thanks!
LoganStrohl
A note on illustrations:
Somebody brought up that their friend assumed my illustrations are AI generated. So I want to clarify: With the exception of the two dancers in “Spaciousness In Partner Dance” (AI generated) and the spider web in “On Realness” (commissioned from Theresa Strohl, my mother), I’ve painted all the illustrations by hand myself. Duncan Sabien has edited them slightly to make them work with LW’s site background.
(Designated Buddhism thread)
I ordinarily do not allow discussions of Buddhism on my posts because I hate moderating them. I haven’t worked out what exactly it is about Buddhism, but it seems to cause things to go wonky in a way that’s sort of similar to politics.
Also, my way of thinking and writing and doing things in general seems to bring out a lot of people who want to talk about Buddhism, and I want my work discussed mostly on its own terms, without it being immediately embroiled in whatever thing it is that tends to happen when people start talking about Buddhism.
One of my moderation rules forbids discussion of Buddhism by default.
Since there was a big old section on meditation in this post, and the type of meditation I described is pretty specifically shikantaza from Soto Zen, I’m designating this here thread as the place where people can talk about Buddhism-related stuff if they want to, just this once, as a treat.
I don’t promise to participate. My other moderation rules still apply.
I have thought fondly of this post several times since I read it.
This post helped me relate to my own work better. I feel less confused about what’s going on with the differences between my own working pace and the pace of many around me. I am obviously more like a 10,000 day monk than a 10 day monk, and I should think and plan accordingly.
Partly because I read this post, I spend frewer resources frantically trying to show off a Marketable Product(TM) as quickly as possible (“How can I make a Unit out of this for the Workshop next month?”), and I spend more resources aiming for the progress I actually think would be valuable (“In the world where I have robustly solved X one year from now, what happened in the intervening twelve months?”).
Outside of academia (or perhaps even inside of it, at this point), our society does not really have a place for monks of the larger magnitudes, so it’s uncomfortable to try to be one. But if I’m going to try to be one, which I absolutely am, it’s awfully helpful to be able to recognize that as what I’m doing. It impacts how I structure my research and writing projects. It impacts how I ask for funding. It impacts how I communicate about priorities and boundaries (“I’m not scheduling meetings this quarter.”)I plot my largest project on a multi-decade timescale, and although there are reasons I’m concerned about this, “lots of other people don’t seem to commit to such things” is no longer among them.
(This is a review of the entire sequence.)
On the day when I first conceived of this sequence, my room was covered in giant graph paper sticky notes. The walls, the windows, the dressers, the floor. Sticky pads everywhere, and every one of them packed with word clouds and doodles in messy bold marker.
My world is rich. The grain of the wood on the desk in front of me, the slightly raw sensation inside my nostrils that brightens each time I inhale, the pressure of my search for words as I write that rises up through my chest and makes my brain feel like it’s breathing through a straw. I know as well as almost anybody what MacNeice called “the drunkenness of things being various”, “incorrigibly plural”. I am awash in details; sometimes I swim, sometimes I drown, and in rare merciful moments, I float.
People talk about missing the forest for the trees; I am a creature of individual leaves. The sticky notes with which I had covered my walls were my attempts to recall every twig and branch I had seen while developing my approach to rationality, ever since I asked myself what the existing art is missing back in 2013. Each page was an attempted portrait of a different tree.
The sentence I somehow pulled together for this sequence—”Knowing the territory takes patient and direct observation”—was my sketch of the entire forest all at once.On that day, it had seemed a literally incomprehensible pile of details, as nearly everything I write about does until some time after I’ve published. Yet after two more years of work on this project, I still think that sketch is not only accurate, but pretty close to complete.
I am proud of this sequence. It’s far from perfect; it’s far from adequate, in fact. And I’ll talk about that, too. But as a first-pass summary of how I think about “Intro to Naturalism”, it’s right to say that overall, I think it may be the best thing I’ve done so far.
*I doubt it’s worth much on its own, though. It was really never meant to be. I tried to make it accessible, but I mostly wrote it for myself. I published it publicly anyway because I figure there’s a (reasonable!) limit to the patience of my funders. I’m delighted and a little surprised that other people have found it useful.
To me, this sequence is a bit like a sextant. Suppose you’re trying to navigate to a particular island off the coast of South Africa. It will not, by itself, get you to your destination. It’s not the shore. It’s not a boat. It’s not even a map. You need an awful lot more than this to sail to Madagascar.
But as the captain of the HMS Naturalism, I felt I had no hope of staying on course without writing down this worldview, in summary and in detail.
*
On course toward what, exactly? What is the project of which this sequence is a small but crucial part?
The journey has four parts, according to my current understanding:1) Intro to Naturalism: My attempt to lay out the perspective from which my branch of rationality is practiced.
2) The Nuts and Bolts of Naturalism (published in early 2023): A straightforward mechanical description of the procedure, as I tend to present it to people learning it for the first time.
3) Naturalism In Practice: A series of accounts of real-life naturalist studies, in frequent dialog with posts from the earlier two sequences, covering a range of topics and demonstrating what naturalism looks and feels like in practice. (This sequence is currently in the works.)
4) THE ACTUAL GOAL [as yet untitled (and unfunded)]: A synthesis of the previous three sequences, perhaps in book form, comprising a comprehensive practical guide to knowing the territory through patient and direct observation.
(There may need to be a part “3.5”, where I refocus for a while on pedagogy and collaboration, before I am ready for 4.)
If I were less awash in details, I imagine I would have been able to start with part 4. But also, I may never have been able to develop such a thing as naturalism.
*
What can I predict about how this sequence will show up in whatever synthesis I eventually create, provided I eventually get to do that?
I think the core summary will be the same: “Knowing the territory takes patient and direct observation.”
Here are some ways I think it will differ from the original.
1.
I may rely much less on the aesthetics of 19th century natural history; indeed, I may completely rename the discipline.
This framing did a ton of work in helping me understand what I was doing and why; but most people do not have my background, and do not find this framework nearly as supportive or inspiring as I have.
For most readers of LessWrong, the word “naturalism” refers to an ontological claim denying the supernatural. They are completely unfamiliar with the approach to biology that also goes by that name, which focuses on knowing a few organisms deeply rather than on the categorization of organisms.
If I do keep the naturalist framing, I’ll need to double down on it and begin with a discussion of the role of the naturalist approach in the history of science, which would be fun for me but perhaps needlessly inefficient.
2.
If there is a major change to the overall summary, it will probably be a result of further developments in my study of “realness”.
During the study I’m currently writing up—one of “Hug the Query”—the primacy of my intuitions involving these observations has come into sharper focus. I’m not the only person who believes “Interlude On Realness” is the most important post in my intro sequence, even though it’s also the one that’s least integrated with the rest of the sequence. I think that a more mature and positively impactful incarnation of “Intro To Naturalism” might put whatever’s going on with “realness” front and center (presumably after I manage to have more coherent thoughts about what is going on with realness).
3.
As Intro to Naturalism suggests, this approach to investigation is in theory extremely general. I have an even larger vision than the one I’ve so far laid out in this review, in which my approach is thoroughly tested and adapted to an enormous breadth of domains, from AI alignment research to metallurgy to computational ethnomusicology. (I maintain an intuition that’s even a little gears-y about the utility of naturalism for AI alignment research especially.)
However, I am by passion and profession a rationality developer. The version of naturalism that I understand best, and that I am best prepared to write about at length, is particularly tailored to the investigation of human cognitive algorithms. Unless someone swoops in and drops a bunch of money on a far more ambitious and speculative project than I currently plan to undertake, and perhaps even finds me some kind of cofounder with a complementary skillset, the final incarnation of this sequence will be more narrowly focused on rationality in particular than was the original.
4.
The structure of the finished work will almost certainly be the exact opposite of my historical publications. Historically, I published the most abstract discussions, then the instructional guidelines, then the fully concrete demonstrations. I did this not because I thought it was a good idea, but because I’m a tiny human with limited cognitive capacity and it was the only way I could manage to write anything in practice.
The almost-always-correct way to write is to move from concrete to abstract. I expect that anything I present from this sequence in the final work will follow a demonstration and a methodological discussion.
5.
I do not think that I had “patient” in sufficiently clear view when I wrote “Patient Observation”, and I still don’t think I’m quite there yet. It may be the wrong term, or it may be overloaded. It’s terribly important, and I think that communicating about it as well as I’d like to will require 1) breaking it down more carefully, and 2) doing so in dialog with contrasting approaches.
I recently found myself claiming to be “at war with the relatively dumb versions of startup culture aesthetics”. I think that to make my point about “patience” to my satisfaction, I will also need to extol the virtues of efficiency, rapid iteration, decisiveness, etc., as I understand them. This entire approach depends fundamentally on patience, and I don’t think it can be wholeheartedly embraced without first safeguarding the fruits of contrasting approaches (or at least explicitly contending with their loss, where they are in fact threatened).
>A visualization where a hose of heavy running water enters at the top of your head and pours out through the pads of your hands results in a pretty solid frame for lateral.
hm i’ve never heard that one! i’ll try it out, thanks!
if you wanna second-guess yourself even harder,
1) look around the room and attempt to produce three instances of something resembling tiny quiet confusion (or louder than that if it’s available)
2) try to precisely describe the difference between surprise and confusion
3) sketch a taxonomy of confusing experiences and then ask yourself what you might be missing
I feel embarrassed that I’m just now reading this. >_< ′ (Ray knows but: I’m the aforementioned “Brienne Yudkowsky”, my name’s just different now.) I enjoyed it; it’s really interesting and valuable to see my thoughts contextualized from the outside and narrativized. It’s usually hard for me to see forests when I’m surrounded by trees.
> There are very few opportunities to practice noticing confusion.
I’m really curious how you relate to this claim six years later.
I wrote up “How To Think Of Things” for CFAR a while back. I probably wanna at least edit it some before making it a top level post, but I’m curious what you think of it.
“What did pregnancy do to your cognition?”
(Interested in responses to this from other people who have been pregnant, but here’s my own answer.)
I think the main thing pregnancy seemed to do to my mind was reduce my associative speed. This had all kinds of effects on the rest of my cognition and experience, because it’s a capacity I rely on almost constantly, but I think this was the central mechanism.I’m not sure I have my concepts carved up right here, but by “associative speed” I mean “the thing that lets your thoughts go far and fast during a babble challenge”. During pregnancy I’d try to do a task like “What does the smell of this chocolate make me think of?” and nothing would come to me for ages and ages (by which I mean a full one to three seconds), and then tiny bits of things would trickle in, but with no vibrancy or motion, no suggestion of more thoughts coming on their tails.
At the height of my mnemonics training, when I was super buff in raw creativity muscles, I’d try something like that and it was an almost overwhelming flood of life-like imaginings that felt effortless, almost like closing your eyes on mushrooms. My brain on pregnancy was the opposite of that, and it felt like death.
I was terrified that my associative speed would stay that low forever. Six weeks postpartum, it’s not quite back to normal yet, but I think it’s close.
>They’re… free? Nothing bad happens when you generate them. You ignore them and move on and consolidate the good ideas later.
I understood BenWr to be suggesting this was false. His pruner is rejecting “bad ideas” for a reason, and perhaps it is a good reason; perhaps bad things do happen if he deliberately lets in more “bad ideas”.
If that were true for people in general, or for a significant minority of people, I’d definitely want to understand what the bad thing is, how it works, whether “having bad ideas” tends to be good on net anyway, and how to mitigate the bad thing if so.
I do think that lots of people—at least 85% of people, in my experiences running this kind of exercise with others—experience some kind of pain or suffering when “trying to have bad ideas”, at least at first. (I did a series of mnemonics workshops before I even started using this kind of thing in rationality training, so n is somewhere around… 350?)
It has always appeared to me that the painful parts of the experience are coming from a combination of “doing new things is hard”, “doing things I’ve trained myself not to do is uncomfortable”, and “social image-based stuff like ‘what if people see this and think I’m bad’ or ’what if I see this and think I’m bad”. All of these concerns are important to address in some way, I claim, for a person to get really good at this. I haven’t actually seen anybody investigate what’s going on for them and then decide that they do not want to gain the skillset. (There certainly are people who decide not to use negatively-valenced emotions when committing things to memory, and who decide to keep their “thinking like a villain” knob turned down pretty low, and these decisions seem similar to “try not to have bad ideas”; but I think they’re not dealbreakers for the central skill, and I think “try not to have bad ideas” probably is.)
However, I think I was much, much worse ten years ago at making space for the people I’m teaching to find their own way of doing things. So maybe if I ran mnemonics workshops today, many more people would pipe up to be like “You know what? This is bad for me. No thank you.”
I don’t know what it says about me that
“Eat it then eat lots of beans then fart while in a handstand.”
was the fourth thing I thought of. Wtf brain.
I started out with the procedure I describe here, as a warmup. I got to number 11 in the first three minutes, then when I started the second half of the procedure I just kept going.
This list took me about 30 minutes, so it’s probably not the same as “the best 50 ideas I can come up with in an hour”. If I were going to do another 30 minutes to make a better list, I think I’d highlight my favorite ideas so far, ask myself what it was like to come up with those ones in particular, and try to adopt more of whatever mental postures those are for the rest of the time. I expect I’d have fewer ideas in the subsequent half hour, but they’d probably be more to my liking, on average.
A different thing I think I could do with that second half hour to make a better list would be to pick several of the items from the first list that seem like they could use further development, perhaps because they have an obvious practical flaw (such as “but there’s no air between the Earth and the moon!”) and take them as prompts, each for three to five minutes.
If I wanted to just explode this list into way more ideas that are all over the place, I’d try the grid method I describe at the bottom of the document I linked above.
Put it on a rocket and light the fuel.
Use a big catapult.
Give it to a gigantic bird.
Eat it then eat lots of beans and fart while in a handstand.
Put it on the hand of a clock, then speed time up a whole bunch so it’s flug really fast off the end.
Hot air balloon.
Blimp.
Hold it and jump really high.
Go around back of the moon and shove the moon into the Earth.
Use an airplane.
Throw it hard with your arm.
Put something really heavy at the end of a slide, put that slide end on the moon, and send the thing down the slide.
Smash an asteroid into the Earth so the part of Earth the object’s on breaks off and smashes into the moon.
Use a portal gun and shove the thing through the portal.
Use a shrink ray so the thing’s so small that… something something particle physics, waveforms, everywhere all at once.
Locate the parallel dimension where the thing’s already on the moon and go there.
Drop something really heavy on the other end of a teeter totter.
Trebuchet.
Fire it from a big gun.
Gigantic bow and arrow.
Use one of those toys with the yoyo thing on rails until it goes so fast it’s flung off the end all the way to the moon.
Eat it, then eat a super poisonous mushroom, then vomit upward.
Send it to exactly the right star then explode the star in such a way that the object is propelled toward the moon.
Put it on a comet that’s headed for the moon.
Put it inside of a soccer ball and kick it really hard.
Pick it up off of the ground way harder than you meant to.
Put it in a Chalmers book then make me read the Chalmers book.
Drop it on the best trampoline ever.
Put it on a ferris wheel that spins really super fast.
Locate a time when it’s already there, then time travel to that time.
Put it on a barbell and give the barbell to one of those guys who drops barbells really loudly at the gym to show off, but dose him with insecurity juice so he drops the barbell so hard it goes all the way through the center of the Earth and out the other side and to the moon.
Tie it to the foot of a gymnast before she does a super flippy flip so she flings it off of the end of her foot and to the moon.
Give the idea to Atrus or one of the other D’ni Myst characters and ask them to write a detailed description of it being on the moon.
Put it in the straw of my water bottle, seal the lid on, and send the bottle from sea level to the top of a tall mountain really fast so the pressure change launches the object to the moon.
Go find a bit of moon rock and put the object on it.
Make a ladder from the Earth to the moon and climb the ladder while holding the object.
Elevator to the moon.
Escalator to the moon.
Staircase to the moon.
Float it on the surface of a pot of water, then heat the water super hot super quickly so it’s exploded off the surface.
Sneak it into an astronaut’s luggage.
Give it to a USPS mail carrier with a moon address and a shittton of stamps.
Put it in an elephant’s trunk, then give him sneezing powder.
Put it in the ocean, then make an enormous wave.
Shove it in a super duper volcano then block up all the other volcanoes so there’s only one way for the pressure to escape.
Put it in a car and tell Yahoo Maps to send it to the bottom of the ocean.
Give it to GPT5 and ask the robot to send it to the moon.
Go back in time to retrieve DaVinci and give him a million dollars to send it to the moon.
Put it on the negative pole of a gigantic magnet and shove the positive pole of another gigantic magnet against it.
Put it on one side of a pancake and flip the pancake way too hard.
Jacob I like this post and I had a good time. Thanks :)
I’m really happy to hear you tried this! Thanks for telling us about it.
>it seems pretty obviously connected to me
I’m curious what happens when you try to spell out why it’s connected.
I think it was something like three to five out of 75 people (so like 5%).
Two of the three people I’m thinking of didn’t tell me all that much detail. Most of my model of what’s going on at least some of the time comes from talking in more depth with just one of them. That’s nowhere near enough information to make any remotely confident generalized claims; but it did seem like enough to include a note of caution.
I think most of the people likely to run into this kind of trouble are autistic. According to my model (which is roughly the “weak central coherence” theory), autistic people are dealing with way more sensory information most of the time, because their top-down processing is relatively weak compared to their bottom-up processing. They’re not pruning stuff like normal. It just hits them all at once, and they can’t organize it.
(I don’t know why I’m saying “them” as though I’m not such a person.)
Departing now from standard stories about how autism works, and veering into my own speculation.
It seems to me that autistics tend to choose one of two strategies for coping with this. (I’m using “choose” very loosely here. It might happen when we’re two years old.) Either we let everything in and become the “primarily sensory sensitive” flavor of autistic, or we dissociate and become the “primarily sensory insensitive” flavor of autistic. (It’s more complicated than that; most of us don’t fall cleanly into one category or the other in every circumstance.)
Some of us freak out when an ambulance goes by, and can’t think straight when there’s a coke can in the same room because it keeps making tiny bubble sounds, and can tell you in great detail about the sensations happening on every inch of skin. Like me.
Others of us are hardly ever aware that our bodies exist, may not notice you’re calling our name when we are standing right next to you, and spend most of our time “with our heads in the clouds”. (IME the clouds are often programming, math, or writing fiction.)According to my story, sensory insensitive autistics have learned to live whatever-kind-of-life-they-have while constantly ignoring almost everything that’s happening to them. They’re not the only sort of person like this; plenty of neurotypicals also have their heads in the clouds almost all of the time. The thing is, head-in-the-clouds neurotypicals are still pretty much fine if they let through a bit more bottom-up data. It’s not what they prefer or are comfortable with, but their basic way of processing information does not rely on never doing this. It doesn’t threaten to break them.
Sensory insensitive autistics, though, depend on this extreme strategy for basic survival. When guided to make a move that would let in the flood—the flood that I constantly swim in, but that they have no practice coping with head-on—they can immediately tell that they’re in danger of drowning, and they go NOPE, NO THANKS, DO NOT LIKE THIS.
Ah, so would I! I think I never actually got to see more info on the outcome. I don’t know whether or not anything was actually compiled. Once it was in Leverage’s hands, I guess I lost track of it somehow, but I don’t remember why.