ricksunny

fake it till you make it. (looks around at real life).

>We still haven’t solved how to help people stay oriented as hundreds of increasingly capable agents change a system at once.

Engineer, meet business school.

One of my favorite LLM use-cases is alternate history extrapolation. Since there’s no formal ‘right’ to a counterfactual, then it can’t be ‘wrong’, and it pretty much always clies me into other events of relevance that are hinge points on my history bead of interest.

…or they just use a knowledge graph under the hood and don’t publish about it.

The OP article is surprisingly candid about their critique - not that the AI-generated style they've identified is bad, but that it is so common. Therein apparently lies one frailty of our human condition - if everyone can do it, it becomes out of fashion at best, or even stigmatized at worse. I feel there's a place where that's a rational instinct, but under an extensive backdrop where the instinct is irrational - aka, 'art'. If someone needed an explanation for 'why can't we have nice things', I think OP's article articulates it very well.

‘knowing what problems are worth solving —- priceless”.

For everything else, there’s Astracard

I’m interested in what Astra’s test for german-language plaintext looks like. No doubt bog standard among cryptanalysts, but not intuitive for the rest of us. Otherwise it’s difficult to determine a goal state that ‘done’ looks like

>We're in the phase where the factions are geographically sorting, identifying allies, and arming.

In this framing what are the factions? What does geographically sorting mean - are the factions moving themselves from place to place?

"…it had women in authority roles, but also in sexy outfits…"

We shall not be surprised that Jeffrey Epstein was exploring remaking this show https://www.justice.gov/epstein (search 'ufo')

Science was presented in a just-so way in our youth. It still is ( #Scicomms ), but back then, whether reading Junior Scholastic or the margin captions of a high school science textbook handed down ex cathedra, there was no communication pathway to pose vexing questions in a publicly visible way as we do now.

We don't even know if Miller or Urey themselves would have (or did) frame their results that way. The honest framing is 'hey this is one way to form amino acids, and maybe that pathway is reflective of the origin of life, but that supposition remains conjecture on our part.'

This was good. I think I want to see a boilerplate on all future substacks I read that says "I read sockpuppet's advice." (bonus points for "here's a hash of my visit log")

As one of the AI-wielding amateurs, I got positive feedback from an audience member on this Ai-gen poster for a talk I gave earlier this summer https://drive.google.com/file/d/1MaawQcupC5XHqWBiTujSpKOKohx...

The clutter is there yes - it's not a sparse 1970s kitsch cultural event promo or 1970s sci-fi book cover (random example I found in 2 minutes https://www.etsy.com/listing/715136429/david-bowie-moonage-d... ) that's for sure - but I wouldn't want to remove any of mine's elements as each one informs the viewer at a glance their decision whether to attend or not.

‘And then OpenAI "had to shed a bunch of dead weight like Sora," as he sees it, "because Anthropic was going for the jugular."’

https://x.com/hilbertspaess/status/2099954626040905834

(Coxon)

surprised the word RAG hasn’t come ho in this thread (except for a likely-LLM-generated-and-therefore-downvoted comment).

>with settings, tools, and context

call me crazy, but I think that this kind of suite, which training-uber-alles people generally dismiss as trivially replicable ‘wrapper’ is actually the differentiating factor for LLM adoption today and moreso into the future.

I’m not dismissing the near-all-out impact of training, but from a competitive busines or industry-structure lens, we’re looking at the three or four big players competing their utmost ultimately, if unintentionally, to turn foundation model access into commodity.

To the capabilities-maximalist minded (typical among engineers - my former life so I’m familiar don’t lack guilt in committing that) folks who will say “Oh the foundation model megacorps will just build out any wrapper whenever one of their third party wrapper plays demonstrates enough adoption, my rejoinder:

Apple did not rebuild an Uber-like app and cut Uber out.

We’ll see how this OpenAI legal services industry wrapper plays out, but I suspect 1) the third party legal wrapper plays will run to other foundation models not doing a legal wrapper, and 2) 3rd party wrappers will do a better job of it since it’s their all-out focus, unlike OpenAI’s whose priorities are necessarily more generalist.

Yes, we all remember the breakout startup failure-arguing quote “Google has entered your space.” That worked for several high profile applications. I believe more of those bets died on the vine than broke-out succeeded however, we only remember the biggest ones that persisted.

If legal services AI turns into one of the Mail or Maps-scale applications of the AI industry, while that would be a fair strategic action counter to the thesis I’ve laid out, the thesis itself would still tolerate it. It’s a question of short-fat tail vs mid-to-long-tail application scope & attractiveness. For example, I think it’s clear that coding is one of these short-fat-tail applications, and the low-no code plays are absolutely having their lunch eaten to acqui-hire ‘death’. I just doubt that the same will persistently transpire facing all professional service wrapper plays.

>We were building 40 Gbps packet sniffers at Google (4x 10 Gbps NICs) and needed switches that could do things like mirror traffic across ports at line rate

I'm sure there were good reaasons for this, but man this just sounds bad. Like that's the spec I think NSA must give to their contractors outfitting 611 Folsom Street's Room 641A

The UX sits in an interesting hybrid place between Gather App and Katmai - would be fun to try out with collaborators sometime

>I can't wait to see Apple Verified™ photos of UFOs flying over the Golden Gate Bridge.

There’s no such thing as a Golden Gate Bridge.

Prove it.

Not denigrating need for such a thing, (and appreciating @hobom 's mention of LSVP I was unfamiliar with) but noting with historical precedent this style of protection reflex does not come without lapses (ahem allegedly): https://en.wikipedia.org/wiki/2001_anthrax_attacks#Bruce_Edw...

Interesting name for an AI company. Isn’t the andon cord the thing that anyone in a 6-sigma manufacturing line can pull to halt it until a concern can be addressed?

does it self-uninstall cleanly (i.e. without leaving a footprint behind) yet? if not I don’t care about this or the next ‘upgrade’, and I can’t believe Homebrew project leaders push upgrades without addressing this basic flaw.

I want to see models of Vannevar Bush's rapid analytical machine computers supplied to the proto-NSA.

See Colin Burke and Pieces of the Action

> Do you think things like the Manhattan Project were a mistake

Yes. At best now the mother of all cautionary tales.

"every hour spent aligning with a co-author is an hour not spent producing output that is legible to an evaluation system." use of the word 'legible'.

"We defund the corridor and then wonder where the corridor conversations went." this is a claude-ist construction.

"This is an old worry wearing new clothes" this is a very claude-ist construction.

(rather ironically... as @embedding-shaep pointed out earlier): "Outsource the writing and you have not accelerated the thinking; you have skipped it." 'not X but punchy-Y' (not to be confused with 'not X but profound-Y'.

I'm not saying the author shouldn't have used an LLM - once in a while I suppose it gets used to good effect. But I wouldn't kid myself that the writing is certainly all human through-and-through.

[Incidentally, in case anyone's wondering where many of these claude-isms come from, just search within lesswrong.com .]

One statement at the end of this excerpt merits scrutiny:

" A variant of these viruses capable of human-to-human spread would therefore be of very high concern. Moreover, it is possible that such a variant could also have capacity for severe disease outside of the respiratory tract. Unlike other influenza variants, H5 viruses (of which this avian virus is one) often show striking brain involvement in cats, foxes, ferrets, and some human cases. A pandemic variant with such properties would be especially concerning due to its potential to increase disease severity, confuse diagnosis, and hinder treatment.

As in the first case study, this research was clearly dual use in nature. Understanding the genetic basis of these specific viral traits could help in the early identification of naturally-emerging versions of the virus—versions with the potential to cause a human pandemic" (emphasis mine)

This statement flirts with the Ron Fouchier (Netherlands) risky-grant-justification thesis, repeated ad nauseam by Peter Daszak in grant applications, who has been cut off from federal funding.

1. Premise: it would be good to surveil for & monitor viruses in nature that are near-ready to spill over to humans from nature;

2. Protocol: We will serially passage bird flu in ferrets until its virulence and/or transmissibility is high. (ferrets are treated as a model mammal stand-in for humans) Comparing genetic changes (mutations) as sequenced along the passaging pathway will tell us what to surveil for in nature.

As has long been debated in recent years, we have no idea if nature (in any given natural instance, if ever) will choose the pathway that serial passaging (in one given lab instance) produced in order to demonstrate higher virulence and/or transmissibility in humans.

Formally, Anthropic's statement recapitulates the premise only, and we don't know what protocol (the "Understanding the genetic basis of these specific viral traits" part) if any was being queried for. Whether a given regulatory regime's policy allows say for purely in silico investigation of same, for the result to be credible it would need to have been leveraging an empirically-verified (i.e. real-world) training set. Generating that training set would risk creating the pandemic that its supporting grant application tries at justifying to prevent.

But the offending prompter and similar user-LLM exchanges' potential to enable GMDs (Grants of Mass Destruction) should remind us to be very much on our guard against epistemologically bankrupt protocol outlines enabled by LLMs & their prompters' motivated reasoning.

proud disclaimer: I am an advisor to BiosafetyNow.

weird obsession. really.

Here, have a Kevin Esvelt, now-tenured professor at MIT: https://www.nytimes.com/2026/04/29/us/ai-chatbots-biological...

disclosure: I am an advisor for BiosafetyNow.

We have somewhere between 0% < X < 100% probability that covid-19 (remember that?) was an accidental lab leak by credentialed virologists not making use of AI in the post-2022 sense.

If I don't trust the credentialed virologists without AI based on perceived precedent, I'm certainly not going to trust them with AI to accelerate their their unabated workstreams.

Note: per my HN profile, I am an advisor for BiosafetyNow.

Good on you. I left my post-university defense job after a year and a half. Of all career decisions, that's one I don't regret, and rather the regret I harbor was for joining in the first place (albeit did so under a combination of limited information & naivete).

The 'will not' list in here I Just learned about today, https://www.solzhenitsyncenter.org/live-not-by-lies it easily maps to the employment situation question.

1000%. My social community if reference is nomads of varying skills. The lower-skilled among them (the ones who were on the digital marketing, crypto,, now into building sales CRM tool) are manifestly demonstrating addictive tendencies around vibecoding) - the chain-smoking ADD guy pulling all-nighters still sticks in my mind. I myself (also low sw-skilled, just not the digital marketing / crypto hustling type) am exhibiting these tendencies in the form of all-nighters that in practice produce very little leverageable value-add. To remedy I try to be conservative in my goals - like only produce a comprehensive english-language PRD instead of any code till the tools get better and stroke the gambling instinct a bit less.

(By the way this is not to say that all nomads are low-skilled or even low-sw-skill, not by a long shot. I’ve met plenty of CS-degreed nomads who I treat as the wizened experts that they are, from whom deferentially elicit their pearls of hard-won software project wisdom. It was one of these even who, non-derisively, recognized the ‘pulling the slot machine lever’ reflex among vibecoders after I posed the apparent addictive tendency. On later describing this to a longtime friend deeply conversant in social science, he immediately responded with a nod and the phrase ‘variable reward response’.)

(Appreciating this entire thread) - side question: you used the word ‘shape’ in the abstract sense, something I never came across until Claude vibecoding came along. Was it common / did you use ‘shape’ in the abstract sense before say 2025?

Also, my coming from being a non-coder, I have a lot of appreciation for the possibilities for project failure one way or another due to a data model or project schema being wrong, even though I still only have a superficial understanding of what either of those concepts even are. . My question is, does your conception of data models in the abstract come from a formal academic course, like an algo’s & data structures course, or from trade-knowledge acquired through the practitioner grapevine?

My dreams of a magnetohydrodynamic hand water-cannon are dashed sniff