herschel

Karma: 182

herschel 4 Apr 2026 18:58 UTC
3 points
0
on: Teaching My Toddler To Read
You would likely be interested in HSRS (Hierarchical SRS). (Not my project but I know the author.) The tl;dr it’s a framework for a. representing hierarchical knowledge dependencies, including grammatical structures, b. “sampling” from points in the tree, where deps are often a class of objects, c. sequencing such samples so that material is introduced only when its deps are satisfied, and d. inferring which parts of the dep tree were likely most responsible for recollection failures and updating their review state accordingly.
My bet would be that you could probably get claude code to write most of the deck basically from where you son is up to ~late elementary school, and from there you have escape velocity. I’d be very curious if you do something like this, I suspect that early literacy edtech is heavily underoptimized etc.

herschel 21 Aug 2025 11:30 UTC
5 points
0
in reply to: Viliam’s comment on: Briefly on MAPLE, and the broader community
This feels somehow like a straw, but reflecting on it briefly it also feels like a hole in my explanation, and maybe that I’m just wrong here.
Maybe a different story I could tell would be that it’s more like “if you want, you can join us in trying to do something really hard, which has power law returns, knowing that the modal outcome is burnout and some psychological damage”, so comparable to competitive bodybuilding, or maybe classical musical training, or doing a startup. (Edit: note, Maple doesn’t include the “modal outcome is moderate psychological damage” part, though neither do the examples really.)

herschel 21 Aug 2025 2:47 UTC
5 points
0
in reply to: Lukas_Gloor’s comment on: Briefly on MAPLE, and the broader community
Thanks for an object level response!
Yup, that’s an accurate enough paraphrase.
I’ll say first, I… don’t actually endorse their model, maybe at all, but this post was to contextualize what the model even is, and that it’s maybe in principle plausible, and that their choices are made with respect to that, rather than just random-spiritual-community-is-bad-just-because-they’re-bad.
(1), people who greatly inspire others almost never started out as followers in a school for how to become inspiring (this is similar to the issues with CFAR, although I’d say it was less outlandish to assume that rationality is teachable rather than sainthood).
I think this is kind of wrong, lots of religious leaders trained within standard institutions within established traditions, lots of musicians get extensive training/coaching in all the aspects of performance besides their instrument etc. This also isn’t really a crux, because:
(2), even if you could create a bunch of particularly virtuous and x-risk-concerned individuals, the path to impact would remain non-obvious from there, since they’d neither be famous nor powerful nor particularly smart or rational or skilled, so how are they going to have an outsized impact later?
So Maple’s theory of change is not necessarily “get people enlightened, and then make sure they’re as agentic as possible”, but more like, get people enlightened, and then some combination of:
- use whatever wisdom they gain to solve technical alignment
  - (this seems mostly just silly to me)
- have them diffuse that wisdom into eg. tech culture, “purifying it from the inside-out’”
  - (again, I don’t think this is likely at all, like I said, but maybe more plausible)
- resolve the incentives of the AI race domestically and internationally… somehow

herschel 21 Aug 2025 2:29 UTC
1 point
0
in reply to: Eli Tyre’s comment on: Briefly on MAPLE, and the broader community
Er, well, it was that plus stuff like this post that Ray linked above (I didn’t necessarily remember this exactly, maybe there was just this one post.) If my gloss landed wrong I’m mostly glad to defer to you there.

herschel 21 Aug 2025 2:26 UTC
3 points
0
in reply to: Eli Tyre’s comment on: Briefly on MAPLE, and the broader community
I’m also pretty displeased with him as a philosopher and sociologist or whatever. I think when he says things like what you’re referring to, it’s both rhetorical abuse, and he’s trying to do something like evoke the “functional core” or “revealed affective posture” or maybe just “vibe”, regardless of whatever the explicit arguments are that some ideology uses. I think this is definitely the right move sometimes but I agree that Soryu kind of sucks at it, or is using mostly inappropriately.
Is this an example of my failing to meet him where he’s at?
Fair, no, obviously giving people a normal kind of fair shot is fine, though this wasn’t really what I meant. The thing that I was criticizing in that line would be more like “criticizing the output of his views without addressing the generators” (which is sorta what you’ve done here, but I’m not bothered by in context), vs “addressing the conception of value in which his claims are made.” So specifically, eg. that awakening is more valuable than ~all kinds of material wealth, that species extinction and also just killing of animals generically are a kind of extreme moral injury, etc. Those are fine to disagree with, and I disagree with them, but ignoring those kinds of cruxes makes the critiques mostly sort of useless/irrelevant.

herschel 21 Aug 2025 2:12 UTC
1 point
0
in reply to: Eli Tyre’s comment on: Briefly on MAPLE, and the broader community
(I might just be wrong here. I had some memory that after the Brent Dill affair there were some “we bungled the whole thing, actually, and people were harmed as a result” type posts on facebook.)

herschel 20 Aug 2025 4:09 UTC
1 point
0
in reply to: Gordon Seidoh Worley’s comment on: Briefly on MAPLE, and the broader community
FWIW there’s a large number of lay Buddhist teachers in the West, many of them formerly ordained, without lineage transmission. The thing that’s weird in Soryu’s case is that he’s teaching outside of his lineage, without having disrobed. My understanding is that his permission to teach was through Shinzen Young, and I’m not aware of that having been revoked. I expect rats don’t really care about lineage, but eg. Shinzen also gave permission to teach to Michael Taft, who teaches meditation in the Bay, and no one seems to be bothered by.
This is very in the weeds but I think the transmission thing mostly hardly matters, if you’re not already pilled on the tradition in its own (again, mythologized) framework here, then mostly each teacher has to independently demonstrate their reliability and authority, etc.

herschel 20 Aug 2025 4:02 UTC
3 points
2
in reply to: Said Achmiz’s comment on: Briefly on MAPLE, and the broader community
No mostly that was Vaniver having been confusing in context. Like, if you just leave out the Catholic part of the explanation it’s just “(mythologized) transmission of authority from the Buddha”, which Soryu doesn’t have and isn’t being misleading about. Even within Catholicism, monastic ordination != priestly ordination, IIRC you can’t have both, but in Orthodoxy you can, but they’re still orthogonal etc.

herschel 20 Aug 2025 3:43 UTC
8 points
1
in reply to: Said Achmiz’s comment on: Briefly on MAPLE, and the broader community
In this case neither misleading nor a lie. Ordination in Zen is generally a much lower threshold, and sometimes orthogonal from Dharma Transmission, and just requires a commitment to the tradition, not some demonstrated understanding.

herschel 20 Aug 2025 0:43 UTC
1 point
0
in reply to: Said Achmiz’s comment on: Briefly on MAPLE, and the broader community
I had that in mind, and some things people wrote after the Brent Dill stuff.

herschel 7 Nov 2024 4:01 UTC
3 points
0
in reply to: gwern’s comment on: Open Thread Fall 2024
this feels like a simplistic model of what’s going on with learning an instrument. iirc in the “principles of SR” post from 20 years ago wozniak makes a point that you essentially can’t start doing SR until you’ve already learned an item, this being obviously for purely sort of “fact” based learning. SR doesn’t apply in the way you’ve described for all of the processes of tuning, efficiency, and accuracy gains that you need for learning an instrument. my sloppy model here is that formal practice eg for music is something like priming the system to spend optimization cycles on that etc—I assume cognitive scientists claim to have actual models here which I suppose are >50% fake lol.
also, separately, professional musicians in fact do a cheap SR for old repertoire, where they practice only intermittently to keep it in memory once it’s been established.

herschel 20 Dec 2023 20:43 UTC
1 point
0
in reply to: Dagon’s comment on: What should be reified?
Yeah, I’m working towards that. Cheap examples would be like, hard sciences have made certain kinds of assumptions (reductionism, primacy of formal models etc) which have been extremely generative. There are lots of locally extremely helpful reifications, certainly various kinds of optimization criteria are very useful to generate strategies etc. A big part of my point is these should sort of always be taken as provisional.
Note that when you say “reification”, my mind replaces it with “model”, “map”, or “focus”. If you mean something else, the source of my confusion is clear.
“Focus” is the best among these but isn’t great. In most cases I think a lot of reifications are upstream of specific models, or generate them or something. Like, reification is the process (implicitly) of choosing weights for tradeoff calculations, but more generally for salience or priority etc. Maybe an ok example would be like, in economics we start trying to construct measures on the basis of which to determine etc etc., and even before these get goodharted we’ve already tried to collapse the complexity of the domain into a small number of factors to think about. We might even have destroyed a number of other important dimensions in this representation. This happens in some sense before rest of the model is constructed.
one source of my confusion may be the use of “reified” as a passive verb, which happens to ideas without specifying the actor.
This may just be sloppiness on my part. I usually mean something like, “a idea, as held by a person, which that person has reified.” Compare to eg. “a loved one” or something like that.

herschel 8 Sep 2023 23:56 UTC
4 points
2
on: herschel’s Shortform
if 👏 you 👏 don’t 👏 have 👏 gnosis 👏 of 👏 phenomenological 👏 insight 👏 then 👏 everything 👏 you 👏 say 👏 about 👏 it 👏 should 👏 be 👏 qualified 👏 as 👏 hearsay 👏 or 👏 inference
What links here?
- (In)appropriate (De)reification by herschel (20 Dec 2023 4:51 UTC; 10 points)

herschel 5 Jul 2023 3:19 UTC
3 points
0
in reply to: Dagon’s comment on: “Reification”
So part of the point is that “real” is phenomenologically very similar to “salient”, and maybe actually the same. I’m sort of on the far end in this respect but this extended usage is actually relatively common in meditation circles I’m in. I maybe could emphasize this point more...

herschel 2 Jul 2023 2:37 UTC
4 points
0
in reply to: Raemon’s comment on: My “2.9 trauma limit”
FWIW I think a lot of “somatic stuff” that’s around is either poor quality or comes with very poor pedagogy. In my model there are also subtle and (even for many teachers) commonly illegible prerequisites. Also filtration processes on students and schools, people writing books, etc.

herschel 2 Jul 2023 1:06 UTC
9 points
3
on: My “2.9 trauma limit”
Thanks, I appreciate you writing this, and I appreciate you sharing about this in public. I’m somewhat excited to see how this conversation might develop more centrally in rat discourse. I think there’s currently some major incommunicable cruxes surrounding these questions, and I’ve had a lot of conversations about this domain with people who are squarely rats, but mostly auxiliary to rat discourse in a way that’s frustrating.
I have some critiques:
- I think this is a plausible contour of how things might go for people who started out relatively healthy, who experience a substantial disturbance and then work to return to something about as stable as their previous configuration. I don’t think this is a good point of reference for people who began this process quite disturbed and have been so for quite a while.
  - Indeed, you refer to people trying to address a morass of confusing and illegible motivational/emotional/etc issues, whose situations I don’t think are described well by this sort of story. I currently have the most weight on Mark Lippmann’s work in this domain, over any other framework, traditional or modern. (Warning: I don’t actually recommend his writing to almost anyone, unfortunately.)
  - There’s some very important questions around how or why people are apparently getting stuck on this emotions/trauma stuff, and whether they should actually be regarded as stuck. This also trades off sharply depending on whether they’re doing vital work otherwise: I hold people working on x-risk in an almost unique category in this respect, where to me even if they might seem to have tons of un-worked-out stuff, I lean towards preferring they continue their work rather than investing hard in healing.
- My impression, from <20 conversations with you ever, is that you have obvious-to-me stuff, that probably remains largely untouched by the emotional work you describe having done. To be clear, this is completely blameless, and should not be a source of shame. However, from my perspective this mostly negates your judgements about what is right or advisable in this domain.
Lastly, and I think you’re not much to blame for this either:
- It’s not clear to me how we should be having these conversations in public. I’m pretty frustrated by a bunch of “Works On My Machine™” posts about this topic all over the internet (and just as well in print), and I’m pretty confused how these conversations should even work, especially in our discourse.
- There’s some problematic epistemological questions relating to claims about phenomenology and psychology, and I don’t think that the existing literatures (modern western psychology, or any of the traditional contemplative schools) have resolved them. For that matter, the techniques and models from the existing literatures mostly aren’t great either, so I’m hardly saying something like “there are established models, please go read these textbooks and stop posting quack physics.”

herschel 18 Oct 2021 4:35 UTC
7 points
0
on: The Colonization of Cults, Nonprofit Organizations, and Society
Just to nip any confusion in the bud (since this came up a couple weeks ago): My name is Herschel, I am not the author of the OP. I spent two years at the Monastic Academy, and have been around the rationalist community since ~2017. [edited to remove the author’s legal name, at their request]

herschel 14 Oct 2021 4:09 UTC
7 points
0
in reply to: HS2021’s comment on: Three enigmas at the heart of our reasoning
Apparently there’s been some confusion, I’m Herschel (I’m leaving out my last name so I’m not trivially googled), the author of this comment is a different person [I’ve removed their name for their privacy]. I was a resident for two years at MAPLE but I haven’t written anything about it publicly, besides a couple brief things on my facebook account.

herschel 24 Mar 2018 19:09 UTC
1 point
0
in reply to: FireItself’s comment on: Mythic Mode
This, please.