivial to test for and yet here we are, being gaslit by lying machines who intentionally do not do the requested work over and over and over and over.
I fully and completely expect a mental health crisis among developers. Being lied to constantly cannot be good for us.
Constant vigilance! is how you get developer PTSD and inability to believe anything you're told. Add the stress of parsing through yet another hyperverbose paragraph of bullshit while having your job threatened? People are not gonna end up in a good place, and this is as inevitable as sunrise.
I dont know what ASD-STE100 is before but I use the exact instruction (without the ASD code) to Claude since the very beginning, and with Opus 5 I have to remind it very often to rephrase the documents
Maybe compute is the real moat (chinese possibly skip around it with distillation), xai is buildouts have been insanely fast (colossus 1 - 100,000 H100 GPUs brought online in 122 days lol) so maybe that explains them catching up
asked grok to give a compute estimate for each:
- SpaceX / xAI: ~1.4 GW (owned Colossus clusters)
- OpenAI: ~2–3 GW (mostly rented/cloud)
- Anthropic: ~1.5–2.5 GW (multi-cloud + xAI lease)
this is a level of hippie delusion i wasnt aware existed unironically
china is never slowing down, therefore the us shouldnt either
reply