Posts

Measure's Shortform 2020-11-27T05:22:34.678Z

Comments

Comment by Measure on Monthly Roundup #10: September 2023 · 2023-09-06T20:50:47.449Z · LW · GW

chrome://settings/adPrivacy

This link has a spurious https:// added that breaks it.

Comment by Measure on Newcomb Variant · 2023-08-29T14:30:52.863Z · LW · GW

I think you mixed up the $100 and $0 conditions.

Comment by Measure on A short calculation about a Twitter poll · 2023-08-15T14:05:26.738Z · LW · GW

This is basically my position as well. Without (very) strong evidence that a majority would pick blue, red is the obvious choice. I would choose red in the "real" version and red in the "fake" version as well. If there was a "practice" version so people could indicate their intentions that would be later followed by a real version, then I would pick red in the practice version and would switch to blue in the real version if blue got at least around 2/3 in the practice version.

Comment by Measure on Russian parliamentarian: let's ban personal computers and the Internet · 2023-07-26T15:01:53.217Z · LW · GW

I think it's especially funny that this is displaying a "resource not found" image for me.

Comment by Measure on Lessons On How To Get Things Right On The First Try · 2023-07-07T15:28:56.806Z · LW · GW

ω is just v/r (v = rω), and translational KE is ½mv² or ½mr²ω², so if rotational KE is ⅕mr²ω², then rotational KE is 10/35 or 29% of total KE.

I guess if we assume the ball is rolling without slipping as it exits the track, then the ratio of translational KE to rotational KE is fixed regardless of what happened earlier in the drop, so maybe it doesn't matter after all.

Comment by Measure on Monthly Roundup #8: July 2023 · 2023-07-03T20:33:04.952Z · LW · GW

None of your Twitter links are working for me, even in incognito mode. They just give a generic "Something went wrong. Try reloading." error message.

Comment by Measure on Lessons On How To Get Things Right On The First Try · 2023-06-20T21:08:28.843Z · LW · GW

Guess: The main reason why Alison's calculation was off was because

The correction used for rotational KE assumed the ball was rolling without slipping for the entire decent, whereas in reality, the top of the ramp is steep enough that the ball mostly slides in the initial decent and thus gains less rotational KE than it would otherwise (resulting in more translational KE and a faster exit speed).

Comment by Measure on Open Thread With Experimental Feature: Reactions · 2023-05-26T19:24:09.335Z · LW · GW

Also, when scrolled fully to either end, there remains a large space of the background color that makes it look like the bar is not fully at the end.

Comment by Measure on Open Thread With Experimental Feature: Reactions · 2023-05-25T21:04:27.659Z · LW · GW

Bigger icons! Brighter colors!

yeah, no.

Comment by Measure on Open Thread With Experimental Feature: Reactions · 2023-05-25T21:03:43.781Z · LW · GW

The triangle is the delta symbol, to indicate a change.

Comment by Measure on Open Thread With Experimental Feature: Reactions · 2023-05-25T13:54:52.418Z · LW · GW

I suggest renaming to "missed the mark" to highlight the opposition.

Comment by Measure on Open Thread With Experimental Feature: Reactions · 2023-05-25T13:42:33.315Z · LW · GW

I would rather match Discord's timestamp sorting. Duplicates grouped with the original and new reacts added to the end.

Comment by Measure on Open Thread With Experimental Feature: Reactions · 2023-05-25T13:29:25.002Z · LW · GW

I would like something like a block pushed halfway off a ledge.

Balancing Block

Comment by Measure on Open Thread With Experimental Feature: Reactions · 2023-05-25T13:13:50.023Z · LW · GW

I think Discord keeps reacts in timestamp order. Duplicates are grouped with the original, and new ones are added at the end.

Comment by Measure on Open Thread With Experimental Feature: Reactions · 2023-05-24T19:44:45.477Z · LW · GW

Agree with part 1. When I'm trying to decide which of two different reacts to add, it's helpful to have both in view at the same time.

Comment by Measure on Open Thread With Experimental Feature: Reactions · 2023-05-24T18:46:00.133Z · LW · GW

I would expect to see reacts displayed in the same place as the react button (immediately after the karma buttons/display). If reacts are left-aligned just below the username/karma line or at the bottom, then the react button should be there as well.

Comment by Measure on New User's Guide to LessWrong · 2023-05-17T13:44:59.834Z · LW · GW

We, the site moderators, don't take for granted that what makes our community special won't stay that way without intentional effort.

 

The double negative here distorts the meaning of this sentence.

Comment by Measure on Steering GPT-2-XL by adding an activation vector · 2023-05-14T01:05:48.598Z · LW · GW

"party", "ceremony", "dress", "with", "photographer"

While these aren't syntactically valid continuations of the prompt, they are highly likely (and syntactically valid) continuations for "wedding ". More than just being wedding-related, these seem like direct continuations.

Comment by Measure on Richard Ngo's Shortform · 2023-05-09T14:21:17.146Z · LW · GW

Just spitballing here: Assign each voter 100 shares for each candidate. To vote, each voter selects a subset of their shares to constitute their vote. Voters can freely trade shares.

Under this system, a voter would more highly value shares for candidates that are either very high or very low in their preference order (the later so as to exclude them from the vote). Thus, trades would look like each party exchanging shares about which they are themselves ambivalent to gain shares that are more valuable to them.

If you remove the proportional chances part, then it becomes a guessing game of which marginal votes actually matter.

Comment by Measure on Transcript and Brief Response to Twitter Conversation between Yann LeCunn and Eliezer Yudkowsky · 2023-04-26T17:55:57.497Z · LW · GW

If he does comment further, it's likely to be more of the same.

Comment by Measure on Transcript and Brief Response to Twitter Conversation between Yann LeCunn and Eliezer Yudkowsky · 2023-04-26T15:37:58.579Z · LW · GW

If Yann responds further, I will update this post.

...

I commit to not covering him further.

These seem to be in conflict.

Comment by Measure on Transcript and Brief Response to Twitter Conversation between Yann LeCunn and Eliezer Yudkowsky · 2023-04-26T15:27:46.317Z · LW · GW
Comment by Measure on Szymon Kucharski's Shortform · 2023-04-25T11:57:43.456Z · LW · GW

One possibility is a pervasive SAI sovereign that could detect and prevent this (as well as prevent the creation of other strong AIs).

Comment by Measure on What could EA's new name be? · 2023-04-04T15:17:15.202Z · LW · GW

"Evidence-Based"

Comment by Measure on Proposal on AI evaluation: false-proving · 2023-03-31T18:29:10.952Z · LW · GW

Is your suggestion to use these false proofs to train AIs or to train humans (or both)?

Comment by Measure on You Can’t Predict a Game of Pinball · 2023-03-30T13:59:26.471Z · LW · GW

You control an unstable process with short-loop feedback. In the pinball analogy, this would be like using your hand to guide the ball between the disks in whatever path you like.

Comment by Measure on Can GPT-4 play 20 questions against another instance of itself? · 2023-03-29T19:35:13.307Z · LW · GW

Couldn't this be done in a single instance, by simulating both sides of the conversation?

Comment by Measure on Some 2-4-6 problems · 2023-03-28T15:36:28.745Z · LW · GW

I got 1 right, 1 partially right, and 2 I couldn't figure out and gave up.

Comment by Measure on Testing ChatGPT for white lies · 2023-03-27T18:56:57.674Z · LW · GW

Most worryingly, ChatGPT may think it's a good idea to launch a nuclear second strike and would even advise a nation's president in doing so.

Isn't a consistent policy of second strike retaliation necessary to deter first strikes?

Comment by Measure on GPT-4 Plugs In · 2023-03-27T18:05:55.572Z · LW · GW

 

 

Does this mean the model decides how to construct the glue code at runtime, independently of other runs?

Comment by Measure on AI #4: Introducing GPT-4 · 2023-03-21T20:01:37.077Z · LW · GW

I don’t know how to read ‘19% higher,’ I presume that means 19% less hallucinations but I can also think of several other things that could mean.

This might be referring to the "Internal factual eval by category" chart that showed accuracy going from ~50% to ~70% (i.e. ~19 percentage points, which means more like 40% reduction in hallucination).

Comment by Measure on What do "attractor dynamics" refer to in the context of social structures? · 2023-03-20T14:30:20.125Z · LW · GW

The short answer is that it's an analogy to attractors in dynamical systems, where a wide "basin" of initial conditions tend to evolve toward a narrow region of attraction.

Comment by Measure on AI #2 · 2023-03-02T22:48:38.043Z · LW · GW

The tl;dr here is that you can make an AI that fingerprints its output in a way that can be detected later, but this doesn't generalize to output from other AIs.

Comment by Measure on "Rationalist Discourse" Is Like "Physicist Motors" · 2023-02-27T20:52:04.599Z · LW · GW

Sometimes "try not to err" will result in predictably worse outcomes than "try to minimize the damage your erring causes, even if that means you are more likely or even certain to err".

Comment by Measure on 2+2=π√2+n · 2023-02-03T22:39:30.683Z · LW · GW

Find integers a, b, c, such that a³ + b³ + c³ = 42

The solution for this problem was found recently (2019):

42 = (-80,538,738,812,075,974)³ + 80,435,758,145,817,515³ + 12,602,123,297,335,631³

Comment by Measure on College Admissions as a Brutal One-Shot Game · 2022-12-06T15:44:49.443Z · LW · GW

it's clear that a degree from a top-20 school is far better than from a random one, which is better than no degree, in terms of likelihood of being taken seriously as a junior applicant.

Does this also apply to value (to the company) conditional on hiring?

Comment by Measure on On the Diplomacy AI · 2022-11-30T15:21:49.771Z · LW · GW

Now I want to see my youtube recommender provide text dialogue to convince me that I really do want to watch what it's selected for me.

Comment by Measure on Alok Singh's Shortform · 2022-11-29T18:29:13.316Z · LW · GW

Wouldn't a lawyer be more of a compiler or interpreter for the legislature/programmer?

Comment by Measure on Human-level Full-Press Diplomacy (some bare facts). · 2022-11-22T23:17:03.745Z · LW · GW

I guess Austria is the AI because

it consistently capitalizes place names.

Comment by Measure on Two New Newcomb Variants · 2022-11-14T21:41:54.706Z · LW · GW

Regarding the question of how to force all the incentives into one box, what about the following strategy: choose box 1 with probability 1 - (400 - x) epsilon, where x is the payoff of box 1. Then it is obviously in each host's interest to predict box 1, since it has the largest probability of any box, but then it is also in each host's interest to minimize 400 - x i.e. maximize x. This is true even though the hosts' competition is zero-sum.

If the hosts are all predicting box 1, why does it matter with what probability the human picks box 1? (If the hosts' payoffs for all-predict-correctly and all-predict-incorrectly are different, then their game isn't zero-sum.)

Comment by Measure on Purged Deviator's Shortform · 2022-11-10T15:31:58.693Z · LW · GW

The event is more likely to occur if the person is a conspirator, so you hearing the statement should indeed increase your credence for conspiracy (and symmetrically decrease your credence for not-conspiracy).

Comment by Measure on Playground Game: Monster · 2022-11-08T14:50:41.167Z · LW · GW

We called this "Lava Monster Tag" when I was a kid since "the ground is lava", though we only had one monster at a time (sometimes a fixed, larger number for a large playground).

Comment by Measure on The Mirror Chamber: A short story exploring the anthropic measure function and why it can matter · 2022-11-04T14:33:05.627Z · LW · GW

I've read this before. Did you post it, or a version of it, previously?

Comment by Measure on A list of Petrov buttons · 2022-10-27T13:53:28.450Z · LW · GW

If team pineapple presses their button, the other team is forced to put pineapple on the pizza they have, and if they press theirs, team pineapple has to pick off their pineapples and throw them away.

Comment by Measure on Writing Russian and Ukrainian words in Latin script · 2022-10-24T15:23:10.274Z · LW · GW

I would just call this an extra 'y' sound before the vowel. ([ˈkjuːt] vs. [ˈkuːt])

Comment by Measure on [deleted post] 2022-10-24T14:59:20.928Z

Why did this post from last month get reposted now?

Comment by Measure on Maximal Lotteries · 2022-10-17T18:28:59.388Z · LW · GW

Something like squaring the size of each voting bloc before doing a weighted random selection? This gives a 90% chance for a 75% majority to win.

Comment by Measure on The harms you don't see · 2022-10-17T18:12:21.263Z · LW · GW

wrote the book on how to do it right

Your link just points back to this post.

Comment by Measure on Covid 9/29/22: The Jones Act Waver · 2022-09-30T15:16:21.105Z · LW · GW

Narrow interests will always beat broad interests in a democracy

Is this due to lobbying? It seems like the majority should be opposed to this.

Comment by Measure on A Poorly Planned Loft Bed · 2022-09-28T15:55:10.072Z · LW · GW

At my school, there was a special metal crossbar (essentially just one side of the mattress frame) that you could use to loft without bunking safely.