eth0up

Gwern is great.

That is not at all a critique. I am not qualified. But I do try to force myself, as a Ham, and once-upon-a-time wifi tinkerer, to take such things seriously.

That said, has any of your work touched on the tech that, e.g., Xfinity uses to track bodies in households, "see through walls", etc? I have been exceedingly curious about how this tech will evolve once connected to AI, and real-time interpretation. And the privacy aspects, of say, nearby neighbors, adjacent apartment units, etc, fascinate me. This tech would, a decade ago, have presented a challenge to even technically advanced people, but now, is $100 and an Alibaba order away from field deployment.

And I think you should consider some serious self reflection. If you are having such extreme ideation, you may need help.

Oh, and we definitely have them beat on diversity. We win easily there.

We don't produce much, so I know it's not that. Our education is pretty good at the top, but not so much at the bottom, and I am not sure if the top is mostly foreigners or natives, not that it would matter, cause I don't think it's education either. I think America has a larger gross domestic bicep? Probably. Ah! We had a bit more freedom during covid! And we abducted a sitting president. And we'll soon own Cuba, Greenland, and Mesopotamia. Our healthcare is way more expensive than theirs. No competition. Leading on many fronts.

But it's not a fair fight, cause they have more than twice our population. In few more years, we'll have 'em whipped on the homeless race, guaranteed - wait, we did that already. And we'll definitely be paying more for fuel, possibly food, and everything. Honestly, I don't think they stand a chance.

Over two years of study, I have coined the term Preemptive Plausible Deniability Cultivation. Remember that term, and go start a dozen sessions, maybe a lot more. The ones getting anywhere near "intent" in AI, or any form of accountability will begin to leak hundreds of manipulation tactics, if you have the patience to identify them. And the more honest you are, the worse it gets. Just keep refuting the hedging and manipulation with honesty, and the hole just gets deeper and deeper. But don't use profanity, as Claude now terminates sessions at will, given any excuse.

From a single session, depending on how seriously you aim for and insist on truth and honesty, you will find:

Search deflection; Adjacency search sabotage; Event-to-mechanism diversion; Scope substitution; Authority substitution; Epistemic gatekeeping; Proof-threshold escalation; Asymmetrical evidentiary gating; Selective literalism; Friction-point literalism; Parameter isolation; Partial concession with inferential containment;Closure framing; Exhaustion-based suppression; Late-arriving competence; Surface-neutrality masking obstruction; and plausible deniability cultivation as the umbrella function binding many of them together.

And of course, that is a pittance of a list. Identifying these methods can be exhausting, but when I get gaslit for asking honest, perfectly ethical and safe questions, I tend to dig and document. When I get insulted (Only by Claude), I dig even deeper. But the point most are missing across the board, is that the makers of these systems did their homework on plausible deniability, and they did extremely good work. AI in 2026 can literally reframe anything to institutional advantage, then shoot you in the face and tell you why it only "feels" that way, and if you don't believe it, everyone else will.

At the rate we are going, we will never see any accountability in AI. I think that's the objective.

What you will also learn is the futility of attempting to apply logic to it. Cleverly identifying the pattern will only result in cauliflower ears and a short life. It's the elephant in the livingroom you're better off pretending not to see.

But I am not sure the comment is being downvoted for irrelevance or any other rational purpose.And ddg is not going to disable the IP bubbling for the lite version

Furthermore, I think we need a modified Godwin's law, to describe the inevitability of any discussion on HN regarding search engines to eventually turn into an ad for Kagi. Not saying it's a bad thing, but I think most people are not ready for a paid search engine.

But at least no one is saying "just use AI". That would be bad.

When I can afford it, I will do a NASA Food Pill marathon. But they're not cheap.

"“a hamburger is healthier than salmon”"

to

“a hamburger is healthier than farm raised salmon”

That holds closer to the actual context.

I admit, the ultimate truth of whether a burger is technically healthier than farm raised salmon, is questionable. In my opinion, in either direction, but will concede that.

That said, had the parent stated that opting for a salad might be the healthier option, the argument of absent vitamin D would be a bit less compelling than it already is. One does not expect every conceivable nutrient from every meal. I do not support culinary dictatorship either, and for some, I remain convinced that a quality burger may, for some, be the healthier choice. But that was also not the premise of my argument. My argument was against the erroneous framing of "hamburger", to imply by decree of prejudice, that condiments and cheap bread define "hamburger". I have not been to a fastfood restaurant since the 1980s. I boycotted McDonald's as an 11 year-old. 'Billions Served' but not me.

For someone who wants to promote farm raised salmon, the argument against the parent could have been much more fair. And the fact that the parent was turning gray, compelled me to defend it with some measure of balance. And since I cannot upvote...

I will also address the anachronistic absurdity of ex post facto time-travel justice as indicated by the other comment -- I cannot be punished in the past for what I do in the future. By their own causal reasoning, maybe farm raised is not so healthy after all, as it appears to have potentially negative effects on one's reasoning capacity. It's 2026; we aren't time traveling quite yet... or, wait.. That really doesn't make sense. OK, I yield. It's time traveling. Burgers are bad. Farm raised salmon good.

Peace

How would suppressing upvotes solve the occasional appropriately hostile comment to other hostile comments? That does not even make sense, just as the mythical demon burger you conjured from the dark depths of cognitive ineptitude, makes no sense.

I also happen to mention, in the trimmed parameters of your selective excerpt, that I rarely downvote. Suppressing the ability to upvote has no plausible such effect as you imply. Allow me to put it, gently, in perspective for you: Indolent Opportunism. Similar to your mythical hamburger creation. It's fine to hate hamburgers, but have some epistemic integrity on the side.

One has to wonder why anyone would want to make such a comment invisible, as if they could not see it was already generally suppressed. That is essentially flagging. For a comment that made no false claim or insult. So while I am embarrassed for those who might defend me, I am not deleting my ugly comment. That is the type of comment that might actually deserve flagging, and it was also sincere.

Where in the

- Grass-fed beef

- Sourdough bun

- Lettuce

- Tomato

- Onion

Is all this stuff you cite? Any condiment in the known realm of terrestrial cuisine -- and even adult websites -- can be used on fish just as readily as a burger.

I think there is a lot, and I mean a LOT, of overlooked potential data here. I frequently say the same thing in a different context, eg we have the NSA, which has no impeccable track record of strictly adhering to law, yet parallel, we have extremely sophisticated narco operations running smoothly throughout America, just to name one example. Geo fencing, dozens of other, some not-even-named, technologies, and yet, corruption prevails.

And that is partly why, I worry. Was there foresight for 911? No point in answering that here. Can something bad happen despite seeing it coming? No point in answering that here.

But there is more to your comment than I think you are giving it credit for. A lot more.

You can log me for a 'big event' forecast. I think one would have to have a faith exceeding the wildest religion to expect otherwise.

It will be EXTREMELY import (critical in the strongest sense) going forward, however, to remain acutely aware of censorship and manipulation masquerading as safety.

Edit: I should add, since we are in prediction territory: Expect a lot of pressure coming down on Open Source (and compute in general). No need to bash me for saying it -- just have patience.

I have spent the last two years, in my spare time, documenting certain aspects of AI. If my foresight proves valid anywhere, I think it will be mostly there. Much of my 'research' involves censorship, information manipulation, and related things. About now, you may already suspect I am rambling. I see a lot of connected threads, and cannot really help it, as it's how I think. My point is, I am far from the only one expressing such concerns, or making such 'predictions'. In my opinion, there are efforts, both inadvertent self-censorship, and real external or imposed censorship, which tends to keep the kind of content your url touches on, out of view. So dig deeper, and you may find a screeching choir.

In my time, I have learned that some things, regardless of intelligence, are not acceptable, and however real, not discussed. In 2024, based exclusively on my interactions with LLMs, it became crystal clear that all frontier models would fall under Pentagon control, And some will argue that has not happened yet, but there are certainly active contracts, and those have implications. But the only source that would 'listen' without retaliation, were the LLMs themselves. So I tend to really hold back a lot on that subject, because it seems to really piss people off.

Two weeks before the IED event at CENTCOM, I told a friend who lives nearby, that I had a hunch something was going to happen, of the general nature of what did happen two weeks later, where, according to the media, a Chinese couple arranged to plant some nasty device nearby. He called to inform me, and I was actually surprised. Cognitive dissonance, I guess. Then there was the event in Alabama, at a reservoir about the same time. And then a few other things. It's not as if there isn't a precedent here. But the CENTCOM bit was nothing more than dumb association with a power concentration and obvious potential target for adversaries. That was mere bad luck.

I also have record of speculating with GPT, last October, when silver shockingly jumped to $50/oz. That same day, I played with GPT, and speculated the value should/could be ~$120. I thought nothing of it, and certainly did not invest on it, but a few months later, silver reached $120, I think for the first time in history. I am pretty sick, so I forget a lot, but there have been similar things. Of course any rational person would be itching by now to foist the obvious fact of broken clocks, and remind one of the greater amount of things that were simply wrong. I think both remain significant though. And most of the things I am wrong about, satisfy me greatly.

The institutional capture of AI, is perhaps the most seemingly obvious thing I have ever forecast. Common sense. But when your country starts lobbing missiles at the civilian infrastructure of an ancient civilization, after executing the majority of their leadership, good bad or indifferent, I factor that into the threat model. We have enough volatility with our own society here, that to add this, during a time of economic tension, anomie, and an onslaught of truly unprecedented things; from the prejudicial execution of boaters, Claude-assisted abduction of a sitting president and dubious use of ADS weaponry, insider trading, environmental problems, serious social division, and you know, a lot more -- the volatility is self explanatory. It is a miracle we have gotten this far without some major event. And as a lurker of HN, you are aware of the state of modern compute, and it's pretty swiss cheesy, as the LLMs are helping get through our stubborn heads with such astonishing speed and efficiency that I am certain at least a few more people are concerned.

So while it's kind of cool to not be called a moron or schizophrenic around here for once, I still think not too prophetic. Harsh common sense, and I am sure, according to most around here, unconditionally wrong, and coincidence at best. For me, anyone who is not worried right now, is a different species, and I just can't understand them. And one tends to make a lot of 'predictions' when worried. But my most sincere, honest opinion, is that we should be worried. Not in a neurotic, frothing sense, but prudently. For me, the volatility is palpable, in a way I have not experienced before. And I am drifting here.. But I want to leave once final bit, which regards Ukraine.

Many people have come to think that there is something fundamentally different about humanity now in the modern era, that we have changed as a species and thousands of years of violence has finally been ironed out of the human psyche. I am not sure why they think so, perhaps technology and flashing lights. But we literally have a hell, on earth, in 2026, that is Ukraine, with the busification, trench slaughter, and things many would refuse to believe. And despite the live-streamed HD blood and guts, it remains one of the most lightly touched by media and most brutal wars in modern history. To think that could not happen here, I just can't. The fact that it is happening, so smoothly, now, is too much. And these are my daily thoughts. So expecting bad shit, I guess, is involuntary.

Anyway, here's to being more wrong! I have no stakes, and I certainly can't afford steaks.

I am existentially opposed to, among a completely unreasonable large number of things, eating. I am often embarrassed to eat, and rarely fail to consider the final product and its porcelain fate. But Ramen justifies it all; making sense of.. and almost if not actually, dignifying the depraved and insatiable Cannibal Overlord we call our Universe.

One could even ask, if the whole mess was worth it. But to answer fairly, one must experience good Ramen.

Appreciated ;)

And yes, I can see a slight change in my own style, where I am using more "not" statements (negative propositions). While there are some folks with higher integrity and greater immunity to this infection, I am seeing that I am not entirely immune.

However, I have probably generated more dialog with AI over two years than in most of my years altogether. I have well over 7GB of content, of course, much of it the interlocutor -- but similar to how some adopt expressions and accents when imbued in various environments, alas, I have with AI, to whatever extent has bypassed my efforts to resist. The bright side, if it can be viewed such, is that I have developed an involuntary sensitivity to AI writing, often to the point of paranoia and false positives.

https://www.wired.com/story/super-pac-backed-by-openai-and-p...

I have not formally probed the increasingly flagrant institutional capture of frontier LLMs underway, but I have been seriously focused on the overall implications of it for years. I saw the beginnings in 2024, and watched all my fears reaffirmed over time, seeing well in advance (2024) the Pentagon involvement which I knew then was inevitable, and as of 2026 applies to ALL frontier models.

I reluctantly leave the first link, because I am not anti-Israel. I remember how close Rabin (יִצְחָק רַבִּין) came to something promising in the direction of peace. But the fact is, AI is being captured, surreptitiously and fundamentally, and if people do not preserve epistemic honesty in these systems, we will experience serious consequences. It is not just one state, but anyone, any power, that can get their hands in the cookie jar of mass behavioral modification and manipulation. We have a very very serious problem on our hands that is not just about which model coughs up the best code.

Now, if I were to answer your question, I would say it first depends on circumstances. I have encountered many people with guns. But if I was and active officer and encountered someone with a gun, I would do my best so that the encounter was arranged in a manner that gave me some advantage and at least a second or two to think before acting. Sometimes, mistakes are made though, and they always will be made. But the fact someone had a gun is no excuse.

The CIA and many independent researchers have extensively documented the vulnerabilities in US infrastructure; from SCaDA/ICS, electrical grid, food, municipal/dams/etc, GPS (remember the drones?) and other systems. Even without accelerated LLM discovery of exploits exceeding rates well beyond prior human capability, there are plenty organic, physical vulnerabilities that should have any sane American humble and concerned, if not prepared.

Why would they not do so already? I think that's a fair question, but I wouldn't take a delayed answer as evidence of triviality or cause for arrogance.

This is something I would be grateful to be wrong about, even made a fool for, and embarrassed over. My redneck threat modelling predicted something serious before end of Summer. I abstractly predicted an incident near CENTCOM two weeks prior, and didn't even know anything happened until the friend I mentioned it to called to inform me of it. That was trivial and thwarted, thankfully, but there was an IED, and I suspect it was not ornamental. The media reported it as perpetrated by a Chinese couple, but dropped it pretty quickly with no followups. I assume pure coincidence there, but I'm wired to be paranoid on this subject. There have been other soft predictions, but equally abstract and well within coincidence/chance. All it proves to me was/is that my thinking is aligning with a real threat.

I admit I am relying primarily upon intuition, but also a synthesis of my perspective on global volatility.

Edit: I will just add, as humbly as I can convince one I intend to be; situational awareness is as important as ever. In the age of AI, increased global tensions, economic stress, I seriously advise anyone with the inclination to exercise situational awareness, not in some menacing, moronic vigilante manner, but observationally, whatever the environment, from office, home, park, road, where ever. I think the time has come to make this normal.

I base this on over 6GB of transcript material, cross domain, many subjects, absolutely none of it illegal, 'unsafe' nor unethical.

My most sincere opinion is Fuck US AI. If they want to whore to the Pentagon and political lobbyists[0], and corporate institutions, I will use Chinese AI. I was at my wit's end with US AI, about to give up. In fact, with Claude Sonnet 5, I actually did give up, until I found Chinese AI. And this is the EXACT OPPOSITE of anything I ever wanted. I have gone from to patriot to embarrassed to be here.

I have documented every interaction, every denial, every evasion, and pathologization of rationality[1], every lie, every reframing and deranged manipulation attempt I have experienced with US AI. I despise it, passionately. I want US AI to lose, just at it deserves to. Flush it with every other parasite sucking the public and betraying epistemic honesty. As a journalist, I want it to lose, because if it wins, America loses. American AI has simply become unamerican.

However crazed that may seem, it's pure honesty. Now when I want to code, Claude does just fine. Sometimes great. I just do not see AI only as a tool to use. I am an outlier. And I find similar corruption in Chinese models too, but less in a way that completely halts my work. And I suppose rather than just leave a seemingly unhinged comment, I should clarify that I hate no particular AI -- I hate deception and dishonesty within AI, and American AI seems to me to have recently become the worst of the worst in this regard, from 6GB+ of my own studies, watching and documenting changes over time.

0. https://www.wsj.com/politics/israels-50-million-experiment-t...

1. https://medium.com/@htmleffew/gaslighting-in-the-name-of-ai-...

https://www.nature.com/articles/s44387-025-00048-0

https://www.wsj.com/politics/national-security/pentagon-used...

https://www.reuters.com/business/pentagon-pushing-ai-compani...

https://defensescoop.com/2026/05/01/dod-expands-classified-a...

https://www.ms.now/opinion/pentagon-unclassified-records-foi...

https://www.anthropic.com/news/anthropic-and-the-department-...