All of Pablo Repetto's Comments + Replies

Early COVID response on LW was a generalized "this is a big deal." I can't find the post that originally caught my eye, but I remember hitting the supermarkets in Buenos Aires, stocking up on masks and hand sanitizer, and two weeks later seeing the city freak the hell out. Jacob's "Seeing the Smoke" was a strong early signal, and Zvi's updates often considered explicit numbers.

I'm glad you like it!

Fixed the footnotes. They were there at the end, but unlinked. Some mixup when switching between LW's Markdown and Docs-style editor, most likely.

I see... so trolling by patenting something akin to convolutional neural networks wouldn't work because you can't tell what's powering a service unless the company building it tells you.

Maybe something on the lines of "service that does automatic text translation" or "car that drives itself" (obviously not these, since a patent with so much prior art would never get granted) would be a thing that you could fight over?

1Dave Orr2y
Certainly if you can predict applications, then you can do as-applied patents. I'm not sure that MIRI or whoever has any particular advantage in predicting applications. Also, what you can get with a patent is mostly licensing fees. If you try to stop someone from using something, you're looking at  years of litigation. In a fast takeoff scenario that doesn't actually get you anything -- by the time litigation is over, so is the game.

You're welcome! I'd like hearing a bit about how it helped, if you are ok with sharing.

Hi! I wrote a summary with some of my thoughts in this post as part of an ongoing effort to stop sucking at researching stuff. This article was a big help, thank you!

I'm glad you enjoyed it! I agree that more should be done. Just listing the specific search advice on the new table of contents would help a lot.

I'm gonna do the work, I promise. I'm just working up the nerve. Saying, in effect, "this experienced professional should have done his work better, let me show you how" is scary as balls.

First of all: thank you for setting up the problem, I had lots of fun!

This one reminded me a lot of D&D.Sci 1, in that the main difficulty I encountered was the curse of dimensionality. The space had lots of dimensions so I was data-starved when considering complex hypotheses (performance of individual decks, for instance). Contrast with Voyages of the Grey Swan, where the main difficulty is that broad chunks of the data are explicitly censored.

I also noticed that I'm getting less out of active competitions than I was from the archived posts. I'm so co... (read more)

Thank you for doing the work of correcting this usage; precision in language matters.

I made some progress (right in the nick of time) by...

Massaging the data into a table of every deck we've seen, and whether the deck won its match or lost it (the code is long and boring, so I'm skipping it here), then building the following machinery to quickly analyze restricted subsets of deck-space.

q = "1 <= dragon <= 6 and 1 <= lotus <= 6"
display(decks.query(q).corr()["win"].drop("win").sort_values(ascending=False).plot.bar())
decks.query(q)["win"].agg(["mean", "sum", "count"])

q is used to filter us down to decks that obey the constr

... (read more)

My counterpoints, in broad order of importance:

  1. If you lie to people, they should trust you less. Observing you lying should reduce their confidence in your statements. However, there is nothing in the fundamental rules of the universe that say that people notice when they are deceived, even after the fact, or that they will trust you less by any amount. Believing that lying, or even being caught lying, will result in total collapse of confidence without further justification is falling for the just-world fallacy.
  2. If you saw a man lying to his child about
... (read more)

If good people were liars, that would render the words of good people meaningless as information-theoretic signals, and destroy the ability for good people to coordinate with others or among themselves.

My mental Harry is making a noise. It goes something like Pfwah! Interrogating him a bit more, he seems to think that this argument is a gross mischaracterization of the claims of information theory. If you mostly tell the truth, and people can tell this is the case, then your words convey information in the information-theoretic sense.

EDIT: Now I'm think... (read more)

The sentence you quote actually sounds like a Harry sentence to me. Specifically the part where doing an unethical thing causes the good people to not be able to trust each other and work together any more, which is a key part of the law of good.

6Alex Vermillion2y
The idea, I believe, is similar to asking death row prisoners if they are innocent. If you establish that you're willing to lie about sufficiently important things for non-obvious reasons, people can't trust you when those reasons are likely to be in play. For Eliezer's stakes, this would be literally all the time, since it would "be justified" to lie in any situation to save the world.

Followed up on this idea and noticed that

A table of winrate as function of number of "evil" cards and "item" cards shows that item cards only benefit evil decks. I considered dragon, emperor, hooligan, minotaur, and pirate to be evil.

  • "No items, all good" wins 55.6%
  • "4 items, no good" wins 58.4%
  • "4+ items, all good" peaks at 37.0% winrate, and drops when adding items

Sorry in advance for an entirely too serious comment to a lighthearted post; it made me have thoughts I thought worth sharing. The whole "Karma convertibility" system is funny, but the irony feels slightly off. Society (vague term alert!) does in fact reward popular content with money. Goodhart's law is not "monetizing an economy instantly crashes it". My objections to Karma convertibility, are:

  1. Exploitability. Put in enough people on a system, and some of them will try to break it. Put in even more people, and they will likely succeed.
  2. Inefficiency. Ther
... (read more)

If there is a place where "pay no attention to that man behind the curtain" is to be disobeyed with prejudice, I'd reckon this is it. ;)

Just the obvious contrarian poke:

Are the decks equally likely? We observe that 412050 decks appear just once, 104483 decks appear twice, etc. Is this distribution compatible with random draws?

There are 342396 rows, with 2 decks each. Solving for the number of valid decks one could make, gives me (straightforward application of counting particle arrangements, imagine you have "coins" to place in "card-type boxes").

Then I just simulated and eyeballed. If I pick 2*342396 random numbers from 1 to 1352078, how many numbers appear just once

... (read more)
4aphyer2y
I don't know why I expected anyone to take things for granted just because I told them not to bother looking into them.

I'm interested! Thank you for the heads-up. I'll be sure to pop by.

Hm. I expected to do terribly on this problem since I hardly exhausted my avenues of research (I didn't even clean up the infohazardous object, despite knowing about it from other's comments). I ended up doing the worst out of everybody, which tracks, but the results are still clustered rather close.

I enjoyed the problem a lot, and I'm very grateful to aphyer for pinging me when he made the problem available. Tragically, I was sick at the time. :P

Brief notes:

  • Unlike Umberto Eco's "How to Write a Thesis", and like johnwensworth's "How To Write Quickly While Maintaining Epistemic Rigor", this does not propose you sacrifice the importance of questions on the altar of rigor
  • There is a pretty big hole on how to find "the 1-3 most prominent pieces on each side". When first encountering a claim on the context of an academic debate, this is usually not an issue. Often, not so much. Obviously, if you simply came up with the claim you have no conveniently laid out pointer to the academic literature.
  • "Just w
... (read more)
3HoldenKarnofsky2y
I broadly agree. I think it is sometimes challenging to find the major pieces on an issue, though rarely super challenging (usually, if I just read the first few pieces I find and search citations backwards and forwards, at some point I find myself running into the same pieces over and over).  

I came in pretty late, so I haven't gotten much to share.

I split my analysis into expected-profit-if-obtained and chance-of-obtaining-per-team. It probably isn't true, but assuming that team selection does not directly affect profit simplifies things a lot.

  1. Predators incur losses, except Euclid humanoids which are profitable.
  2. Safe objects are profitable.
  3. Euclid objects are profitable, except locations and replicators which are neutral.
  4. Keter objects incur losses, except humanoid organics which are profitable.
  5. When objects we obtain
... (read more)

And here is my impressions after seeing the spoilers and solution.

This post should be spoilered. Use >! to hide information that could tip somebody off, which in this case is everything.

1Pablo Repetto2y
And here is my impressions after seeing the spoilers and solution.

Maybe it is. Feynman's abacus story suggests that he (and colleagues) were familiar with lots of specific numbers and that it matters, somehow. Perhaps I should pick up the habit. Or perhaps that's backwards, and there's some particularly useful skill tree that, as a side effect, results in learning to recognize lots of numbers. Either way, just knowing that this is a common thing among the mathematically inclined is worth knowing.

3GuySrinivasan2y
If I had to guess, I'd guess that the largest contributor towards viewing numbers like that was probably my courses taught from https://www.amazon.com/Discrete-Combinatorial-Mathematics-Applied-Introduction/dp/0201199122/ in university.

I see. The spikiness is a tipoff that the numbers are being generated by some simple underlying process. I'm still not clear about why primes, though.

I'm guessing the idea is looking out for multiplicative processes, like looking out for the hump-tail shape of the distribution? Multiplying numbers together is an addition on their multiplicities-of-factors representation, so nd6 can never generate a number with a prime factor of 7 or higher. But I'm not explicitly hearing that as the rationale, so it feels like "primes are bound to show up, just keep an eye out for them".

2GuySrinivasan2y
Oh. Um, I just see a lot of numbers as their prime factorization so it was obvious something unusual was going on. Probably not helpful to you, there. But I guess it's similar to what gjm said. Like how you'd notice if everything was divisible by 10 because everything ended in 0s, but not quite so clear.

Thank you gjm and Guy for the responses! It lifts my spirits immensely to know that my work is being received favorably.

Reproduced verbatim so both recipients are notified.

Thank you gjm and Guy for the responses! It lifts my spirits immensely to know that my work is being received favorably.

Reproduced verbatim so both recipients are notified.

I've written my independent work into a post. There is also a preface with some explanation as to the choice of tools.

1Pablo Repetto2y
My thoughts after reading the solution and spoiler comments.

It's great fun! I'm very pleased you are enjoying the posts. :)

I'm late for the party. I put my blind analysis on a full post, and will be going through all the problems in order.

2Pablo Repetto2y
Retrospective thoughts after seeing the solution.

Certainly for crux-hunting, you need two people who are fundamentally collaborating.

It has been pointed out to me that therapy is analogous to depositions in a way relevant to your argument: in therapy both patient and therapist are there with the stated purpose of resolving emotional tensions in the patient, but the patient can prove unhelpful or actively oppose the therapist's probes.

I think this is an example of an interaction that is collaborative in principle, but where techniques designed for adversarial interactions may do good.

2supposedlyfun2y
I agree with that particular observation about therapy, having been therapized effectively myself (big ups to therapy!), but I fundamentally trusted that the therapist knew what ze was doing--based on diplomas on the wall, recommendations from trusted sources, and the first session with zir. The time that I did not feel that fundamental trust, I didn't go back. Thus, when a therapist asked me a question that bothered me, I believed that 1) ze had a good professional reason for asking, so I should figure out the answer and 2) my discomfort about the question was worth investigating as well.  And maybe even 3) the discomfort is worth pushing through because it's a signal of a possible avenue for a breakthrough, and given that I value previous breakthroughs at $2,000 cash or maybe more if I thought hard about it, the discomfort is exciting in a way.     Contrast: If I'm a deponent, and the lawyer asks me a question that bothers me, my (sensible) belief is that the truthful answer is bad for my case, meaning that the amount of money I'm going to {win/lose} is going to go {down/up} when I answer. If you thought that answering a certain question truthfully would cost you $10,000 USD, you wouldn't answer it if you didn't have to. Thus, another crucial distinguishing element of a deposition is that the deponent will face serious consequences if ze doesn't sit for zir deposition: if the deponent is a party to the lawsuit, a judgment will probably be entered against zir, and if the deponent is a non-party who has been properly served with a valid deposition subpoena, ze will be held in contempt and theoretically jailed until ze does comply. Few people, by contrast, are compelled by sanction of law into therapy or crux-hunting. 

"I don't know" can be a accurate. I think the advice is intended against people playing dumb, like Bill Clinton's "depends on what the meaning of the word 'is' is" or this witness denying knowledge of what a photocopier is. I know I've pulled this bullshit on myself at least once.

Using strategies that still work when some people act adverserial with you and try to deceive you is in line with being rational.

I think this gets close to the insight that motivated my post: a part of ourselves often tries to curl into a ball and deny reality to avoid emotional stress, interacting with that part of you is kind of adversarial.

No and no.

When I appear in the room I observe that I exist, but I couldn't have observed that I didn't exist, thus I don't update.

When a minute passes I observe that I survived, but I couldn't have observed that I didn't survive, thus I don't update.

This comment did not deserve the downvotes; I agree with asking for disclosure.

It does deserve criticism for tone. "Alarmist and uninformed" and "AGI death cult" are distractingly offensive.

The same argument for disclosure could could have been made by "given that LW's audience has outsized expectations of AI performance" and "it costs little, and could avoid an embarrasing misunderstanding".

To expand on 5:

I may be explaining Scrum for a job interview, and completely forget that the sprint review is a thing. Ask me about the sprint review however, and I can make a cogent case for (or against) the necessity of the dev team being involved (customer interactions are the purview of the project owner! agile methodologies emphasize cutting red tape! or something on those lines).

I use notes as reminders/pointers rather than longform descriptions (adopted from "The Bullet Journal Method", ch 2 "Events"). This helps with three things:

  1. reviewing (looking back on my month),
  2. remembering ideas when they are relevant,
  3. building a big-picture view while reading.
  1. Not very, still developing habit.
  2. Mostly for work-related minutae, but trying to expand that.
  3. Yes. I keep a stack of clipped together A8 sheets of paper on me at all times (essentially as a "RAM" extension), write a (modified) Bullet Journal, and keep an Obsidian vault with more topic-oriented permanent notes (this is also where book summaries go). I also use Zotero for digital collection management.
  4. Sort of. I've tried a few different ideas (Zettelkasten, Bullet Journal, Getting Things Done, Cornell notes, plus some hacked-together monstruosities) for severa
... (read more)
3Pablo Repetto3y
To expand on 5: I may be explaining Scrum for a job interview, and completely forget that the sprint review is a thing. Ask me about the sprint review however, and I can make a cogent case for (or against) the necessity of the dev team being involved (customer interactions are the purview of the project owner! agile methodologies emphasize cutting red tape! or something on those lines). I use notes as reminders/pointers rather than longform descriptions (adopted from "The Bullet Journal Method", ch 2 "Events"). This helps with three things: 1. reviewing (looking back on my month), 2. remembering ideas when they are relevant, 3. building a big-picture view while reading.

Some time ago I noticed this trend among people I respect on Twitter (motivating examples). It seems to me that there is a consensus view that openness has a damaging effect on discourse.

This view does not seem to stem from the problem addressed by "Well-Kept Gardens Die By Pacifism" and "Evaporative Cooling of Group Beliefs" -the gradual decline of a community due to environmental exposure- but rather from the problem that you percieve: the reputational hazard of public fora.

My current stance on public discourse is that it serves as a discovery mechanism:... (read more)

6DirectedEvolution3y
That connects with my piece on the EA Forum, Articles are Invitations. That's a nice way to look at it. Over time, I've come to see article writing, including scientific articles, as primarily ways to coordinate people to work together, and only having a secondary purpose to convey information. That doesn't mean that their information-conveying purpose is unimportant, but rather that their coordination function is extremely important and often neglected.

I have not yet read the book, but from a recommendation I infer that Leo Strauss' "Persecution and the Art of Writing" goes into considerable depth on the mechanics of using vagueness as a dogwhistling mechanism.

And remember: on-site backups are not backups. Even if you are working on a git repo, it is surprisingly easy to screw up your commit history so badly you have to burn it all down and start over, and that is not by far the worst possible disaster.

It is a direct response to a quotation from the article, so not really.

I guess I want to be "a normal [...] man wearing a female body like a suit of clothing."

Is that weird? Is that wrong?

Okay, yes, it's obviously weird and wrong, but should I care more about not being weird and wrong, than I do about my deepest most heartfelt desire that I've thought about every day for the last nineteen years?

I know this misses most of the point of the article, but I also believe it's worth pointing out: I don't think a male wanting a female body form is any weirder or wronger that a male wanting to be 2 inches taller, buff, and having 20/20 eyesight.

PS: I did try reducing "weird and wrong" to their components. Result of the excercise: I find the OP uncontroversially "statistically rare" or "heterodox", but neither "viscerally repulsive" nor "morally reprehensible". I can see the value of explicitly reducing complex concepts in the general case, but I'm not sure it was worthwhile for this instance.

3wunan3y
Can you taboo the words "weird" and "wrong" in that comment?

May I propose "appraisals" as a substitute for "opinions"? It is more precise, in that it implies judgement of worth

1Logan Riggs3y
Thanks! Changed to "social appraisals". Someone's opinion is definitely a loaded term which may lead to pattern matching. I'm also fine with more novel phrasing since it's explained immediately after.

Who wouldn't watch an ad for $10? Hell, at 10 cents per 1-minute ad, that is near minimum wage.

8Elizabeth3y
Anyone for whom executive function is a limiting reagent, unless the $10 is exceptionally significant to them.  If it's a one-off transaction, it costs more than just the minute to watch the ad- it's the time and energy to evaluate the offer, see if there are any catches, refocus after disrupting your thoughts, fight against any impact the ad would have on you (they wouldn't keep making this deal if it wasn't positive EV for them. Maybe they're mistaken, maybe you're an exception- but that takes thought to figure out)... The fact that it's a pretty good hourly rate doesn't matter if it's not actually running long enough to amortize the evaluation costs.

Appreciated. To my rss it goes.

Currently grappling with this problem (compsci undergrad). I'm pulling ideas from Allen's Getting Things Done, Carroll's Bullet Journal, and Ahrens' How to take smart notes

Hanabi

It is an excellent game to really get the concepts of priors and subjectively objective probabilities. Try a round and ask "what is the probability that the next card on the deck is a 5?", interesting discussion ensues.

Load More