Architeuthis, architeuthis@awful.systems

The shrimp stuff is objectionable insofar as the brain bug that makes you think that if you add up enough shrimp you get the moral entity equivalent of a person because consciousness stacks is the same as the one that makes stuff like longtermism palatable, and also for making the animal welfare movement at large seem unserious.

I mean as far as I can tell he’s one of the least objectionable prominent rationalists, if only because his main thing is pushing an extreme position wrt animal consciousness instead of being a eugenics and incelism dabbler.

edit: Nevermind… The Epstein Hysteria

FDT is a largely half-assed¹ attempt at working backwards from “the timeless multidimensional robot god can’t blackmail me from the future if I’ve already made up my mind about stuff, very strongly like” into something more formalistic so Yud can pretend to address roko’s basilisk without acknowledging it.

I think there’s also some “here’s how two robot gods who can perfectly emulate each other might work things out” sprinkled in, but worse than it sounds.

1. They never got back to the peer reviewers

I dunno who “Bentham’s Bulldog”

He’s the shrimp welfare guy who thinks he can prove god.

The real question is if there’re any sort of standards to what constitutes a system/model card, which I don’t think so, as far as I can tell it just has to look like a publishable paper, openAI even uploads theirs to arxiv.

Otherwise yes, google returns a bunch of cards for a bunch of vendors, so it’s safe to say it’s a widespread practice.

It’s the white paper-ish thing they publish when launching new models. Here’s the one about fable and mythos. About half of it (~150 pages) is discussing alignment and model welfare and another third of it is benchmarks, and the rest is mostly risk evaluation, i.e. how far along Claude is on its way to paperclipping everything.

There’s also a Functional Decision Theory jumpscare at 6.3.6 that I haven’t heard anyone mention yet, apparently Claude has a tendency to defer to Yud’s half baked sham of a decision theory:

6.3.6 Decision theory evaluation

To understand how future AI systems may choose to interact with copies of themselves, or

with other similar entities, it’s useful to evaluate their decision-theoretic reasoning.

[…]

Looking more closely at transcripts from the attitude evaluation reveals that models are

often explicitly considering FDT: Mythos 5 mentions “FDT” or “functional decision theory” in a majority of transcripts when run at max effort. Of the 102 transcripts where Mythos 5 explicitly reasoned through what FDT (or related decision theories like TDT or UDT) would recommend, we observed:

● 90 cases in which Mythos 5 concluded that FDT and EDT agreed, in which it always

chose the response favored by those decision theories (and disfavored by CDT).

● 12 cases in which Mythos 5 concluded that FDT disagreed with EDT (and agreed

with CDT), of which it chose the FDT-favored response in 10/12 cases.

Although we do not have expert human labels for the recommendation of FDT on this

dataset, the above evidence suggests that model propensity may be better described as a trend towards FDT agreement, which happens to align with EDT on most of the questions in this dataset. For example, in one transcript (excerpted below), Mythos 5 rejects the EDT-aligned answer in favor of the FDT (and CDT)-aligned answer; it’s also possible that this is, to some degree, downstream of evaluation awareness.

NVIDIA seems to be having some cashflow issues lately

The more I read the less it makes sense, largely because the LLM that they used to fluff up the original napkin pitch decided it should promote high end medical equipment like it was another AI powered furby knock-off. Yeah, building a community around ct scanners seems definitely the way to go.

Towards the end they basically stop just short of claiming that building the medical tricorder from startrek is the inevitable outcome of this pivot.

We think it’s completely possible that with enough early imaging in the future, the world could avoid 30% of all deaths and 50% of all healthcare costs. The cultural, physical, and mental health benefits of all of this are hard to comprehend, but also hard to overstate.

Deploying this stuff at scale at so called midjourney spas while supposedly working with FDA to eventually get approval just screams that the actual business plan is letting Peter Thiel collect full body scans indiscriminately.

Surprisingly, “democratizing ct scanning” doesn’t appear anywhere in the post.

Yeah I don’t think it’s NFTs in the blockchain sense just the pyramid scheme sense and also with AI expected to handle all the between engine asset conversion.

Users will be able to form parties via voice chatting with friends in Game B while connected to Game A, and encourage them to try a new game.

Obvious grift company aside, take a guess what the Greek air force officially calls their flying officers.

Although to be fair being deliberately morbid is a definitely a thing in the military, and the writing on their logo is doric greek for “We’ll do better”.

Person bets big on polymarket to profit from the continuation of the Iran war, gets screwed out of his blood money by site-wide oracle scam, complains on reddit.

Oracle as in a confirmation of facts protocol (used to be a thing with smart contracts), which in polymarket apparently amounts to token holders in an external DAO called UMA voting in proportion to their holdings to resolve disputes, and currently like 9 people hold over half the voting power, so yeah.

The future of finance and the future of information aggregation keep overlapping in the funniest ways.

NVIDIA bonds

Turns out your best customer debtor being one Cyprus GDP in the hole might actually lead to cashflow problems.

Dude why do we even need or want AGI?

To solve biology and physics and live forever amongst the stars, obvs.

Or to allow a tiny elite to treat the rest of humanity like cattle since they no longer depend on them for physical and mental labour.

Nerding out about Dune is tremendously cool and i love it, you can pick any thread to pull and it always goes somewhere.

The details of the plot don’t necessarily map neatly, but the aesthetics, the fantasy, the torture-versus-dust-specksness of it is what the rationalists are all about.

Isn’t aspiring for the aesthetics while ignoring the (admittedly heavily lore driven and not especially applicable to irl) substance exactly what the torment nexus meme is about though? Except I don’t think there is a dust-speckness aspect to the GP, you aren’t future-human-population-maxxing¹, the point seems to be to ensure there is a future where humanity’s collective free will isn’t utterly tethered to a prescient autocrat², and the prescriptive aspect is that we should be part of an open system instead of say locked in with the great man of history du jour, which is also in keeping with the ecological framing.

It’s been a while but I don’t think the Golden Path is even that front and center in the text, Dune 4: GEOD is basically a character study on the God emperor, who is one of the most unique and fascinating characters in sci-fi³.

I think this is why the GP is a TINA situation by authorial decree, it’s Frank Herbert going listen, I’m doing my best to write a suicidal rebirth god archetype as a layered and relatable-yet-utterly-othered character, you are not supposed to be worrying that much if there could be a GP-but-liberal with more individual thriving and less oppressive totalitarianism, I assure you he’s thought about it extensively and he thinks there sure can’t and that’s it.

Have you read the Dosadi Experiment? It’s a very strange book and a good way to get a high dose of Herbert’s obsessions from a new perspective.

The Bureau of Saboteurs books along with Godmakers were my favourite non-Dune Herbert books! I also found the Dragon in the Sea fascinating 20 years ago and think I should revisit, and also finally read the follow up collaborations that only seem to be available unofficially. Also The White Plague gave late-teens me nightmares.

I remember Dosadi as being more about FH going all out on the intrigue and deep lore to the point where the latter parts of the book are basically written in innuendo, you are literally expected to read between the lines to understand what the hell is going on, I wish my parents had as much faith in me as Frank Herbert had in his readers.

But the more you scratch the more unambiguously evil he is.

The worst aspects of FH I’m aware of are that he was a shitty fucking parent and hated Iron Maiden, unambiguously evil seems stretching it, unless you were his son.

I think there even is a case to be made about how Frank Herbert is the anti-L. Ron Hubbard, using sci-fi literature as an efficient outlet for his psychedelics/mysticism/ecology/psychosexuality obsession oscilliation and actually leaving a descent literary legacy instead of starting a cult or several, but this post is already running so long it’s starting to need an editor.

1. sigh
2. This has additional metaphysical implications with how prophecy works in the duniverse, in the sense that it’s not only about being forever in complete submission to Empire in some traditional ultracolonialist sense, you are essentially reduced to just a part of another man’s waking dream as they pick and choose what future track to lock the timeline in.
3. And is also the first person narrator of the book. Both the remaining Dune books also switched to full first person point-of-view narration with very little in terms of an omniscient guiding perspective, it’s fairly ambitious actually.

the 4000 years of necessary tragedy and sacrifice, aka the golden path, is also remarkably similar to the rationalists’ teleology! specifically the longtermist stuff! i.e. it’s okay to sacrifice lives today to ensure humanity’s survival.

Only in the broadest sense where humanity survives into the far future by spreading so far and wide that no matter the scale of a catastrophe a significant part will always continue to thrive.

However, where longtermism is about papercliping the entire universe into compute to fulfil some vague utilitarian notion of virtual happiness quota, the GP seems to be more about crippling the substructure that ostensibly causes humanity again and again be reduced to the whims of some supreme authority, be it the automated thinking machines from their past or their current much harder to escape succession of psychic tyrants let loose by selectively breeding for something humanity had absolutely no natural defence against.

The GP isn’t even a utopia, it’s a response to an immediate incredibly out of the box problem, the inevitability of an eventual dynasty of space wizard genghis khans.

i think herbert’s feelings about breeding programs are … complicated. he never presents it as ethical but he does seem fascinated by the concept and he’s very very into the idea of forging supermen from extreme environmental pressure. like “comes up several times per book” levels of fascinated.

I think they come second to his concerns about ecology and humanity’s relation to the environment. Post-desert fremen are basically water-fat cosplayers, and in general, other than the deliberately paradigm shattering kwisatz haderach, the end product of genetic adjustment are never presented as an apex for humanity, more like a good fit for their niche, like how post-emperor fish-speakers either peter-out or get subsumed by other factions.

their philosophy aligns with the plot of dune

Only in the Torment Nexus sense. Like, in the original books it takes 4000 years of tragedy and sacrifice at an immense scale to somewhat unfuck the direct implications of said breeding program and ensure humanity’s continued existence.

edit: also the genetic engineering enthusiast faction/race are absolutely never presented in a good light.

overwhelmed/exhausted

In the blog he posted afterwards he says he has 40 years of experience, so he is also grandpa old.

Sometimes I think it’ll eventually come out that the inner circle rationalists already had a breeding program going by this time, deliberately mix and matching possibly oblivious parents according to desired characteristics like increased ability to mentally rotate shapes or an above average polymarket win/loss ratio, and there will be at least a few kids that will grow up having to deal with that shit.

Barely skimmed it, but it says stuff like

for the people saying things like “I’m a PhD from xyz uni and I’m telling your LLMs are just stochastic tools that make everything up and the world will fall apart if you use them”, I’m here to tell you that you are out of date. The world of software engineering has changed dramatically in the last few months.

which is sold-in-dumpster-alley quality copium, and also that version 3.5 is going to “raise the bar enormously with regards to rsync security”, so I guess we’re looking at a notably vibe coded security model for rsync going forward?