MichaelDickens

• MichaelDickens 27 Jun 2022 19:20 UTC

  21 points

  11

  on: LessWrong Has Agree/​Disagree Vot­ing On All New Com­ment Threads

  I am pretty uncertain about whether this change is good, and I don’t think anyone can confidently say it is or isn’t good. But no other forum with voting does this (AFAIK), so it’s good to try it and see what happens.

  Something to think about: What sorts of observations might constitute evidence in favor of or against this system?

How can we in­crease the fre­quency of rare in­sights?

[Question] Why does gra­di­ent de­scent always work on neu­ral net­works?

• MichaelDickens 27 Mar 2024 2:00 UTC

  11 points

  1

  in reply to: Tenoke’s comment on: My In­ter­view With Cade Metz on His Re­port­ing About Slate Star Codex

  When the NYT article came out, some people discussed the hypothesis that perhaps the article was originally going to be favorable, but the editors at NYT got mad when Scott deleted his blog so they forced Cade to turn it into a hit piece. This interview pretty much demonstrates that it was always going to be a hit piece (and, as a corollary, Cade lied to people saying it was going to be positive to get them to do interviews).

  So yes this changed my view from “probably acted unethically but maybe it wasn’t his fault” to “definitely acted unethically”.

• MichaelDickens 19 Mar 2024 0:06 UTC

  9 points

  5

  in reply to: johnswentworth’s comment on: On Devin

  people have repeatedly told me that a surprisingly high fraction of applicants for programming jobs can’t do fizzbuzz

  I’ve heard it argued that this isn’t representative of the programming population. Rather, people who suck at programming (and thus can’t get jobs) apply to way more positions than people who are good at programming.

  I have no idea if it’s true, but it sounds plausible.

• MichaelDickens 6 Mar 2024 21:30 UTC

  9 points

  0

  in reply to: niplav’s comment on: If you weren’t such an idiot...

  Just last week I wrote a post reviewing the evidence on caffeine cycling and caffeine habituation. My conclusion was that the evidence was thin and it’s hard to say anything with confidence.[1]

  My weakly held beliefs are:

  1. Taking caffeine daily is better than not taking it at all, but worse than cycling.

  2. Taking caffeine once every 3 days is a reasonable default. A large % of people can take it more often than that, and a large % will need to take it less.

  I take caffeine 3 days a week and I am currently running a self-experiment (described in my linked post). I’m currently in the experimental phase, I already did a 9-day withdrawal period and my test results over that period (weakly) suggest that I wasn’t habituated previously because my performance didn’t improve during the withdrawal period (it actually got worse, p=0.4 on a regression test).

  [1] Gavin Leech’s post that you linked cited a paper on brain receptors in mice which I was unaware of, I will edit my post to include it. Based on reading the abstract, it looks like that study suggests a weaker habituation effect than the studies I looked at (receptor density in mice increased by 20–25% which naively suggests a 20–25% reduction in the benefit of caffeine whereas other studies suggest a 30–100% reduction, but I’m guessing you can’t just directly extrapolate from receptor counts to efficacy like that). Gavin also cited Rogers et al. (2013) which I previously skipped over because I thought it wasn’t relevant, but on second thought, it does look relevant and I will give it a closer look.

• MichaelDickens 8 Sep 2022 15:27 UTC

  9 points

  on: MichaelDick­ens’s Shortform

  What’s going on with /​r/​AskHistorians?

  AFAIK, /​r/​AskHistorians is the best place to hear from actual historians about historical topics. But I’ve noticed some trends that make it seem like the historians there generally share some bias or agenda, but I can’t exactly tell what that agenda is.

  The most obvious thing I noticed is from their FAQ on historians’ views on other [popular] historians. I looked through these and in every single case, the /​r/​AskHistorians commenters dislike the pop historian. Surely at least one pop historian got it right?

  I don’t know about the actual object level, but a lot of /​r/​AskHistorians’ criticisms strike me as weak:

  • They criticize Dan Carlin for (1) allegedly downplaying the Rape of Belgium even though by my listening he emphasized pretty strongly how bad it was and (2) doing a bad job answering “could Caesar have won the Battle of Hastings?” even though this is a thought experiment, not a historical question. (Some commenters criticize him for being inaccurate and others criticize him for being unoriginal, which are contradictory criticisms.)

  • They criticize Guns, Germs, and Steel for...honestly I’m a little confused about how this person disagrees with GGS.

  • Lots of criticisms of popular works for being “oversimplified”, which strikes me as a dumb criticism—everything is simplified, the map is always less detailed than the territory.

  • They criticize The Better Angels of Our Nature for taking implausible figures from ancient historians at face value (fair) and for using per capita deaths instead of total deaths (per capita seems obviously correct to me?).

  Seems like they are bending over backwards to talk about how bad popular historical media are, while not providing substantive criticisms. I’ve also noticed they like to criticize media for not citing any sources (or for citing sources that aren’t sufficiently academic), but then they usually don’t cite any sources themselves.

  I don’t know enough about history to know whether /​r/​AskHistorians is reliable, but I see some meta-level issues that make me skeptical. I want to get other people’s takes. Am I being unfair to /​r/​AskHistorians?

  (I don’t expect to find a lot of historians on LessWrong, but I do expect to find people who are good at assessing credibility.)

• MichaelDickens 1 Apr 2023 1:03 UTC

  8 points

  12

  in reply to: LVSN’s comment on: Ex­po­sure to Lizard­man is Lethal

  If you disagree but can’t succinctly explain, I would suggest doing one of these things:

  1. Write a long comment explaining your disagreement

  2. Write a short comment stating your specific points of disagreement, with a disclaimer that you don’t have time to fully justify your beliefs

  Your comment is being downvoted (I suspect) because it does neither of these, instead it indirectly insults the author without providing any information as to why you disagree. IMO this sort of comment doesn’t really contribute anything—all I know is that you disagree, I have no idea what’s going on inside your head, so I’m not learning anything from it.

• MichaelDickens 11 Nov 2021 20:25 UTC

  8 points

  in reply to: Alexander Gietelink Oldenziel’s comment on: Dis­cus­sion with Eliezer Yud­kowsky on AGI interventions

  10 million dollars will probably have very small impact on Terry Tao’s decision to work on the problem.

  That might be true for him specifically, but I’m sure there are plenty of world-class researchers who would find $10 million (or even $1 million) highly motivating.

• MichaelDickens 14 Apr 2022 19:43 UTC

  7 points

  in reply to: gbear605’s comment on: Lies Told To Children

  I thought it was obviously fiction, but I didn’t know that it was set in Dath Ilan, and the fact that it’s set in Dath Ilan would give away that the red hair thing is fake.

• MichaelDickens 26 Apr 2024 1:20 UTC

  6 points

  0

  on: MichaelDick­ens’s Shortform

  Have there been any great discoveries made by someone who wasn’t particularly smart?

  This seems worth knowing if you’re considering pursuing a career with a low chance of high impact. Is there any hope for relatively ordinary people (like the average LW reader) to make great discoveries?

• MichaelDickens 24 Mar 2022 3:18 UTC

  6 points

  in reply to: jefftk’s comment on: Jet­lag, Nausea, and Diar­rhea are Largely Optional

  I “know” that nausea can be handled with a pill, but it had never occurred to me to carry around a couple anti-nausea pills.

• MichaelDickens 24 Feb 2024 7:48 UTC

  5 points

  6

  on: Con­tra Ngo et al. “Every ‘Every Bay Area House Party’ Bay Area House Party”

  The contextualizer/​decoupler punch is an outstanding joke.

• MichaelDickens 2 Dec 2022 16:40 UTC

  4 points

  on: The Fal­lacy of Gray

  A related pattern I noticed recently:

  • Alice asks, “What effect does X have on Y?”

  • Bob, an expert in Y, replies, “There are many variables that impact Y, and you can’t reduce it to simply X.”

  Alice asked for a one-variable model with limited but positive predictive power, and Bob replied with a zero-variable model with no predictive power whatsoever.

• MichaelDickens 28 Mar 2022 19:07 UTC

  4 points

  in reply to: TLW’s comment on: Is Me­tac­u­lus Slow to Up­date?

  You see this sort of thing with acquisitions. Say company A is currently priced at $100, and company B announces that it’s acquiring A for $200 per share. A will jump up to something like $170 per share, and then slowly increase to $200 on the acquisition date. The $30 gap is there because there’s some probability that the acquisition will fall through, and that probability decreases over time (unless it actually does fall through, in which case the price drops back down to ~$100).

• MichaelDickens 3 Aug 2021 2:42 UTC

  4 points

  in reply to: lsusr’s comment on: Covid 7/​29: You Play to Win the Game

  The high-level explanation I’d give for this is that smart people make better decisions in general, and certain classes of bad decisions are also illegal. So perhaps the reason smart people follow rules more isn’t that they’re more inherently rule-abiding, but that they behave in more reasonable ways, and rules tend to be reasonable (obviously not always, but they’re more reasonable than if they were assigned at random).

• MichaelDickens 29 Mar 2024 18:30 UTC

  3 points

  0

  in reply to: romeostevensit’s comment on: Dis­in­cen­tives for par­ti­ci­pat­ing on LW/​AF

  I find that sort of feedback more palatable when they start with something like “This is not related to your main point but...”

  I am more OK with talking about tangents when the commenter understands that it’s a tangent.

• MichaelDickens 16 Jun 2023 19:42 UTC

  3 points

  0

  in reply to: Dagon’s comment on: I still think it’s very un­likely we’re ob­serv­ing alien aircraft

  Ok, fair point, I was going too far in assuming that the sort of engineering necessary was physically impossible.

• MichaelDickens 23 Jan 2023 0:05 UTC

  3 points

  1

  in reply to: shminux’s comment on: Ve­gan Nutri­tion Test­ing Pro­ject: In­terim Report

  Fortified milks usually don’t contain much iron. The soymilk in my fridge (Silk unsweetened) has 120% RDA of B12 but only 6% RDA of iron.

• MichaelDickens 31 Dec 2022 17:46 UTC

  3 points

  8

  in reply to: Eli Tyre’s comment on: Sazen

  Without the outline, the stars look really skinny. To my eye, it looks much more like an anteater.