elliotec

A charitable read of this article isn't even necessary to understand the point of this is the opposite of the poster you replied to. Basic discernment at most.

They don't, did you read it? It's clearly a satirical art project exposing the bullshit.

Did just that with your comment. Seems like an exercise in discernment.

You’d think the most powerful AI could make a website that doesn’t completely break usability by destroying scroll.

EU and China are perfect inversions of your point.

China suppressing ethnic, cultural, and linguistic diversity for homogeneity has caused immeasurable suffering and imperialistic destruction.

EU explicitly embracing ethnic, cultural, and linguistic diversity for heterogeneity has brought more peace and prosperity than ever before to a continent in perpetual conflict for nearly all of human history.

Apparently it may surprise you that driving on both the left or right side of the road depends on which country of the EU you’re in.

Another great example of the perverse idea that “same == better”

I don't know where "here" is, but yes, translation can be hard. That's fine. It doesn't mean we can't get along or have a peaceful world full of diversity in language and culture. We just need to get better at nuance and understanding.

A robots.txt explicitly allowing LLMs has nothing to do with whether it was made for humans or not and is generally the right move assuming you want your site picked up and recommended by AI tools. No different than allowing search engine crawlers to pick it up and the authors focus on that is bizarre when the rest of the evidence is so stark.

We know with extremely high certainty that classical psychedelics effect other animals similarly given the neuroscience [1]. But this is not a classical psychedelic. We don't know how it works yet, what receptors it operates on, or what the pharmacology is doing.

So would dogs see little humans or little dogs? We can't say for sure. But the best guess might actually be most likely little dogs, or maybe little smells of little dogs. Dogs show conspecific preference (species-focus) while we show facial preference [2] and they have a strong focus on olfactory sense while we're more visual. The vividness might show up in ways we can't really perceive through smell like they can.

And the answer to "does that mean the fungus has targeted humans genetically" is almost definitely no. The study the article is based on [3] mentions that lilliputian hallucinations predate this mushroom, appear across cultures' folklore, and are reported in alcohol withdrawal, dementia, and macular degeneration. The effect seems fundamental to how the human mind and brain work. So it's more likely that the mushroom is disrupting something and the brain is resolving that disruption into little people, because that's what it does with pattern matching.

This mushroom is an ectomycorrhizal symbiote of the Yunnan pine tree, is ancient with the genus being at least 5 million years old compared to modern humans 300k, and requires undercooking which isn't compatible with human targeting so it doesn't make sense to have any genetic human target.

The fungus supplies a non-targeted disruption and the resulting little people content we describe is the human brain's response. A dog getting the same disruption would resolve it into whatever a dog's machinery produces, or maybe they metabolize it into nothing recognizable.

Pretty damn cool.

[1] https://consciousnesslibrary.org/research/29070-correlation-... [2] https://www.jneurosci.org/content/40/43/8396 [3] https://consciousnesslibrary.org/research/27751-phylogenomic...

They just said they work on Gemma at Deepmind. That stands to reason that as of time of writing, we should assume the answer is yes.

Rails 8's default Active Job backend is solid_queue. It lives in the same single Postgres as everything else (including solid_cache). No Redis or anything to have to wrangle. Recurring scheduling is built in. Another big plus for using Rails for this.

As I understand, it uses Postgres's "FOR UPDATE SKIP LOCKED" job claim feature. I don't run most jobs concurrently because they need to build on each other. Postgres advisory locks sit on top for mutual exclusion and solid_queue handles queuing. It's akin to good_job which is the pg-native thing from before solid_queue existed, but I just stuck with the easy Rails default and this is the first time I've even thought to remember how it works.

Thanks! One minor note, it's got pretty much all accessible research on psychedelics, but it's not exactly specific to psychedelics. About half the corpus is research on meditation/contemplative traditions and philosophy of mind/consciousness theory. They all go pretty well together though!

Hey, thanks so much for the kind words and taking a look!

Search observation: What you're noticing with search is the currently intended behavior, but you raise an interesting point. I'm using tsvector for Postgres full-text search over title, abstract, etc. with prefix matching, specifically in English. The stemmer doesn't decompose Sanskrit so "brahmavihara" and "vihara" are separate tokens matching legitimately separate papers. Then it falls back to the embedding geometry to search by meaning instead of strings.

That said, there's enough Sanskrit going on in contemplative papers that a synonym/variant layer for Pali and Sanskrit vocabulary is a pretty good idea. There's a lot of German and Portugese in the corpus as well, so upgrading the search to account for all these languages is a great feature idea.

Tech question: There's actually zero LLM involvement at all regarding citations or references in this project. That all comes from OpenAlex API. LLMs only do relevance gating, study detail abstraction, and evidence synthesis summaries.

Google has a moat in the sense they have no API and don't license anything to anyone, so nobody builds on it and the open ecosystem of sources I'm using is the same as what AI products use. Gemini is trained on the same sources this library uses (and recently, this library itself) and the Google Scholar vault is in my opinion, pretty useless given the open ecosystem available.

Research question: the most mind-blowing results I've discovered are in an unexpected direction - it is wild how thin and low-confidence the evidence is for basically any claim regarding psychedelics, due primarily to functional unblinding and low sample size and diversity. The hype cycle is way beyond science can keep grounded.

Here's an example (disclaimer, it's my own published research!): A recent study reported an effect size for mebufotenin exceeding the theoretical maximum for antidepressants, implying methodological issues account for the results. https://consciousnesslibrary.org/research/52248-blinding-int...

Another "mind-blowing" observation is the sheer volume of consciousness work outside the psychedelic hype cycle. A full half of the corpus is on meditation or philosophy of mind, not related to psychedelics. And we still are so far from understanding the "hard problem.

Development question: I have plans to extend this indefinitely, and your factoid idea is interesting. The "study at a glance" might have most of what you're asking about already on each article, but taking that a step further, it could be cool to have a cross-corpus browsable feed of those. I'll also be integrating data from a project by Josie Kins (maker of https://effectindex.com) on psychedelic substance topics in particular for dose info and potentially trip reports, but that's a little orthogonal to what you're suggesting.

Let me know if you mean something more about factoids than the "study at a glance" and the potential to have those be aggregated in a browsable way and I'll see what I can think of.

I haven't bit off more than I can chew yet, my plan is to work on it until that's the case OR there's just nothing left to do in the scope of the project.

Thanks a lot for your comment!

Yeah I’ve used mainly Claude code and recently kimi with pi for dev. Honestly the vast majority of features and improvements happened in the last few months with AI enablement. The Fable fake-outs really pushed me to optimize my time with it.

On experience reports - I decided to keep it purely academic at first but will soon be working on integrating data from a tool by Josie Kins (the maker of https://effectindex.com).

I also don’t know of anywhere that aggregates reports from meditation and dreams etc, but that’s a really interesting idea. Maybe this can be the place!

It does not - yet. I explicitly chose to keep it academic for now, but I considered over and over if I want to pull in subjective experience or legal content and decided to focus on academia for the initial push.

That said, I've been working with Josie Kins who runs https://effectindex.com/ and she's been building out a new system that incorporates categorized subjective effects on psychedelics, dosage and harm reduction info, with a huge amount of trip reports from all over the internet that we'll integrate into the individual topics sections probably in the next couple months.

Thanks for taking a look! Yeah, these programs are very new and pretty uncommon (for now).

As for why I chose Rails - I just know Rails really well. I've been using it since 2012 or so, and it's basically muscle memory now. The most intensive parts of the app really would lend themselves to Python tooling and I did end up hand-building a lot of the data engineering stuff that there for sure exist plugins for in Python, but the trade off of using what I know well outweighed that by a lot. Plus the rest of it are perfectly CRUD-able server-rendered pages. Pretty Railsy aside from the data eng stuff.

I also started this before AI coding harnesses were available. I shelved it for a long time after the initial few rounds, but the work I did in the beginning was substantial enough to not just start over. If I had Claude Code with Fable when beginning, it might look a lot different. But I do like knowing everything about what I built and understanding every line.

On your AI ethics question... I do think it's possible, eventually. That's why I hedged a bunch on it being a long term goal. If open weight models gain steam and hardware acquisition becomes manageable, it would be a dream to run a self-hosted self-trained model in my house (which is powered by solar, so the renewable thing is real). It is especially realistic if some funding comes through for hardware.

Thank you! I couldn't agree more. Big questions to answer, and your framing is exactly why I went into this.

I've been building https://consciousnesslibrary.org, an open library for psychedelic and consciousness research that pulls from over a dozen academic APIs every 20 minutes, dedupes them, organizes by relevance, and synthesizes evidence.

I just finished my master's in Psychedelics and Consciousness Studies, and built the library I wanted while studying. I started it over a year ago, but put it down until my practicum advisor suggested I dust it off and go hard on it during my last term.

As of this writing, there are over 35,700 papers by over 82,200 authors organized into 37 topics, growing daily. There's a 2D map of the whole corpus and evidence syntheses on the topic pages, plus the ability to generate your own syntheses with any prompt.

It's built with Rails 8.1, uses one Postgres (for full-text search, vectors, and job queue), and runs on a $24/mo DigitalOcean droplet that I just upgraded from the $12 tier and deploy to with Kamal. Total LLM spend so far is ~$28 in DeepSeek-V4-Flash, trending down after a big one-time backfill push.

Relevance was the hardest part to get right by far. Ensuring we keep papers on LSD the psychedelic vs Lumpy Skin Disease, or Ketamine for depression vs anesthesia in cats was no small feat. The design ended up being a cheap keyword prefilter, then an LLM rubric that accepts or rejects papers. More on that here: https://consciousnesslibrary.org/docs/article-pipeline.html

It's free, no ads, no signup to read anything, and a registered 501(c)(3) I fund myself for now. I'd really like to know what breaks or any other feedback, and would love to answer any questions about it!

I built an open library for psychedelic and consciousness research that pulls from over a dozen academic APIs every 20 minutes, dedupes them, organizes by relevance, and synthesizes evidence.

I just finished my master's in Psychedelics and Consciousness Studies, and built the library I wanted while studying. I started it over a year ago, but put it down until my practicum advisor suggested I dust it off and go hard on it during my last term.

As of this writing, there are over 35,700 papers by over 82,200 authors organized into 37 topics, growing daily. There's a 2D map of the whole corpus and evidence syntheses on the topic pages, plus the ability to generate your own syntheses with any prompt.

It's built with Rails 8.1, uses one Postgres (for full-text search, vectors, and job queue), and runs on a $24/mo DigitalOcean droplet that I just upgraded from the $12 tier and deploy to with Kamal. Total LLM spend so far is ~$28 in DeepSeek-V4-Flash, trending down after a big one-time backfill push.

Relevance was the hardest part to get right by far. Ensuring we keep papers on LSD the psychedelic vs Lumpy Skin Disease, or Ketamine for depression vs anesthesia in cats was no small feat. The design ended up being a cheap keyword prefilter, then an LLM rubric that accepts or rejects papers. More on that here: https://consciousnesslibrary.org/docs/article-pipeline.html

It's free, no ads, no signup to read anything, and a registered 501(c)(3) I fund myself for now. I'd really like to know what breaks or any other feedback, and would love to answer any questions about it!

You're not helping though. If you want that to be different, focus on different things rather than bringing attention to what you're complaining about.

Could also play around with the multiplier logic, and do more cookieclicker-esque type gamification.

Is it open source? I'd be down to play around with it on a branch i can push up if you're interested.

It's fun! Cool way for new projects to compete to get a backlink, assuming the hall of fame persists forever.

Maybe more pomp and circumstance for crossing some number milestones. A confetti explosion when you double the lead, a cool animation when you cross 10k, something like that.

Maybe for winners of the day, adding the meta description of the link or something.

Better you holding the bag than me. Unfortunately all of us will be holding parts of it.

If that ends up being true, they should be very surprised.

We're looking at ~$725B combined hyperscaler capex in 2026 (on a path to $1.08T by 2028) against roughly $25B of AI service revenue in 2025 on $250B+ of infrastructure spend. By 2030, the global data center build-out will require $6.7 trillion in capital expenditure. If hyperscalers require a 25% return on AI-specific capex, the industry needs to generate ~$169B in AI-attributable revenue annually by end of 2028.

A flexible compute market plus three well-capitalized competitors plus Google's internal silicon means nobody gets to hold price. SpaceX's public offering is the best signal we have on this type of thing and it's down 15% from offer price.

It's incredibly unlikely that BOTH OpenAI and Anthropic will be "obscenely" profitable in the next few years. Also very unlikely that even one will be "obscenely" profitable in the next few years.

It is more likely (but still not very) that they both will be simply profitable (not obscenely).

The most probable scenario is that ONE will be somewhat profitable (probably Anthropic) and the other still burning.

The comment I made above the reply you're responding to was at 9 votes then a few seconds after that reply was made, it dropped to 4. Suspicious indeed, I've never seen that type of swing on HN.

The sources say over 40% of incremental demand will need to be met with gas and coal over the next 4 years, which have 30-40 year book lives. So we're locking in that much more non-renewable energy for at least a generation.

The water thing is a separate issue, and it's not about "water use" itself.

I'm mostly talking about incredibly environmentally destructive data centers and energy use on the current trajectory. Not even thinking about water problems, the gas and coal requirements alone are climate disasters. [1] [2]

But if it is a bubble, which many would argue is the case, an AI capex bust with it's circular financing collapse is one of, if not THE top threat to global financial stability. ~$725B combined hyperscaler capex in 2026 against roughly $25B of AI service revenue in 2025 on $250B+ of infrastructure spend. Not exactly a promising situation. [3] [4]

This is immediate-picture, I'm not talking about a singularity extinction event. Just what's actually happening right now and the trajectory of the next 4 years.

[1] https://www.publicpower.org/periodical/article/electricity-d...

[2] https://www.iea.org/reports/energy-and-ai/energy-supply-for-...

[3] https://www.tftc.io/bis-annual-report-2026-ai-bubble-circula...

[4] https://alcapitaladvisory.com/research/intelligence/ai-infra...

I'm cautiously optimistic that this is the beginning of a much needed correction toward sobriety with regards to AI investment. The macro effects of all the money going into AI are apocalyptic. Maybe we can take some deep breaths and move forward a little smarter if the hype fest slows down a tad. And if we're lucky, eventually be able to afford RAM again.

Indeed, the core of this misalignment is that we have a foundational difference of opinion on that.

The only caveat I'd push is not that I believe that "individuals are mostly fine," but that as a whole, separate from the state apparatus, I believe humanity collectively has its own best interests in mind and is capable of achieving more good than states or coercive hierarchies can or want to achieve.

Thanks for the civil discourse!