nosuchthing

This paper is an incredible read: TESCREAL hallucinations: Psychedelic and AI hype as inequality engines

https://akjournals.com/view/journals/2054/7/S1/article-p22.x...

  > "Researchers have called attention to the ways that the hype promoting psychedelics as miracle cures 
  replicates preceding claims about the efficacy of SSRIs and other antidepressants in prior decades. 
  As the drug historian David Herzberg articulated in conversation with UC Berkeley's The Microdose:

    There’s been an enormous amount of money invested in psychedelics as people hope that they 
    can be the real Prozac in the same way that Prozac hoped it would be the real Valium and 
    Valium would be the real barbiturates, which would be the real morphine. 
    There’s a long history of hoping that maybe this time, it’s not so complicated; 
    maybe there is a simple switch to change people without having to change any [other] aspect of their [lives].

  While others have noted similarities between the earlier SSRI hype and the ongoing hype for psychedelic medications,
   the rhetoric of psychedelic hype is tinged with utopian and magico-religious aspirations that have no parallel 
   in the discourse surrounding SSRIs or other antidepressants. I argue that this utopian discourse provides insight 
   into the ways that global financial and tech elites are instrumentalizing psychedelics as one tool 
   in a broader world-building project that justifies increasing material inequality. 
   This elite project reveals how medicalized psychedelics can potentially undermine the very prosocial and 
   pro-environmental outcomes that the field's funders insist psychedelics will promote. 
   To understand the envisioned role of psychedelics within this elite project, this paper analyzes a different 
   parallel hype, revealing correspondences between the psychedelic industry hype and the concurrent 
   hype surrounding artificial intelligence (AI), including the Large Language Models (LLMs) that power ChatGPT. 
   The presence of these parallels is understandable when one considers their underlying affinities, 
   like two blooms from one plant: the same Silicon Valley and venture capital forces are investing 
   enormous amounts of capital to develop both as cultivars in their own image, 
   selecting for desired traits that further the existing socioeconomic order.

LLMs can't access the training data that's less than the statistically most common token, so they use a random jitter.

With that randomness comes statistically irrelevant results.

It's a type of cognitive bias not much different than an addict or indoctrinated cult follower. A subset of them might actually genuinely fear Roko's basilisk the exact same way colonial religion leveraged the fear of eternal damnation in hell as a reason to be subservient to the church leaders.

hyperstitions from TESCREAL https://www.dair-institute.org/tescreal/

It looks like most of Peter's projects are just simple API wrappers.

Peter's been running agents overnight 24/7 for almost a year using free tokens from his influencer payments to promote AI startups and multiple subscription accounts.

  Hi, my name is Peter and I’m a Claudoholic. I’m addicted to agentic engineering. And sometimes I just vibe-code. ...  I currently have 4 OpenAI subs and 1 Anthropic sub, so my overall costs are around 1k/month for basically unlimited tokens. If I’d use API calls, that’d cost my around 10x more. Don’t nail me on this math, I used some token counting tools like ccusage and it’s all somewhat imprecise, but even if it’s just 5x it’s a damn good deal.

  ...  Sometimes [GPT-5-Codex] refactors for half an hour and then panics and reverts everything, and you need to re-run and soothen it like a child to tell it that it has enough time. Sometimes it forgets that it can do bash commands and it requires some encouragement. Sometimes it replies in russian or korean. Sometimes the monster slips and sends raw thinking to bash.

Peter has been running agents overnight for months using free tokens from his influencer payments to promote AI startups and multiple subscription accounts:

  Hi, my name is Peter and I’m a Claudoholic. I’m addicted to agentic engineering. And sometimes I just vibe-code. ...  I currently have 4 OpenAI subs and 1 Anthropic sub, so my overall costs are around 1k/month for basically unlimited tokens. If I’d use API calls, that’d cost my around 10x more. Don’t nail me on this math, I used some token counting tools like ccusage and it’s all somewhat imprecise, but even if it’s just 5x it’s a damn good deal.

  ...  Sometimes [GPT-5-Codex] refactors for half an hour and then panics and reverts everything, and you need to re-run and soothen it like a child to tell it that it has enough time. Sometimes it forgets that it can do bash commands and it requires some encouragement. Sometimes it replies in russian or korean. Sometimes the monster slips and sends raw thinking to bash.

https://github.com/steipete/steipete.me/commit/725a3cb372bc2...

  OpenAI has deleted the word 'safely' from its mission (November 2025)

https://theconversation.com/openai-has-deleted-the-word-safe...

Thread: https://news.ycombinator.com/item?id=47008560

Other words removed:

   responsibly
   unconstrained
   safe
   positive

either way there's a lot of money pumping the agentic hype train with not much to show for it other than Peter's blog edit history showing he's a paid influencer and even the little obscure AI startups are trying to pay ( https://github.com/steipete/steipete.me/commit/725a3cb372bc2... ) for these sorts of promotional pump and dump style marketing efforts on social media.

In Peter's blog he mentions paying upwards of $1000's a month in subscription fees to run agentic tasks non-stop for months and it seems like no real software is coming out of it aside from pretty basic web gui interfaces for API plugins. is that what people are genuinely excited about?

Can you elaborate on this more or point a link for some context?

That's not actually research though. The LLM API is only requesting a few of the top search results from a search engine and then adding those web pages to its context window.

That might work for simple tasks, but it's easily susceptible to prompt injection attacks and there's no way to validate the quality if it's statistically novel enough and outside of the core training data.

"Is curing patients a sustainable busines model?" - Goldman 2018

https://www.investmentwatchblog.com/goldman-sachs-asks-in-bi...

Many of the biggest medical innovations have come from publicly funded university researchers, which then license or give away their findings to private businesses.

LTO9 is only 18TB.

The LTO compression ratio is theoretical and most peoples data will be incompatible with native LTO compression method used.

"The Dawn of Everything" by David Graeber is a great, more recent alternative with a lot more context around the non-linear trajectory of history - the modern myths of linear progressive societal progress from savages, to agriculture, to cities and centralized technological futurism.

Graeber also explores the question what defines a society, and how at certain points some groups of people identified their culture through "schismogenesis" more so in oppositional context to against other group(s)

It's a massive book, but really refreshing and full of delightful little anecdotes and footnotes all through out.

Manipulating the The Keynesian Beauty Contest is the metagame, especially after the era of /r/Wall-street bets swelled up. What do you think Jim Kramer gets paid to do, report facts?

By 2017 teenagers were running quantitative trading algorithms and semantic analysis trading bots they found from influencers and github. Around that time private subscription fee telegram and discord "whale" insider trading groups often had multiple levels of insider trading market manipulation running pump and dump schemes. Marks would pay to join the private chat group and at a certain time of day a ticket symbol would be released for "insiders" to buy before it's promoted in public channels. The outer circle was just providing liquidity for the deeper insiders to exit with easy profit.

Most VC gambits are not much different. Sell the news. Unleash the rumors.

In a post-industrial economy there are no more economic problems, only liabilities. Surplus is felt as threat, especially when it's surplus human labor.

In today's economy disease and prison camps are increasingly profitable.

How do you think the investor portfolios that hold stocks in deathcare and privatized prison labor camps can further Accelerate their returns?

upwards of 60% of the ethanol used in combustion engines is completely wasted.

Mass layoffs and the wellness camp industry will easily account for $600+ billion a year in contracts, for at least a few years.

https://www.teenvogue.com/story/rfk-wellness-farms-us-disabi...

Slop influencers like Peter Steinberger get paid to promote AI vibe coding startups and the agentic token burning hype. Ironically they're so deep into the impulsivity of it all that they can't even hide it. The latest frontier models all continue to suffer from hallucinations and slop at scale.

  - Factory, unconvinced. Their marketing videos are just too cringe, and any company that tries to get my attentions with free tokens in my DMs reduce my respect for them. If you're that good, you don't need to convince me by giving me free stuff. Additionally, some posts on Twitter about it have this paid influencer smell. If you use claude code tho, you'll feel right at home with the [signature flicker](https://x.com/badlogicgames/status/1977103325192667323).


  + Factory, unconvinced. Their videos are a bit cringe, I do hear good things in my timeline about it tho, even if images aren't supported (yet) and they have the [signature flicker](https://x.com/badlogicgames/status/1977103325192667323).

https://github.com/steipete/steipete.me/commit/725a3cb372bc2...

New York City had hybrid in-person and remote as an option, so only about 4 months were remote for most students. Seems like a perfect A/B test case to test the theory that a remote learning was actually associated with cognitive decline 5 years later. Meanwhile every year since 2020 there have been studies published from nearly every country on the neurological and cognitive damage from mild covid cases. Seems to be more of a cognitive dissonance issue, that people don't want to confront the elephant in the room. 100 years ago when TB and Polio were taken more seriously than today, the common solution was fresh air. Schools today are even more sealed off to the point where a simple lack of oxygen starts to noticeably reduce cognitive ability after a few minutes.

Funding for schools is often based on Average Daily Attendance.

Most studies are showing children with neurological damage from continued exposure to covid outbreaks in schools multiple times a year.

Lockdowns did not last more than a few months for the vast majority of school districts in the US.

Most schools across the US were available for in person teaching by April and June of 2020. By August of 2020 the vast majority of schools were opened back up.

https://ballotpedia.org/School_responses_to_the_coronavirus_...

The science for what's actually causing cognitive decline is linked more to the neurological damage from poor ventilation and lack of hygienic conditions in the schools causing kids to get sick multiple times a year, directly causing neurological damage.

  In the most expansive study of its kind, researchers have for the first time shown serious and prevalent symptoms of long COVID in kids and teens. The August study, published in the Journal of the American Medical Association, is among the first large comprehensive studies of the disorder in this age group. The study, which followed 5367 children, found that 20% of kids (ages 6-11) and 14% of teens met researchers' threshold for long COVID.

  We know that COVID harms the brain. Neuroinflammation, brain shrinkage, disruption of the blood-brain barrier and more have been documented in adults, as have cognitive deficits. These deficits have been measured as equivalent to persistent decreased IQ scores, even for mild and resolved infections. Millions of people have, or have experienced, “brain fog.” What, then, do we guess a child’s COVID-induced “trouble with focusing or memory” might be?

  When you put together the estimate that 10 to 20 percent of infected kids may experience long-term symptoms, that many of the most common symptoms affect cognition, energy levels and behavior, and that children are being periodically reinfected, you have a scientific rationale to partly explain children’s widely reported behavioural and learning challenges.

https://www.scientificamerican.com/article/long-covid-is-har...

https://jamanetwork.com/journals/jama/article-abstract/28227...

https://www.nature.com/articles/s41598-025-96191-4

https://www.cdc.gov/mmwr/volumes/71/wr/mm7131a3.htm?s_cid=mm...

https://theconversation.com/long-covid-puzzle-pieces-are-fal...

https://www.washingtonpost.com/education/2024/07/23/covid-te...

https://jamanetwork.com/journals/jamapediatrics/fullarticle/...

https://www-news--medical-net.cdn.ampproject.org/c/s/www.new...

Debunked and no evidence, all claims came from biased sources: https://www.pbs.org/newshour/world/how-2-debunked-accounts-o...

https://en.wikipedia.org/wiki/Sexual_and_gender-based_violen...

Hannibal directive: https://www.haaretz.com/israel-news/2024-07-07/ty-article-ma...

https://en.wikipedia.org/wiki/Hannibal_Directive#Claimed_use...

Isn't ATProto just a compromised version of Activity Pub, basically designed around an excuse to force all users into a data mining firehose structure like twitter used to have only there's no privacy features or federation for moderation controls?

The game is social engineering to "exit" for early investors to bag holders through an IPO just like an ICO. The irony is with crypto everyone knows it's a scam, but with IPO's there's still true believers who think the business operations will be sustained. Uber is the case study. They've just hid their financials in shell companies that assume a future monopoly of the entire planet.

https://boingboing.net/2019/05/08/forever-unprofitable.html

Metadata is never deleted, only copied to more and more archives and databases.

You're always being added to a list, and all of your metadata decides what lists you'll end up on in the future.

Thanks. That's not really bypassing iPhone before first unlock. It says only 'partial' metadata, so it's likely just looking at encrypted blobs and making guesses just like file recovery tool would on an encrypted drive. So it's a bit of a marketing gimmick to "leak" that document

  > The document does not list what exact types of data are included in a “partial” retrieval and Magnet declined to comment on what data is included in one. In 2018, Forbes reported that a partial extraction can only draw out unencrypted files and some metadata, including file sizes and folder structures.

Where has there been any allegations iPhone before first unlock has been bypassed?

GrapheneOS isn't quite as secure in the real world. Pixels continue to have baseband and OOBConfig exploits that allow pushing zero interaction updates, or system memory access.

Okay fair point, so it seems we're miscommunicating the semantics. In math, there's proofs, so I'm not sure what term is better to say other than disprove, or to say the original claims of the Tragedy of the Commons has been refuted as dependent on false pretext or premise.

The meta problem with The Tragedy of the Commons is it was communicated almost as if it were some law of physics but the rules are synthetic and when the framework of ToC enters the real world it is invalidated and defeated by the human systems it exists in. ToC became a reference as a narrative weapon to justify social engineering policies, where lobbyists claim it as some sort of scientific finding. Ostrom researched case studies of common pool resources, and identified how to prevent the tragedy of antisocial exploitation through a system of accountability, auditing, and explicit retaliations. ToC is basically just libertarian/anarcho-capital dynamics.

So sure, ToC behavior exists and is a problem. Ostrom found the solutions to fix the problem.

I guess that's a fair point. It will be interesting to see how unregulated AI plays out.

It seems like one other aspect to this is a question of how these systems are all very new and we're already seeing addiction and psychosis from adults using them. Apparently there's laws in China that limit the use of social media and video games for anyone below a certain age, and same with the use of LLM tools. There's mandatory education and training on what LLMs are for certain grade ranges.

At least there's some transparency with open weight models. With closed models it's harder to audit for censorship or bias. Even with "open weight" models there's no transparency with training datasets.

Elinor Ostrom proved The Commons can be collectively owned and managed. It's nothing to do with the scale being small or large if you actually read her research studies and the game theory models she published you would know that but instead you seemed to have use an LLM to frame your argument.

The main findings from Ostrom's research on the success of managing commonly owned pooled resources, and how the Tragedy of the Commons was wrong, is basically just enforcing audits, and retaliating against selfish individuals as part of a system of rules and social reputation. The other major game theory element is pointing out humans interact many times not just once, and that much of the original game theory models had arbitrary rules like players cannot communicate, or assuming the game only happens once, or assuming that the game cannot change or new players may join the game and punish the players that have been mismanaging social trust and resource policies.

"Conventional wisdom says that common ownership is a bad idea. “That which is owned by all is cared for by none.” Therefore, all scarce resources should either be owned privately by individuals or be regulated by central authorities. Or should they? Elinor Ostrom rejects that conventional wisdom. Based on numerous empirical studies of user-managed fish stocks, pastures, woods, lakes, and groundwater basins, she concludes that common property is often well tended."

There's a study of the grasslands in Mongolia, but that's just renforcing Ostrom's findings that the lack of auditing or policies for sustainability is what drives unsustainable behavior. Mongolia started to increase livestock and cashmere exports and their economy has not invested in modernizing or maintaining sustainable industrial scale livestock practices. Goats for cashmere has been a major source of income for Mongolian nomads but because they are essentially living out libertarian economics it's just more proof to validate the game theory factors Ostrom won the nobel prize for.

https://www.nobelprize.org/prizes/economic-sciences/2009/ill...

From the Nobel Prize website:

    Elinor Ostrom identifies seven keys to successful cooperation:
  • Rules clearly define entitlements.
  • Conflict resolution mechanisms are in place.
  • Duties stand in reasonable proportion to benefits.
  • Monitoring and sanctioning is carried out either by the users themselves or by someone who is accountable to the users.
  • Sanctions are graduated, mild for a first violation and stricter as violations are repeated.
  • Decision processes are democratic.
  • The rights of users to self-organize are clearly recognized by outside authorities`