Sorted by New


Gauging the conscious experience of LessWrong

Could you imagine the feeling of lying on a carpet without a shirt on (ie the feeling of a carpet on your torso) ?

Somewhat... it's too diffuse. I can imagine the effect at single spots, the whole thing at once doesn't really work. (I get "glitchy partials", brief impressions flickering and jumping around, but it's not forming anything consistent / stable.)

What about a spider crawling across your hand ?

Back of the hand is manageable (it's "only" tracking of 9 points - 8 legs plus occasional abdomen contact) and it can even become "independent" and surprise me with what direction it will move in next, front is basically impossible. (Glitchy partials again.)

I am very jealous of your ability to ignore your thoughts[.]

If there's usually not much happening that reaches the conscious level, that's really not that hard. I've talked at length with people who have a near-constant internal monologue and I get that that's much harder. I just notice that it's also relatively easy for me to ignore other constant or near-constant things like loud buzzing noise (server fans?), most extreme smells (rotting/feces) where others flinch away, etc. but basically impossible to ignore constantly changing impressions. (I currently have a construction site in front of the house and I can't work at all while they're active, even with earplugs.)

[I am very jealous of your ability to] track north.

Again, I have to constantly keep an "inner eye" on that or it comes loose and then it's broken / de-synced. A question from someone that makes me think just a little bit too much can be enough to break it, unless I remember to explicitly save the current heading and reinstate the thing afterwards. (I suspect you can do/imagine something similar and it'll get less costly and more precise over time of using it.)

I wonder if there's any tangible overlap between brain function and fields of interest.

Probably - you're more likely to do stuff that's easier for you, which makes it easier... But I expect that to be fairly weak / to have lots of noise on top. (External expectations, stuff that you don't even know exists, etc. etc.)

I guess what also plays a big role is how you approach your brain / how you model it & yourself / what expectations you place on it. I treat mine as a substrate that can spawn lots of independent processes of varying size and with more or less expressive interfaces, and "I" just happen to be one that's fairly big and stable. So making that arrow pointing north is basically just me spawning yet another small process on that substrate that subsequently does its own thing (unless swapped out because of capacity constraints), likewise the imagined spider can become independent and choose its own direction, because I fully expect it to be an independent process and not something that I have to manually control.

(Treating your brain as some sort of "expectation realizer" seems to be a powerful model/perspective, but totally expect really weird "replies" when you try to apply that to external stuff. (Like weird feelings, sudden anxiety, strange preferences, etc. guiding you in the direction that your brain thinks is best - be careful (and keep track of) what you wish for.) Internal seems to be relatively safe compared to that.)

Gauging the conscious experience of LessWrong

Also one of my friends struggles with verbal thinking and thinks mostly implicitly, using concepts, if I understood that correctly, and they have a strong preference for non-verbal signs of affection (physical contact, actions, quality time etc.).

Same here. Not thinking in words at all, very strong preference for touch or very simple expressions. Over the years with my SO, we basically formed a language of taps, hugs, noises, licks, sniffs, ... (E. g. shlip tongue noise - Greetings! / I like you. / ... (there are even tonal variations - rising / higher pitch is questioning, flat is affirmative), sniff-sniff noise - What's wrong? Etc. - So a possible exchange could be seeing them sitting on the sofa, looking unhappy - sniff-sniff (What's wrong?) - grumble (Bad mood.) - shlíp? (Affection?) - shlīp. (Yes.) - hug. That's much less exhausting than doing the same in words.)

Gauging the conscious experience of LessWrong

Some more details on each of the categories in order:

Visual - I don't really see things, I just get some weird topological-ish representation. E.g. if I try to imagine a cube, it's more like the grid of a cube / wireframe instead of a real object, and it's really stretchy-bendy and can sometimes wobble around or deform on its own. And attributes like red / a letter printed on a side etc. are not necessarily part of the face but often just floating "labels" connected by a (different kind of) line that goes "sideways" out of 3-space? o.O Even real objects like a tree are abstracted (and it depends on what I focus on), e. g. it's more like a cylinder with a "fuzzy sphere" on top and a "fuzzy cone" below. Or maybe I focus on the bark, then that gets more detail but everything else becomes even less defined. Or maybe it's just general "tree-ness", then there's just some elongated blob and I can't even query whether it has sharp corners/edges or not without it changing shape to become more defined (and then it could end up either way, so no clue).)

Sound - sometimes, I get runaway "earworms" ("stuck tune" doesn't really fit) that morph over time and evolve away from the source material. Sometimes there's two of them at the same time, fighting each other or working in harmony. (If that's (the illusion of) two full orchestras, it can get quite exhausting...) I lack the music education / knowledge to write down or play what I hear.

Taste - negligible, mostly on the "floating tags" level. "Here be salty."

Smell - was on the same level, but getting a lot better now that I consciously care about smell in real life.

Touch - mostly negligible but sometimes I can cook up scary real stuff. (Like imagining a friend biting my neck and hairs raising up and also noticing how the muscles around that spot would move if that were real, even if I didn't consciously follow that before, and then later validating the memory of that against the real thing. That was so surprising that I still remember it.) I find it a lot harder to imagine touch on extremities than the torso. E.g. I can't really imagine a stubbed toe, the feeling of walking across carpet, grabbing a cabbage with my hand, but I can vividly imagine a drop of water running down my chest or a spider crawling across it. (Lower resolution is easier to fake?)

Internal monologue / thinking in words - I can't imagine how that would be. Even when talking, I don't really know (on the word level) what I'm going to say before I actually say it. I generally don't think in words and have a hard time translating back and forth from words into my internal representation and back out. There's lots of concepts that I use that I can't really describe because they're several abstraction levels up from stuff that maps to known external words, so I'd have to explain/translate several layers of wordless stuff before I could even describe the perspective from which these concepts arise. (And the "floating tags" mentioned above also aren't words but just "nodes" that somehow represent the stuff.)

Memory - no clue how to assess that, I don't have a baseline to compare against. Some stuff gets regularly forgotten (names, anyone? .), other stuff (like small random details) just sticks around far longer than justified. (I also tend to notice lots of weird details.)

Another potentially notable thing about memory: I can't randomly access or scan memories, but if I have a specific query I tend to get results, including context (e.g. why this memory is probably trustworthy, stuff that helps locate it in space/time etc. etc.)

Mind control - I can't stop an ongoing battle of two orchestras or other things that make themselves hard to ignore, but for stuff that's not as overpowering, I can mostly just ignore it and it's gone?

Synesthesia - I think sometimes I get whiffs of it, like 3 smelling slightly cake-y and 7 rather bloody / putting a metallic taste in the back of my mouth, but it's rare / hard to notice. (Might be imagined, but the associations are consistent.) I also get some angle-dependent coloration on narrow grids, not sure what that is. (I know that you can supposedly trigger that effect by looking at training images and it'll persist for a long time, but I didn't (knowingly) look at any similar images, and also it's doing a full 360° / full color cycle for me vs. that effect only getting you like 2 or 3 colors IIRC?)

Some other stuff:

I can feel even pretty small heat sources across fairly large distances, I notice the exaggerated impression / it presents as a tingling sensation on my face, or my hands if I'm searching with them (in which case I also notice it along my arms unless they're covered). E.g. I can locate a small desk lamp that's 8°C above room temperature from the other side of the room with my eyes closed and after the lamp was turned off (so I don't react to any stray light), or notice on what side of a sofa a human has been sitting on.

Several years ago I did an experiment where I tried to do as much as possible with my eyes closed for the better part of a week. That seriously improved my ability to navigate with eyes closed. I unlearned a lot of that, but I still like to e. g. shower with eyes closed because it reduces the amount of sensory data coming in. Sometimes, when I'm stressed / close to sensory overload, closing my eyes doesn't help and the falling drops of water create a fairly stable "image" of the bathtub / wall / curtain. (It's like every drop's impact creates a small gray circle that fades over ~0.2-0.5 seconds and that's approximately aligned with the normal of the surface. Together, from all the drops, I get something resembling a picture of the surroundings. The dots are relative to my head, not the external world, so if I move my head, they move along and I get garbage until the old dots fade and things stabilize again. Aaaaaaaaaa exhausting.)

I generally have trouble navigating unknown areas, but if I reserve a few brain cycles to manually track "north" (or any direction really), that persists and I can navigate relative to that. (The marker that I use is like an "arrowless arrow", a "shapeless shape"? It's really weird trying to describe it. It's pointy and has a clear designated "pointy end" but it's not really "physical" (even in mind space). It generally floats above my head and can turn really quickly - I can jump around, whirl in place, ... and it just stays glued to the target.) However, as soon as I forget to keep it active (even just for a moment) it comes loose and no longer updates relative to my movements. "Keeping it active" isn't "thinking at it" but just observing it do its thing, don't really know how to describe that. It also doesn't work when I'm being moved (cars, trains, ...), it seems like my guts (body location-wise) do the tracking and they don't get (all) slow rotations.

When I do mathematical proofs, I don't really think, I just "feel" my way through the tree of possible derivations, pick a step, see where I end up, repeat. If that doesn't work, I get stuck. I lack the experience with manually/consciously enumerating the top layers (or this combinatorial stuff is just really exhausting for me?), and so I can basically just wait a few days and try again until I get some progress.

The same "either it works intuitively or not at all" applies to a lot of things that I do, from other technical stuff like coding to things like cooking. Usually whatever I cook tastes good to great, but sometimes it comes out weird and then I have no clue how to fix it. (That's getting better, I'm slowly accumulating things to check / try - like aim for 0.5-1% salt (saliva is around 0.4% and anything below tends to taste weird), check if it needs more acid, add a bit of sugar, etc., but there's still a big difference between things just working and something being not quite right.)

Why are delicious biscuits obscure?

I suspect the water content of honey/treacle (estimating 15-20%) will lead to more gluten formation, which risks causing a chewy instead of crumbly texture. (If you're not adding any water at all, you're not getting gluten strands.) Butter also has some water (around 15%), so you generally don't knead these kinds of dough for long. (Same goes for shortbread, scones, ...)

Hence, I guess any flour should do if you know how to handle it / are careful not to overwork the dough.

Final Version Perfected: An Underused Execution Algorithm

Apropos SRS and needing a second list: I'm currently experimenting with using Anki as a to-do tool. (Every task becomes a card, I "learn" the deck, and for each task get the choice of Again/Hard/Good/Easy with listed intervals, e.g. <15min, 1d, 6d, 27d. If I want to see the task again quickly (Again/Hard), I have to spend at least 5 minutes on it before clicking the button. Otherwise, it's ok to bury it by clicking 'Easy' on it.)

This deals nicely with "nice-to-haves" that I don't want to kill yet. Whenever they come up, I just send them away to next month/quarter/year/... - that way, I don't ever have to make the hard decision to kill the idea, I could always change my mind about this later, because I know that I will be reminded eventually. These "non-tasks" are also eating a fraction of the mental capacity that they'd otherwise require, because (a) I don't get a growing list of "things that would be nice to do eventually" that I'd have to manually comb through (instead every couple of days there will be 1 or 2 of them mixed in with the review), and (b) I know that they'll pop back up eventually, so I don't have to worry about keeping track of them or finding them again if I ever change my mind. (To a lesser degree, it also forces me to start early on tasks due in several months, to keep the interval growth in check.)

The only problem I had so far was tasks that are inactionable right now (e.g. taking out the trash at 5am, because noisy) needed to be 'skipped' and reviewed later, leading to "ugh not in the mood right now" skips of actually actionable tasks, leading to less stuff getting done.

FVP looks like it might fix this. So for the month of December I'll be running the following modified algorithm:

  1. Go through the list of tasks in Anki, for each either do it right away or put it on the (paper) list, or delegate it to future-me (i.e. pick the long interval Good/Easy option).
  2. Do FVP on the paper list.
  3. 3-strikes rule for "list leftovers": If I say I do it but don't get around to it, the counter increases. Once it hits three, I must "delegate to future me". (Counter resets whenever I explicitly declare not to work on it, not sure yet if it should also reduce for streaks... For the experiment, I'll go with yes and 3-days streaks for removing one mark.)

Let's see how long I'll stick to this.

The New Frontpage Design & Opening Tag Creation!

Some (fairly low-priority) glitches / feature requests surrounding tag filtering, in decreasing order of perceived importance.

Front page filtering / tag order: Request: Currently they're in order of being added, with no way to reorder. I'm already getting confused by that… Manual reordering might be nice, but just some imposed order (most likely name or weight adjustment) would be better than the current state.

Front page filtering / adjusting tag weights in the popup: Glitch: When changing the tag weights via the up/down arrows next to the 'other' field, the size of element(s) to the left changes when reaching/leaving one of the pre-defined values, which changes where your clicks go. (Worst case, Firefox re-layouted such that the second click went to the 'remove tag' button… oops.) Ideally, the buttons stay where they are so I don't have to (remember to) wait nor reach for the keyboard and manually compute the adjustment in my head, and instead can just click a few times.

All tags page: Request: Mark newly added / edited tags in some way. (I went through the whole list to add tags that I want to track. New tags are still being created all the time. Currently, it seems the only way to find all those new/updated tags is to go through the whole list again.)

Site Redesign Feedback Requested

I'd actually prefer to have a true dark mode, which probably won't be coming anytime soon… (unless the team steals the dark/light mode stuff from Gwern maybe?)

Luckily, user style sheets are a thing! There's a dark mode style sheet from 2018 on userstyles.org and thanks to its extreme simplicity, I'm happy to report that it still works on the new site design. (Unfortunately, that simplicity also means that it's not very good… c'est la vie. Maybe I'll make a better one when some urgent deadline approaches some day in the following months, maybe not.)

If you prefer white, setting this as your user style sheet might get you 80%-95% of the way there:

.PostsItem2-background {
    background: #f4f4f4;
body, .Layout-main {
    background: white;
.PostsItem2-bottomBorder {
    border-bottom: solid 2px white;

(You may want to fiddle some more with the post / comment borders, other than that I noticed no problems.)

Of course, I'm not sure how stable those names are… So here's some actual feedback / questions on that side of the design:

  • Some CSS classes do seemingly unnecessary things (e.g. .PostsItem2-isRead overrides the background color of the normal .PostsItem2-background but applies the exact same value, same for .Layout-main / body), which makes it potentially harder to customize. (Because !important in the user style overrides both classes it actually still works fine for PostsItem2-isRead, but .Layout-main needs an explicit extra rule and just deleting that background color override in the original CSS doesn't seem to break anything.) Do you want a list of weird spots like that? (If yes, I'll make a crazy style in 1-2 weeks and see what breaks when modding / what can (probably) safely go away.)
  • stable names for most things would be really useful, but PostsItem2 looks fairly unstable / generated? How stable will those names be?

Other than that, I agree with the "looks good"!

__nobody's Shortform

Observation: It should generally be safe to forbid non-termination when searching for programs/algorithms.

In practice, all useful algorithms terminate: If you know that you're dealing with a semi-decidable thing and doing serious work, you'll either (a) add a hard cutoff, or (b) structure the algorithm into a bounded step function and a controller that decides whether or not to run for another step. That transformation is not adding significant overhead size-wise, so you're bound to find a terminating algorithm "near" a non-terminating one!

Sure, that slightly changes the interface – it's now allowed to abort with "don't know", but that's a transformation that you likely would have applied anyway. Even if you consider that a drawback, not having to deal with potentially non-terminating programs / being able to use a description format that cannot represent non-terminating forms should more than make up for that.

(I just noticed this while thinking about how to best write something in Coq (and deciding on termination by "fuel limit"), after AABoyles' shortform on logical causal isolation with its tragically simple bit-flip search had recently made me think about program enumeration again…)

Baking is Not a Ritual

The key point is this: The big difference between baking and cooking is that baking involves much more chemistry than cooking, which means that altering the recipe without understanding what you're doing is much more likely to result in failure. (Bad substitutions/ratios in cooking means the result might taste/look a bit strange, but overall it'll likely be fairly close to the original. Bad substitutions/ratios in baking means you probably get a brick / dust / other unidentifiable garbage.)

Thus, if people approach baking like cooking, they probably fail. Repeatedly. Hence the ritual thinking.

What are objects that have made your life better?

Erasable gel ink pens in lots of different colors.

Working on paper still beats tablets etc. sometimes, and instead of crossing out stuff and trying again, you erase and overwrite – looks much cleaner, even if this was your very first draft / rough diagram / whatever. Instead of copying / re-writing the whole thing a half-dozen times or more to get a final clean version, you copy maybe once, at most twice, often not at all. (You just erase small mistakes that happen while making the clean version, instead of starting over yet again.)

Beats colored pencils by a wide margin, both in handling when writing/drawing as well as in ease of erasure. (The ink becomes completely invisible when heated, no need to scrape pigments out of the crevices of the paper / abrade the paper surface.)

Muji (the "Japanese Ikea") had great ones, but they got rid of most colors (no more green/cyan/purple/…, only black/red/blue.) Luckily, lots of others are producing them now, so I can get new ones when mine (and their refills) finally run out.

My only warning: If you're writing double-sided in a notebook with thin paper, don't be too vigorous when erasing. Normal corrections are no problem, but taking out a whole shaded diagram might also erase parts on the back. Other than that, while I'm not sure how long-term stable these inks are, my 5+ year old notes still look fresh. (I still made backup photos just in case…)

Load More