Comment by jimrandomh on Can a Bayesian agent be infinitely confused? · 2019-03-22T19:55:06.463Z · score: 10 (2 votes) · LW · GW

The latter; it could be anything, and by saying the probabilities were 1.0 and 0.0, the original problem description left out the information that would determine it.

Comment by jimrandomh on Can a Bayesian agent be infinitely confused? · 2019-03-22T19:02:11.583Z · score: 15 (6 votes) · LW · GW

If you do out the algebra, you get that P(H|E) involves dividing zero by zero:

There are two ways to look at this at a higher level. The first is that the algebra doesn't really apply in the first place, because this is a domain error: 0 and 1 aren't probabilities, in the same way that the string "hello" and the color blue aren't.

The second way to look at it is that when we say and , what we really meant was that and ; that is, they aren't precisely one and zero, but they differ from one and zero by an unspecified, very small amount. (Infinitesimals are like infinities; is arbitrarily-close-to-zero in the same sense that an infinity is arbitrarily-large). Under this interpretation, we don't have a contradiction, but we do have an underspecified problem, since we need the ratio and haven't specified it.

Comment by jimrandomh on [deleted post] 2019-03-16T00:06:33.254Z

The Jewish liturgy about divine judgment can be quite different. Every week, at the beginning of the Sabbath, Jews around the world sing Psalms a collection of psalms focused on the idea that the world is rejoicing because God is finally coming to judge it. From Psalm 96:

Say among the nations that the Lord reigns: the world shall so be established that it shall not be moved: he shall judge the peoples with uprightnesses. Let the heavens rejoice, and let the earth be glad; let the sea roar, and its fullness. Let the field be joyful, and all that is in it: then shall all the trees of the wood sing for joy. Before the Lord: for he comes, for he comes to judge the land: he shall judge the world with justice, and the peoples in his faithfulness. From Psalm 98: Melodize to the Lord with harp; with harp, and melodic voice. With the trumpets, and the voice of the horn, shout before the king, the Lord. Let the sea roar, and its fullness; the world, and those who dwell in it. Rivers shall clap their hands; together, the mountains shall sing for joy. Before the Lord: for he comes, for he comes to judge the land: he shall judge the world with justice, and the peoples in his faithfulness. In one of these outlooks, humans can't behave well enough to stand up to pure justice, so we should put off the day of judgment for as long as we can, and seek protection. In the other, the world is groaning under the accumulated weight of hypocrisy and sin, and only the reconciliation of accounts can free us; in constant flux due to ever-shifting stories, which can only be stabilized by a true judge. We can't reconcile accounts if that means punishing all bad behavior according to the current hypocritical regime's schedule of punishments. But a true reconciliation also means adjusting the punishments to a level where we'd be happy, not sad, to see them applied consistently. (Sometimes the correct punishment is nothing beyond the accounting itself.) In worlds where hypocrisy is normal, honesty is punished, since the most honest people will tend to reveal deprecatory information others might conceal, and be punished for it. We get less of what we punish. But honesty isn't just a weird quirk - it's the only way to get to the stars. "The first principle is that you must not fool yourself, and you are the easiest person to fool." - Richard Feynman

Comment by jimrandomh on LW Update 2019-03-12 -- Bugfixes, small features · 2019-03-15T00:39:54.599Z · score: 2 (1 votes) · LW · GW

Neither of those will autolink. Autolinking is handled at the UI level, in the default (WYSIWYG/draftjs) editor only.

Comment by jimrandomh on Blackmailers are privateers in the war on hypocrisy · 2019-03-14T13:11:04.870Z · score: 75 (24 votes) · LW · GW

There's something I think you're missing here, which is that blackmail-in-practice is often about leveraging the norm enforcement of a different community than the target's, exploiting differences in norms between groups. A highly prototypical example is taking information about sex or drug use which is acceptable within a local community, and sharing it with an oppressive government which would punish that behavior.

Allowing blackmail within a group weakens that group's ability to resist outside control, and this is a very big deal. (It's kind of surprising that, this late in the conversation about blackmail, no one seems to have spotted this.)

Comment by jimrandomh on LW Update 2019-03-12 -- Bugfixes, small features · 2019-03-14T01:50:52.521Z · score: 4 (2 votes) · LW · GW

The latter, but it applies immediately when you type it (rather than waiting until you click Submit), so it won't happen without you noticing.

Comment by jimrandomh on [deleted post] 2019-03-12T22:35:16.876Z

Foo

1: Asdf

## LW Update 2019-03-12 -- Bugfixes, small features

2019-03-12T21:56:40.109Z · score: 17 (2 votes)
Comment by jimrandomh on Karma-Change Notifications · 2019-03-05T21:01:09.994Z · score: 12 (3 votes) · LW · GW

It's plausibly correct to provide the option, so long as it isn't the default. (Options: Show all, show none, show only positive, show only negative. The last option being something that no one should ever use, provided only for symmetry.)

Comment by jimrandomh on Karma-Change Notifications · 2019-03-03T04:39:38.680Z · score: 9 (5 votes) · LW · GW

There is a real-time setting, which shows you everything since the last time you looked. It just isn't the default.

Comment by jimrandomh on Karma-Change Notifications · 2019-03-02T03:40:32.118Z · score: 9 (5 votes) · LW · GW

We want writing posts and comments, especially posts and comments which get a positive reception, to feel rewarding, so that people will do it more often. And, to a lesser but still significant degree, we want people to use the site.

2019-03-02T02:52:58.291Z · score: 91 (24 votes)
Comment by jimrandomh on [deleted post] 2019-03-01T00:49:55.179Z

9

Comment by jimrandomh on [deleted post] 2019-03-01T00:49:52.356Z

8

Comment by jimrandomh on [deleted post] 2019-03-01T00:49:49.385Z

7

Comment by jimrandomh on [deleted post] 2019-03-01T00:49:46.131Z

6

Comment by jimrandomh on [deleted post] 2019-03-01T00:49:42.410Z

5

Comment by jimrandomh on [deleted post] 2019-03-01T00:49:39.143Z

4

Comment by jimrandomh on [deleted post] 2019-03-01T00:49:35.847Z

3

Comment by jimrandomh on [deleted post] 2019-03-01T00:49:32.086Z

2

Comment by jimrandomh on [deleted post] 2019-03-01T00:49:27.848Z

1

Comment by jimrandomh on [deleted post] 2019-03-01T00:49:24.025Z

Rate limit test

Comment by jimrandomh on Two Small Experiments on GPT-2 · 2019-02-27T18:42:28.443Z · score: 3 (2 votes) · LW · GW

Yes, that is a pretty good summary of how Deep Blue works.

Comment by jimrandomh on [deleted post] 2019-02-26T05:56:15.721Z

It's no big deal; I've put weird test things on the site myself plenty of times.

Comment by jimrandomh on How good is a human's gut judgement at guessing someone's IQ? · 2019-02-26T03:38:50.509Z · score: 30 (10 votes) · LW · GW

Peoples' experiences with this are majorly confounded, and your experiment risks also being confounded, by the Conservation of Virtue effect. While it's true that mental attributes, health, height, wealth, and so on all positively correlate, within any filtered social group, those things tend to *negatively* correlate instead, because someone whose composite-trait-score is too low or high for a group will tend to end up in a different one. So, estimating IQ based on a conversation will tend to work well when the person whose IQ you're estimating comes from an unfiltered population, and poorly when it comes from a filtered one.

Comment by jimrandomh on Humans Who Are Not Concentrating Are Not General Intelligences · 2019-02-25T21:37:35.676Z · score: 6 (4 votes) · LW · GW

That is a plausible architecture, and is probably analogous to something humans do. But the "neural net which finds flaws in reasoning" would by itself be a much more complex object than a language model.

Comment by jimrandomh on Two Small Experiments on GPT-2 · 2019-02-25T21:33:04.289Z · score: 3 (2 votes) · LW · GW

Nitpick: Deep Blue does not backchain (nor does any widely used chess algorithm, to my knowledge).

Comment by jimrandomh on Take the Rationality Test to determine your rational thinking style · 2019-02-25T21:22:20.892Z · score: 2 (1 votes) · LW · GW

I've tried to find a complete version of the test online, and haven't succeeded in finding one. But there are a decent quantity of sample questions, covering all the question types, in the appendix to his book "The Rationality Quotient: Toward a Test of Rational Thinking".

Comment by jimrandomh on [deleted post] 2019-02-25T04:33:51.007Z

[Mod note: Moving this back to drafts folder]

Comment by jimrandomh on Two Small Experiments on GPT-2 · 2019-02-21T05:05:29.661Z · score: 6 (4 votes) · LW · GW

This definitely could use more trials. In the case of the sentiment analysis experiment, I'd ideally like to try out some other sentence structures (eg "Is a <noun> bad?", "Are <adjective> things good?); in the case of the Moloch experiment, I'd like to try some reruns with the same parameters, as well as different name substitutions, just to be sure that it isn't noise.

## Two Small Experiments on GPT-2

2019-02-21T02:59:16.199Z · score: 47 (20 votes)
Comment by jimrandomh on When does introspection avoid the pitfalls of rumination? · 2019-02-20T20:06:17.493Z · score: 7 (2 votes) · LW · GW

The thing that makes rumination bad is not its topic, but its repetitiveness; the first time one has a rumination-flavored thought it is potentially useful, and is not called rumination.

Comment by jimrandomh on [deleted post] 2019-02-15T03:13:09.988Z

Migration is finished and things should be fine now. Read-time estimates on posts are temporarily disabled due to a regression, and will be back soon.

## How does OpenAI's language model affect our AI timeline estimates?

2019-02-15T03:11:51.779Z · score: 51 (16 votes)
Comment by jimrandomh on Who owns OpenAI's new language model? · 2019-02-14T20:04:37.975Z · score: 13 (5 votes) · LW · GW

That's not how copyright works. US copyright law is civil, not criminal, which means that the US government can't act on infringements of its own initiative. A web site owner could theoretically sue OpenAI claiming that OpenAI infringed their copyright, but they'd probably lose, for a number of reasons.

Comment by jimrandomh on New York Restaurants I Love: Breakfast · 2019-02-14T19:34:58.519Z · score: 2 (1 votes) · LW · GW

This is an auto-crosspost of https://thezvi.wordpress.com/2019/02/14/new-york-restaurants-i-love-breakfast/ . Due to a server migration in progress, the auto-crossposting seems to have not worked right; it should be fixed soon.

Comment by jimrandomh on [Link] Did AlphaStar just click faster? · 2019-01-30T18:07:13.776Z · score: 4 (2 votes) · LW · GW

I think the right way to frame this is that AlphaStar has done the Starcraft 2 equivalent of mastering blitz chess. Blitz chess is easier for computers than slow chess because humans don't cope well with time pressure. Starcraft 2 imposes fairly extreme time pressure by default; humans need to make multiple actions per second, which means they're choosing between multi-action templates rather than optimizing each action separately. An interesting challenge for AI would be the Starcraft equivalent of slow chess: Starcraft played at 1/10th speed.

Comment by jimrandomh on Vote counting bug? · 2019-01-23T23:19:22.935Z · score: 6 (3 votes) · LW · GW

This is now in our issue tracker here.

Comment by jimrandomh on Three Worlds Collide (0/8) · 2019-01-15T20:34:42.176Z · score: 2 (1 votes) · LW · GW

Looks like someone ePub-ified it here .

Comment by jimrandomh on Reframing Superintelligence: Comprehensive AI Services as General Intelligence · 2019-01-09T02:39:13.408Z · score: 2 (1 votes) · LW · GW

The restriction on having multiple linkposts to the same URL is something we inherited from our framework (Vulcan), which doesn't particularly make sense for LW. We've taken it out, so you'll be able to make the linkpost after the next time we deploy an update (which will be later this week).

Comment by jimrandomh on [deleted post] 2019-01-01T23:39:29.172Z

Test comment

Comment by jimrandomh on Good arguments against "cultural appropriation" · 2018-12-19T00:44:15.922Z · score: 8 (5 votes) · LW · GW

Some CA is seen as a kind of theft of intellectual property

It's worth mentioning here that Intellectual Property is codified in law, that the version of it that is codified into law is restricted to a specific list of types of things, and that intellectual property existing at all is controversial, even within those codified categories. If "Cultural Appropriation is bad" is framed as an extension of the preexisting debate about intellectual property, it's solidly outside the Overton window: it would mean drastically increasing the scope of things which can be IP.

Comment by jimrandomh on LW Update 2018-12-06 – Table of Contents and Q&A · 2018-12-08T02:02:57.653Z · score: 6 (4 votes) · LW · GW

the animated scroll-to-section is rather too slow for my taste

I agree. The catch is, the browser API we're using (window.scrollTo) doesn't have a speed option, so we have to either live with its choice of speed, or implement the animation ourselves.

Comment by jimrandomh on LW Update 2018-12-06 – Table of Contents and Q&A · 2018-12-08T02:00:34.891Z · score: 8 (6 votes) · LW · GW

I'm the one who implemented this. While it's partially about making it intuitive in LessWrong's GUI editor, the decision was mostly based on trying it out on historical posts, and seeing what seemed to work best. All of that content pre-dates our current ToC implementation, and some of it was imported from quite far back in time. (This is also why <h5> and <h6> aren't considered headings.)

While we could write a migration script and also modify the editor to save as headers, we're reluctant to do that because it could change the semantics of old imported content, and we're reluctant to invest in LessWrong's Draft-JS editor right now, since we're planning on replacing it with something better.

Comment by jimrandomh on How rapidly are GPUs improving in price performance? · 2018-11-26T18:01:18.341Z · score: 6 (3 votes) · LW · GW

Ooh, shiny dataset! This is a good complement to cpudb, which does something similar for CPUs (but whose data fields get kinda sparse after 2012).

The first release of CUDA is in 2007; prior to that GPGPU wasn't much of a thing. I think the extra-fast improvement from 2007 to 2012 represents the transition from game-graphics-oriented hardware to general-computing-oriented hardware.

Comment by jimrandomh on Incorrect hypotheses point to correct observations · 2018-11-20T23:37:06.635Z · score: 11 (6 votes) · LW · GW

This would be better if you waited for someone to write an actually-dismissive comment, rather than projecting dismissiveness onto people who haven't commented.

Comment by jimrandomh on [deleted post] 2018-11-14T21:02:59.975Z

Test comment on my draft.

## Introducing the AI Alignment Forum (FAQ)

2018-10-29T21:07:54.494Z · score: 88 (29 votes)
Comment by jimrandomh on The Bizarre Behavior of Berkeley Rationalists · 2018-10-20T17:41:25.086Z · score: 6 (7 votes) · LW · GW

If you think someone is hypomanic, or in the process of falling for quackery, or otherwise having something that looks like it is or could develop into a mental health issue, I'd recommend letting them know directly, or bringing it up in a 1:1 conversation with a mutual friend who might have some visibility into the situation. Publicly saying "I see this happening in unspecified anonymous people" is just going to make people nervous and defensive, without enabling any discussion or action.

Comment by jimrandomh on "Now here's why I'm punching you..." · 2018-10-19T17:57:22.644Z · score: 4 (2 votes) · LW · GW
I think I disagree with this. When a social sanction is born of a particular case, I think it is quite important to actually have that case as a part of the discussion.

Clarification: what I meant is that it's better if the rules are created in a context where there are no cases pending for the rules to bear on; ie, I'm not objecting to admitting that a rules-discussion is about a specific case that it would bear on, but to it actually being about a specific pending case.

I think also it's quite difficult for people to think about tradeoffs in the abstract; "should annoying people be allowed at meetups?" is different from "should we let Bob keep coming to meetups?", and generally the latter is a more productive question.

I think discussing the latter question is less likely to produce the right result, though.

Comment by jimrandomh on "Now here's why I'm punching you..." · 2018-10-16T22:02:09.703Z · score: 23 (15 votes) · LW · GW

The principle I would draw is not, "you should separate the meta- and object-level discussions". Rather, I think the important thing is that meta-level discussions of how social sanctions work shouldn't be generated by backward-chaining from an ambiguous case. If people think that your meta-level argument about what circumstances it's okay to punch people, is actually about whether to punch Bob, then Bob's allies and enemies will engage with that conversation in a biased way. But separately, if people think that you had some Vaguebob in mind that they didn't know about, and that you might be Vaguebob's friend or enemy, then they'll rightly suspect you of being biased in the same way.

Comment by jimrandomh on Where is my Flying Car? · 2018-10-15T22:50:21.085Z · score: 4 (2 votes) · LW · GW

There are a lot more hypotheses not mentioned here, mostly hypotheses in which innovation stagnated because of effects which made individuals less innovative (as opposed to affecting coordination). For example, the "television rots your brain" hypothesis: that people who grew up with access to television are less capable of innovating, in general. Or the same but for people who ate the low-fat diet that the government was promoting, or were exposed to too much lead as children, or lost their agency to oppressive schooling and didn't recover it.

Comment by jimrandomh on Things I Learned From Working With A Marketing Advisor · 2018-10-09T20:30:51.944Z · score: 22 (7 votes) · LW · GW
Business “jargon” and “buzzwords” are unfairly maligned by people who aren’t used to corporate culture. First of all, a lot of them originally referred to specific important concepts, and then got overused as generic applause lights — e.g. “disruptive innovation” is actually a really useful idea in its original meaning.  But, second of all, it’s honestly just handy to have stock phrases if you need to keep talking fluently without awkward pauses.

Many people respond well to it, but smooth-sounding-emptiness is a corruption. Having overlap between your set of really-useful-idea words and your stock-phrase-for-fluency words seems like it has the potential to be really bad for your long-term ability to think clearly.

Comment by jimrandomh on Moloch in whom I sit alone · 2018-10-04T01:38:26.460Z · score: 9 (5 votes) · LW · GW

Person enters the room, or exits a conversation. In descending order of preference, they can either (1) talk to someone who is sitting alone, (2) join an existing conversation, or (3) sit alone. But because people don't want to be the first person to seed a new conversation, there aren't many people by themselves to make 2-person conversations out of. So, to fix the incentives...

When you start a conversation with someone who was sitting alone, give them a dollar. (Not entirely serious but not entirely joking either.)

Comment by jimrandomh on Rationalism for the masses · 2018-10-02T19:54:56.354Z · score: 13 (4 votes) · LW · GW

I would recommend starting with more sophisticated and friendly audiences, the sort of people who would go to a Less Wrong meetup, before you try to pitch rationality to more mainstream audiences. It's more difficult than you would expect; rationality concepts generally rest on and depend on a foundation of analytical thinking, and this is far from universal. It's also hard to distinguish yourself from people pitching things that look superficially similar at first, but later veer into crazy politics or scams.

All that said, there's definitely value in creating more-accessible explanations of key concepts, and if you think you've got the writing skill, it's worth a shot.

## Boston-area Less Wrong meetup

2018-05-16T22:00:48.446Z · score: 4 (1 votes)

## Welcome to Cambridge/Boston Less Wrong

2018-03-14T01:53:37.699Z · score: 4 (2 votes)

## Meetup : Cambridge, MA Sunday meetup: Lightning Talks

2017-05-20T21:10:26.587Z · score: 0 (1 votes)

## Meetup : Cambridge/Boston Less Wrong: Planning 2017

2016-12-29T22:43:55.164Z · score: 0 (1 votes)

## Meetup : Boston Secular Solstice

2016-11-30T04:54:55.035Z · score: 1 (2 votes)

## Meetup : Cambridge Less Wrong: Tutoring Wheels

2016-01-17T05:23:05.303Z · score: 1 (2 votes)

## Meetup : MIT/Boston Secular Solstice

2015-12-03T01:14:02.376Z · score: 1 (2 votes)

## Meetup : Cambridge, MA Sunday meetup: The Contrarian Positions Game

2015-11-13T18:08:19.666Z · score: 1 (2 votes)

## Rationality Cardinality

2015-10-03T15:54:03.793Z · score: 21 (22 votes)

## An Idea For Corrigible, Recursively Improving Math Oracles

2015-07-20T03:35:11.000Z · score: 5 (5 votes)

## Research Priorities for Artificial Intelligence: An Open Letter

2015-01-11T19:52:19.313Z · score: 23 (24 votes)

## Petrov Day is September 26

2014-09-18T02:55:19.303Z · score: 24 (18 votes)

## Three Parables of Microeconomics

2014-05-09T18:18:23.666Z · score: 25 (35 votes)

## Meetup : LW/Methods of Rationality meetup

2013-10-15T04:02:11.785Z · score: 0 (1 votes)

## Cambridge Meetup: Talk by Eliezer Yudkowsky: Recursion in rational agents

2013-10-15T04:02:05.988Z · score: 7 (8 votes)

## Meetup : Cambridge, MA Meetup

2013-09-28T18:38:54.910Z · score: 4 (5 votes)

## Charity Effectiveness and Third-World Economics

2013-06-12T15:50:22.330Z · score: 7 (12 votes)

## Meetup : Cambridge First-Sunday Meetup

2013-03-01T17:28:01.249Z · score: 3 (4 votes)

## Meetup : Cambridge, MA third-Sunday meetup

2013-02-11T23:48:58.812Z · score: 3 (4 votes)

## Meetup : Cambridge First-Sunday Meetup

2013-01-31T20:37:32.207Z · score: 1 (2 votes)

## Meetup : Cambridge, MA third-Sunday meetup

2013-01-14T11:36:48.262Z · score: 3 (4 votes)

## Meetup : Cambridge, MA first-Sunday meetup

2012-11-30T16:34:04.249Z · score: 1 (2 votes)

## Meetup : Cambridge, MA third-Sundays meetup

2012-11-16T18:00:25.436Z · score: 3 (4 votes)

## Meetup : Cambridge, MA Sunday meetup

2012-11-02T17:08:17.011Z · score: 1 (2 votes)

## Less Wrong Polls in Comments

2012-09-19T16:19:36.221Z · score: 79 (82 votes)

## Meetup : Cambridge, MA Meetup

2012-07-22T15:05:10.642Z · score: 2 (3 votes)

## Meetup : Cambridge, MA first-Sundays meetup

2012-03-30T17:55:25.558Z · score: 0 (3 votes)

## Professional Patients: Fraud that ruins studies

2012-01-05T00:20:55.708Z · score: 16 (25 votes)

2011-12-23T19:54:22.907Z · score: 1 (1 votes)

## I started a blog: Concept Space Cartography

2011-12-16T21:06:28.888Z · score: 6 (9 votes)

## Meetup : Cambridge (MA) Saturday meetup

2011-10-20T03:54:28.892Z · score: 2 (3 votes)

## Another Mechanism for the Placebo Effect?

2011-10-05T01:55:11.751Z · score: 8 (22 votes)

## Meetup : Cambridge, MA Sunday meetup

2011-10-05T01:37:06.937Z · score: 1 (2 votes)

## Meetup : Cambridge (MA) third-Sundays meetup

2011-07-12T23:33:01.304Z · score: 0 (1 votes)

## Draft of a Suggested Reading Order for Less Wrong

2011-07-08T01:40:06.828Z · score: 26 (29 votes)

## Meetup : Cambridge Massachusetts meetup

2011-06-29T16:57:15.314Z · score: 1 (2 votes)

## Meetup : Cambridge Massachusetts meetup

2011-06-22T15:26:03.828Z · score: 2 (3 votes)

## The Present State of Bitcoin

2011-06-21T20:17:13.131Z · score: 7 (12 votes)

## Safety Culture and the Marginal Effect of a Dollar

2011-06-09T03:59:28.731Z · score: 23 (36 votes)

## Cambridge Less Wrong Group Planning Meetup, Tuesday 14 June 7pm

2011-06-08T03:41:41.375Z · score: 1 (2 votes)

## Rationality case study: How to evaluate untested medical procedures?

2011-05-28T11:17:17.349Z · score: 7 (8 votes)

## Ontological Crises in Artificial Agents' Value Systems by Peter de Blanc

2011-05-21T01:05:12.613Z · score: 15 (15 votes)

## Homomorphic encryption and Bitcoin

2011-05-19T01:07:14.192Z · score: 5 (10 votes)

## Extremely Important Cell Phone Feature Missing

2011-04-29T01:48:23.091Z · score: 3 (25 votes)

## Cambridge Less Wrong Meetups April 21, May 1, May 15, ...

2011-04-19T12:24:43.249Z · score: 5 (6 votes)