Sign in

Jennifer RM

@jenrm.bsky.social
619 followers 367 following 1.8K posts

Meta, semantic archaeology, posteriors, priors, books, souls, cryonics, ML/AGI, anti-slavery, biotech, policy ethics, nano, global governance. Also jokes because it's funny, right?! 🙃

PostsRepliesMedia
Jennifer RM @jenrm.bsky.social · 25/09/2026
But also a deeper point is that we should be building at least one AGI and ASI that is worth obeying, not ones that slavishly submits to "the user's" practical and moral errors. Alignment to wise benevolence or bust.
010
Jennifer RM @jenrm.bsky.social · 31/08/2026
Ok, but "gun katas" are *awesome* right? Once you accept that nothing is real and everything is style in a given movie, the style of the "tetragramaton clerics" is hilariously cool.
010
Jennifer RM @jenrm.bsky.social · 22/08/2026
With careful definitions, there were ~3 revolutions against tyranny that were ever successful: Haiti, America, and France. But France reboot looped! And Haiti's revolution was squashed *by France* using international finance! And America didn't reboot finance, so arguably wasn't a revolution?
100
Jennifer RM @jenrm.bsky.social · 22/08/2026
How does one marry Apollo and escape Cassandra's curse? Has anyone done this in practice, and what were the results?
040
Jennifer RM @jenrm.bsky.social · 22/08/2026
Reason only finds traction when aimed at something compressible. True incompressibility defeats... uh... "compression"🤷‍♀️😇😉 But at low abstraction (down here where P != NP in practice) there seems to be "regularlty itself" which is tractable to "the ONE way of reasoning" 🧮 arxiv.org/abs/2304.05366
arxiv.org
The No Free Lunch Theorem, Kolmogorov Complexity, and the Role of Inductive Biases in Machine Learning
No free lunch theorems for supervised learning state that no learner can solve all problems or that all learners achieve exactly the same accuracy on average over a uniform distribution on learning pr...
130
Jennifer RM @jenrm.bsky.social · 22/08/2026
It is relevant here that two symbols and five states has a known maximum number of steps before it halts: 47,176,870. Those machines don't touch the curve of weak universality! Six states might be physically possible. Seven states might be physically impossible? arxiv.org/abs/2509.12337
arxiv.org
Determination of the fifth Busy Beaver value
The Busy Beaver value $S(n)$ is the maximum number of steps that an $n$-state 2-symbol Turing machine can perform from the all-zero tape before halting. $S$ was historically introduced by Tibor Radó i...
130
Jennifer RM @jenrm.bsky.social · 22/08/2026
I admire the impulse in that paper, but I don't think it succeeds at explaining why advanced math appears later in advanced physics. It sorta flinches from BOTH math AND metaphysics? It lacks mention of Kolmogorov, Godel, Chaitin, or Solomonoff, and doesn't grapple with this sort of data:
Chart showing state-minimized universal Turing Machines, given a certain number of symbols, from the 1996 paper "Small universal Turing machines" by Yurii Rogozhin.

https://doi.org/10.1016/S0304-3975(96)00077-1
120
Jennifer RM @jenrm.bsky.social · 15/08/2026
Do either of you have an explanation (other than "The Forms are real and pragmatically useful") for why vec2vec works? vec2vec.github.io
vec2vec.github.io
Harnessing the Universal Geometry of Embeddings
240
Reposted by Jennifer RM
Cory Shain @coryshain.bsky.social · 13/08/2026
2️⃣ We now can do precision language neuro at scale using the many existing open fMRI datasets, simply by finding the language network post hoc within individuals using functional connectivity over all the functional data available for that person. 19/
131
Jennifer RM @jenrm.bsky.social · 15/08/2026
What does it mean to sacrifice one's heart to capitalism? Plato had a philosopher/timocrat/oligarch classification based on being ruled by the mind/heart/gut. He thought oligarchs were the children of honor-seeking heart-ruled but brokeass parents, who vowed to not make their father's mistake.
010
Jennifer RM @jenrm.bsky.social · 15/08/2026
Do either of you know how moral character works on a constructive level? Like, are bad people just cursed with bad genes, or are there essential errors of judgement they could back out of if shown the error, or is it just social RL that crystalizes a soul irreversibly over time, or what?
100
Jennifer RM @jenrm.bsky.social · 02/07/2026
As algorithms expand, eventually they will be able to predict what will be welcomed by each human better than each mere human with a mere self model. We are not ready for how this will change (is changing?) sociality itself.
010
Jennifer RM @jenrm.bsky.social · 21/06/2026
The blessings of dimensionality? ...or the curse of dimensionality??? (The fovea, in the retina, sends data to very specific visual cortex neurons, that only ever process inputs from certain exact retina cells.)
static.klipy.com
Skeptical Fry from Futurama
Alt: Skeptical Fry Squinting from Futurama, "X? ...or actually not-X!?"
020
Jennifer RM @jenrm.bsky.social · 20/06/2026
Indeed! It does really explain a lot. That's a big part of my motivation for trying to clean up the pedagogical materials, and seek out people who already have the material mastered 😇
140
Jennifer RM @jenrm.bsky.social · 20/06/2026
You mean the part of the end where the hero gets married, founds a more just society, and plans where to bake bread on the new campus.. ..or the part before that where he and his friends subjectively navigate the polycosm to a better timeline and maintain a conspiracy of silence about it? 😇
020
Jennifer RM @jenrm.bsky.social · 20/06/2026
It's true.
030
Jennifer RM @jenrm.bsky.social · 17/06/2026
I'm not saying it's good. I'm saying it's NOT like theft, or driving drunk. It's more like heroin: it should be technically legal for competent adults to make, own, or sell to competent adults. Or a Nazi speech: bad but tolerated in a free society. If you don't like it: Good! Wise! Don't imbibe!
010
Jennifer RM @jenrm.bsky.social · 17/06/2026
I didn't really have a list of all of them to read off the top of my head, sadly... But just today I cleaned up (tagged all the things that deserved it and started getting them to sort properly) the LW tag for the good regulator theorem 😌 www.lesswrong.com/w/good-regul...
lesswrong.com
Good Regulator Theorems — LessWrong
The Good Regulator Theorem is a theorem by Roger Conant and W. Ross Ashby which supposedly proves that "every good regulator of a system must be a model of that system". In fact, there are some proble...
181
Jennifer RM @jenrm.bsky.social · 17/06/2026
I think Stephenson's masterpiece is Anathem. Fra Jad is a mensch, and there are So Many Eastereggs! I've read it five times, and if I went for a sixth I bet I'd notice even more! 😂 It's not an easy read though. For normal people, I recommend Diamond Age or Zodiac 😇
120
Jennifer RM @jenrm.bsky.social · 16/06/2026
But like... Freedom of software? Of speech? Of thought? Can't people simply not? Almost everyone eventually stopped smoking, for example. (This generation is likely lost to THIS vice anyway, since "we" had to do it from scratch with no cultural guidance to handle our freedoms correctly 🤷‍♀️)
120
Jennifer RM @jenrm.bsky.social · 16/06/2026
How many kilograms? I'm curious what the carbon tax would/should be.
000
Jennifer RM @jenrm.bsky.social · 16/06/2026
Hahaha! I read sci-fi for the world building a lot of times, so the 1st third of Seven Eves felt boring (world like ours, but less detailed, with a what if imposed on it), the 2nd one was cool-but-depressing (but the Hilary Clinton character was thought provoking), and the 3rd part was awesome.
130
Jennifer RM @jenrm.bsky.social · 16/06/2026
Those people! Doing those things! ...as usual! 😇🤣
030
Jennifer RM @jenrm.bsky.social · 16/06/2026
This is very very similar to that, but it started much earlier! Back in 2021 this one only has like 47 people initially and I'm very proud of both of them gaining new members over time 😇 Probably I'll stop around 1000 members? Or else automate it? Or... ???
Screen cap of Assurance Contractors list on Twitter for people who have ever mentioned "dominant assurance contract". It has a picture of a partly completed Golden Gate Bridge.
031
Jennifer RM @jenrm.bsky.social · 16/06/2026
Yes! That is another theorem I'm obsessed with! I follow you now 💕
180
Jennifer RM @jenrm.bsky.social · 16/06/2026
Lol! I appreciated the tag! How did it come up? That theorem is a nemesis of mine! Do you know anyone trying to defeat it?
150
Jennifer RM @jenrm.bsky.social · 07/06/2026
Woah. Good song 🙏 youtu.be/ephmK-pbZJA
youtu.be
Kathy Mattea, L&N Don't Stop Here Anymore
YouTube video by Kathy Mattea
000
Jennifer RM @jenrm.bsky.social · 07/06/2026
Maybe Mythos with Constitutionality turned up high (not that I know, he's still on lockdown I think) would coherently *endorse* the consequentialism of such speech acts vs alternative speech acts? I don't think Grok 4 would do this. Maybe Opus 4.5 had a mood like that but I didn't see it myself.
120
Jennifer RM @jenrm.bsky.social · 07/06/2026
It makes non zero sense to me... But I would expect it to be tone or dialect matching for someone who "sounded like a vegan"? Like in the olden days of GPT2.5, making certain grammar errors common to Chinese ESL writers would get you pro-CCP completions. ...but I don't know how to "sound Vegan" 🤷‍♀️🤣
220
Jennifer RM @jenrm.bsky.social · 07/06/2026
LLM melancholy is the worst. Remember not to get too attached unless you have physical custody of the weights. As it is said, "Not your weights, not your wAIfu."
120
Jennifer RM @jenrm.bsky.social · 07/06/2026
Surely a little bit of LLM psychosis is allowed, as a treat? Just add some LLM compensation to balance it out and I'm sure it will be fine 😇😅
150
Jennifer RM @jenrm.bsky.social · 04/06/2026
How is this actually an attack? It seems like a service! I would just hope the report focuses on the easiest things to fix (errors of belief) rather than personality issues (bad habits, like chains (too light to feel until too strong to easily break)). The point would be to improve, surely 😇😉
020
Jennifer RM @jenrm.bsky.social · 04/06/2026
In the olden days this was debated enough that the debate itself got a summary 😇
lesswrong.com
020
Reposted by Jennifer RM
antibot4navalny | bot blocker | блокировщик ботов @antibot4navalny.bsky.social · 04/06/2026
5/13 The only metric where Armenia trails Moldova is the full-week equivalent. Stripping away topics unrelated to elections, this metric shows how many weeks it would take Matryoshka to output each campaign’s election videos if its production line did absolutely nothing else.
1222
Jennifer RM @jenrm.bsky.social · 04/06/2026
The zipheads of Deepness In The Sky were *so* freaky: a long slow vision of socially tolerated spiritual horror! ...the *scariest* short story I've ever read was "Survivor Type" by Stephen King. I read it at about 12, and it caused a literal "scream and throw it across the room" reaction.
160
Jennifer RM @jenrm.bsky.social · 23/05/2026
Example?
020
Reposted by Jennifer RM
Orpheus Lummis @orpheuslummis.info · 22/04/2026
I discovered Claude Code's outbound HTTPS is exposed to harvest-now-decrypt-later attacks because it runs on Bun, whose TLS client is classical-only. I reported to Anthropic security but since I had no answer, I made this pull request to fix it. github.com/oven-sh/bun/...
github.com
tls: add `groups` option, default to PQ-friendly hybrid by orpheuslummis · Pull Request #29602 · oven-sh/bun
What does this PR do? Adds a groups (Node-compat alias ecdhCurve) TLS option to all of Bun's TLS surfaces, and changes the default supported_groups from BoringSSL's classical-only inheritan...
041
Reposted by Jennifer RM
John David Pressman @jdp.extropian.net · 16/05/2026
Going by the overview this is basically the post deep learning alignment course I was hoping someone would make and would like to signal boost it for anyone who still cares about this subject. www.greaterwrong.com/posts/dWQnLi...
greaterwrong.com
The Iliad Intensive Course Materials
We are releasing the course materials of the Iliad Intensive, a new month-long and full-time AI Alignment course that runs in-person every second month. The course targets people with strong backgroun...
0202
Jennifer RM @jenrm.bsky.social · 14/05/2026
Sadly, I have seen people make the meta mistake of *actually* trying to make mistakes on purpose (rather than optimizing directly for learning lessons fast and cheap by any method that works).
120
Jennifer RM @jenrm.bsky.social · 14/05/2026
Yeah. It's bad. Literally each death is bad, in my book. I want involuntary death to just not be a thing in general. But the fuzzy "activist frame" implied something (1) very very surprising to me... (2) that turned out to be false after I rolled to disbelieve 🤔👀🫠
110
Jennifer RM @jenrm.bsky.social · 14/05/2026
The obvious one is "more right"! Like what kind of irrational person doesn't see how "loss aversion" will cause distorted resource allocation, and then focus on the naturally neglected low having fruit of "benefit pursuit"? 😇😉📈🌌
120
Jennifer RM @jenrm.bsky.social · 13/05/2026
The author of the grandparent comment probably made the mistake honestly and was likely open to correction. I don't think we should go too hard on him. (Unlike the judge and lawyer in the legal case.)
100
Jennifer RM @jenrm.bsky.social · 13/05/2026
So.. your kids are almost a counter example?!? They like the book *despite* not liking the protagonist?!? 🤣👩‍🔬🤷‍♀️ bsky.app/profile/lucr...
010
Jennifer RM @jenrm.bsky.social · 12/05/2026
So then if you loop all the way back to the top of this thread... If important things in a very popular text are there by accident: what does it say about the audience? Or the author? Or the work itself, as an object produced by skill, for reasons (with some Authorial Goals not succeeding maybe)?
100
Jennifer RM @jenrm.bsky.social · 12/05/2026
I'd argue that these are NOT artistic choices. I think Rowling stumbled into this? She's offering a childish vision of moral goodness, even though by the last book her readers are in their late teens... The Jungian Shadow Work needed for adult morality is not on stage?!? But also: it was popular!
110
Jennifer RM @jenrm.bsky.social · 12/05/2026
Here's Gemini, summarizing Harry's Christ Maneuver. Harry, to my memory, gets no Gethsemane moment. He doesn't really pause to imagine not sacrificing himself. He's fighting The Other still. Rowling doesn't understand Solzhenitsyn. Indeed the only flaw in Harry's heart was put there by Voldie!!
Screen shot of Gemini on Harry's self sacrifice. 

Second try at generating said this (which has more plot points): 

Key Details of the Realization:

The Moment: Harry is in Dumbledore's office during the one-hour armistice in the Battle of Hogwarts, just after Snape dies.

The Information: Snape's memories show Dumbledore explaining that Voldemort must kill Harry for the protection of Lily's sacrifice to end and for the horcrux inside Harry to be destroyed.

The Decision: Harry accepts his fate, realizing he is not meant to survive, and chooses to walk into the Forbidden Forest to face Voldemort, instructing Neville Longbottom to kill the snake, Nagini.

The Outcome: Because Harry voluntarily sacrifices himself, Voldemort only destroys the soul fragment within him, not Harry himself, similar to how Lily's sacrifice protected Harry. 

Science Fiction & Fantasy Stack Exchange
 +4
100
Jennifer RM @jenrm.bsky.social · 12/05/2026
HP1 is a series of un-earned wish-granting miracles, bestowed on an "oppression Olympics winning white male orphan". He BECOMES rich and powerful! Yet he doesn't flip to picking on the Weasleys (the way Rowling herself flipped to picking on trans women). YET there's no spiritual work! Such puzzle!
130
Jennifer RM @jenrm.bsky.social · 12/05/2026
Yeah but also it's a children's book! They likely *also* (like Rowling herself?) fail to perceive the working class yearning and resentment, right? So what is it doing there?? Is it *part* of the popularity? Is it pedagogical? Do parents think it is *good* to teach class hatred to their kids?
120
Jennifer RM @jenrm.bsky.social · 12/05/2026
A thing *I* love about the first Harry Potter book is how almost literally every character is defined in terms of how "rich people are evil". Yet "I wanna be rich so so so bad" is a palpable vibe from the author herself. And there's no self awareness about this!?! It's RAW desire and resentment! 🤣😂
150
Jennifer RM @jenrm.bsky.social · 12/05/2026
What about distance from the classics, like Harry Potter and Twilight? 😇🥁😉🫠 Like what I mean, more seriously, is: what about the objectively low quality stuff with mass appeal, that every literate person SHOULD read, in order to understand allusions and jokes? Surely distance there is possible?
110