Less Wrong is a community blog devoted to refining the art of human rationality. Please visit our About page for more information.

Kickstarting the audio version of the upcoming book "The Sequences"

26 Rick_from_Castify 16 December 2014 01:01AM

LessWrong is getting ready to release an actual book that covers most of the material found in the Sequences. 

There have been a few posts about it in the past, here are two: the title debate, content optimization.

We've been asked if we'd like to produce the audiobook version and the answer is yes. This is a large undertaking. The finished product will probably be over 35 hours of audio.

To help mitigate our risk we've decided to Kickstarter the audiobook.  This basically allows us to pre-sell it so we're not stuck with a large production cost and no revenue. 

The kickstarter campaign is here: https://www.kickstarter.com/projects/1267969302/lesswrong-the-sequences-audiobook

If you haven't heard of us before we've already produced some sequences into audiobooks.  You can see them and listen to samples which are indicative of the audio quality here.

Rationality Quotes December 2014

6 Salemicus 03 December 2014 10:33PM

Another month, another rationality quotes thread. The rules are:

  • Please post all quotes separately, so that they can be upvoted or downvoted separately. (If they are strongly related, reply to your own comments. If strongly ordered, then go ahead and post them together.)
  • Do not quote yourself.
  • Do not quote from Less Wrong itself, HPMoR, Eliezer Yudkowsky, or Robin Hanson. If you'd like to revive an old quote from one of those sources, please do so here.
  • No more than 5 quotes per person per monthly thread, please.
  • Provide sufficient information (URL, title, date, page number, etc.) to enable a reader to find the place where you read the quote, or its original source if available. Do not quote with only a name.

The Bay Area Solstice

18 Alex_Altair 03 December 2014 10:33PM

 

As the holiday season approaches, we continue our tradition of celebrating the winter solstice.

This event is the offspring of Raemon's New York Solstice. The core of the event is a collection of songs old and new, silly and profound, led by the well-calibrated Bayesian choir. There will be bean bag chairs and candles. There will be campfire and chocolates (in case of dementors).

When: The Bay Area Solstice will be held on 13 December at 7:00 PM.

Where: We've rented the Humanist Hall, at 390 27th St, Oakland, CA 94612.

All humanists or transhumanists are welcome. We'll be diving our minds into the nature of the universe, both good and bad. We'll stare into the abyss of death, and into the radiance of our ability to remove it. We will recognize each other as allies and agents.

We're glad to provide aspiring rationalists with an alternative or addition to any holiday celebrations. There is an expected attendance of around 80 people. 

Get your tickets here! And if you'd like to help us put it together, PM me.

MIRI Research Guide

52 So8res 07 November 2014 07:11PM

We've recently published a guide to MIRI's research on MIRI's website. It overviews some of the major open problems in FAI research, and provides reading lists for those who want to get familiar with MIRI's technical agenda.

This guide updates and replaces the MIRI course list that started me on the path of becoming a MIRI researcher over a year ago. Many thanks to Louie Helm, who wrote the previous version.

This guide is a bit more focused than the old course list, and points you not only towards prerequisite textbooks but also towards a number of relevant papers and technical reports in something approximating the "appropriate order." By following this guide, you can get yourself pretty close to the cutting edge of our technical research (barring some results that we haven't written up yet). If you intend to embark on that quest, you are invited to let me know; I can provide both guidance and encouragement along the way.

I've reproduced the guide below. The canonical version is at intelligence.org/research-guide, and I intend to keep that version up to date. This post will not be kept current.

Finally, a note on content: the guide below discusses a number of FAI research subfields. The goal is to overview, rather than motivate, those subfields. These sketches are not intended to carry any arguments. Rather, they attempt to convey our current conclusions to readers who are already extending us significant charity. We're hard at work producing a number of documents describing why we think these particular subfields are important. (The first was released a few weeks ago, the rest should be published over the next two months.) In the meantime, please understand that the research guide is not able nor intended to provide strong motivation for these particular problems.


Friendly AI theory currently isn't about implementation, it's about figuring out how to ask the right questions. Even if we had unlimited finite computing resources and a solid understanding of general intelligence, we still wouldn't know how to specify a system that would reliably have a positive impact during and after an intelligence explosion. Such is the state of our ignorance.

For now, MIRI's research program aims to develop solutions that assume access to unbounded finite computing power, not because unbounded solutions are feasible, but in the hope that these solutions will help us understand which questions need to be answered in order to the lay the groundwork for the eventual specification of a Friendly AI. Hence, our current research is primarily in mathematics (as opposed to software engineering or machine learning, as many expect).

This guide outlines the topics that one can study to become able to contribute to one or more of MIRI’s active research areas.

continue reading »

Rationality Quotes November 2014

8 elharo 07 November 2014 07:07PM

Another month, another rationality quotes thread. The rules are:

  • Please post all quotes separately, so that they can be upvoted or downvoted separately. (If they are strongly related, reply to your own comments. If strongly ordered, then go ahead and post them together.)
  • Do not quote yourself.
  • Do not quote from Less Wrong itself, HPMoR, Eliezer Yudkowsky, or Robin Hanson. If you'd like to revive an old quote from one of those sources, please do so here.
  • No more than 5 quotes per person per monthly thread, please.
  • Provide sufficient information (URL, title, date, page number, etc.) to enable a reader to find the place where you read the quote, or its original source if available. Do not quote with only a name.

2014 Less Wrong Census/Survey

87 Yvain 26 October 2014 06:05PM

It's that time of year again.

If you are reading this post and self-identify as a LWer, then you are the target population for the Less Wrong Census/Survey. Please take it. Doesn't matter if you don't post much. Doesn't matter if you're a lurker. Take the survey.

This year's census contains a "main survey" that should take about ten or fifteen minutes, as well as a bunch of "extra credit questions". You may do the extra credit questions if you want. You may skip all the extra credit questions if you want. They're pretty long and not all of them are very interesting. But it is very important that you not put off doing the survey or not do the survey at all because you're intimidated by the extra credit questions.

It also contains a chance at winning a MONETARY REWARD at the bottom. You do not need to fill in all the extra credit questions to get the MONETARY REWARD, just make an honest stab at as much of the survey as you can.

Please make things easier for my computer and by extension me by reading all the instructions and by answering any text questions in the simplest and most obvious possible way. For example, if it asks you "What language do you speak?" please answer "English" instead of "I speak English" or "It's English" or "English since I live in Canada" or "English (US)" or anything else. This will help me sort responses quickly and easily. Likewise, if a question asks for a number, please answer with a number such as "4", rather than "four".

The planned closing date for the survey is Friday, November 14. Instead of putting the survey off and then forgetting to do it, why not fill it out right now?

Okay! Enough preliminaries! Time to take the...

***


[EDIT: SURVEY CLOSED, DO NOT TAKE!]

***

Thanks to everyone who suggested questions and ideas for the 2014 Less Wrong Census/Survey. I regret I was unable to take all of your suggestions into account, because of some limitations in Google Docs, concern about survey length, and contradictions/duplications among suggestions. The current survey is a mess and requires serious shortening and possibly a hard and fast rule that it will never get longer than it is right now.

By ancient tradition, if you take the survey you may comment saying you have done so here, and people will upvote you and you will get karma.

On Caring

88 So8res 15 October 2014 01:59AM

This is an essay describing some of my motivation to be an effective altruist. It is crossposted from my blog. Many of the ideas here are quite similar to others found in the sequences. I have a slightly different take, and after adjusting for the typical mind fallacy I expect that this post may contain insights that are new to many.

1

I'm not very good at feeling the size of large numbers. Once you start tossing around numbers larger than 1000 (or maybe even 100), the numbers just seem "big".

Consider Sirius, the brightest star in the night sky. If you told me that Sirius is as big as a million earths, I would feel like that's a lot of Earths. If, instead, you told me that you could fit a billion Earths inside Sirius… I would still just feel like that's a lot of Earths.

The feelings are almost identical. In context, my brain grudgingly admits that a billion is a lot larger than a million, and puts forth a token effort to feel like a billion-Earth-sized star is bigger than a million-Earth-sized star. But out of context — if I wasn't anchored at "a million" when I heard "a billion" — both these numbers just feel vaguely large.

I feel a little respect for the bigness of numbers, if you pick really really large numbers. If you say "one followed by a hundred zeroes", then this feels a lot bigger than a billion. But it certainly doesn't feel (in my gut) like it's 10 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 times bigger than a billion. Not in the way that four apples intenally feels like twice as many as two apples. My brain can't even begin to wrap itself around this sort of magnitude differential.

This phenomena is related to scope insensitivity, and it's important to me because I live in a world where sometimes the things I care about are really really numerous.

For example, billions of people live in squalor, with hundreds of millions of them deprived of basic needs and/or dying from disease. And though most of them are out of my sight, I still care about them.

The loss of a human life with all is joys and all its sorrows is tragic no matter what the cause, and the tragedy is not reduced simply because I was far away, or because I did not know of it, or because I did not know how to help, or because I was not personally responsible.

Knowing this, I care about every single individual on this planet. The problem is, my brain is simply incapable of taking the amount of caring I feel for a single person and scaling it up by a billion times. I lack the internal capacity to feel that much. My care-o-meter simply doesn't go up that far.

And this is a problem.

continue reading »

2014 iterated prisoner's dilemma tournament results

59 tetronian2 30 September 2014 09:23PM

Followup to: Announcing the 2014 program equilibrium iterated PD tournament

In August, I announced an iterated prisoner's dilemma tournament in which bots can simulate each other before making a move. Eleven bots were submitted to the tournament. Today, I am pleased to announce the final standings and release the source code and full results.

All of the source code submitted by the competitors and the full results for each match are available here. See here for the full set of rules and tournament code.

Before we get to the final results, here's a quick rundown of the bots that competed:

AnderBot

AnderBot follows a simple tit-for-tat-like algorithm that eschews simulation:

  • On the first turn, Cooperate.
  • For the next 10 turns, play tit-for-tat.
  • For the rest of the game, Defect with 10% probability or Defect if the opposing bot has defected more times than AnderBot.

continue reading »

The Future of Humanity Institute could make use of your money

52 danieldewey 26 September 2014 10:53PM

Many people have an incorrect view of the Future of Humanity Institute's funding situation, so this is a brief note to correct that; think of it as a spiritual successor to this post. As John Maxwell puts it, FHI is "one of the three organizations co-sponsoring LW [and] a group within the University of Oxford's philosophy department that tackles important, large-scale problems for humanity like how to go about reducing existential risk." (If you're not familiar with our work, this article is a nice, readable introduction, and our director, Nick Bostrom, wrote Superintelligence.) Though we are a research institute in an ancient and venerable institution, this does not guarantee funding or long-term stability.

Academic research is generally funded through grants, but because the FHI is researching important but unusual problems, and because this research is multi-disciplinary, we've found it difficult to attract funding from the usual grant bodies. This has meant that we’ve had to prioritise a certain number of projects that are not perfect for existential risk reduction, but that allow us to attract funding from interested institutions.

With more assets, we could both liberate our long-term researchers to do more "pure Xrisk" research, and hire or commission new experts when needed to look into particular issues (such as synthetic biology, the future of politics, and the likelihood of recovery after a civilization collapse).

We are not in any immediate funding crunch, nor are we arguing that the FHI would be a better donation target than MIRI, CSER, or the FLI. But any donations would be both gratefully received and put to effective use. If you'd like to, you can donate to FHI here. Thank you!

Simulate and Defer To More Rational Selves

114 BrienneStrohl 17 September 2014 06:11PM

I sometimes let imaginary versions of myself make decisions for me.

I first started doing this after Anna told me (something along the lines of) this story. When she first became the executive director of CFAR, she suddenly had many more decisions to deal with per day than ever before. "Should we hire this person?" "Should I go buy more coffee for the coffee machine, or wait for someone else deal with it?" "How many participants should be in our first workshop?" "When can I schedule time to plan the fund drive?" 

I'm making up these examples myself, but I'm sure you, too, can imagine how leading a brand new organization might involve a constant assault on the parts of your brain responsible for making decisions. She found it exhausting, and by the time she got home at the end of the day, a question like, "Would you rather we have peas or green beans with dinner?" often felt like the last straw. "I don't care about the stupid vegetables, just give me food and don't make me decide any more things!"

She was rescued by the following technique. When faced with a decision, she'd imagine "the Executive Director of CFAR", and ask herself, "What would 'the Executive Director of CFAR' do?" Instead of making a decision, she'd make a prediction about the actions of that other person. Then, she'd just do whatever they'd do!

(I also sometimes imagine what Anna would do, and then do that. I call it "Annajitsu".)

In Anna's case, she was trying to reduce decision fatigue. When I started trying it out myself, I was after a cure for something slightly different.

Imagine you're about to go bungee jumping off a high cliff. You know it's perfectly safe, and all you have to do is take a step forward, just like you've done every single time you've ever walked. But something is stopping you. The decision to step off the ledge is entirely yours, and you know you want to do it because this is why you're here. Yet here you are, still standing on the ledge. 

You're scared. There's a battle happening in your brain. Part of you is going, "Just jump, it's easy, just do it!", while another part--the part in charge of your legs, apparently--is going, "NOPE. Nope nope nope nope NOPE." And you have this strange thought: "I wish someone would just push me so I don't have to decide."

Maybe you've been bungee jumping, and this is not at all how you responded to it. But I hope (for the sake of communication) that you've experienced this sensation in other contexts. Maybe when you wanted to tell someone that you loved them, but the phrase hovered just behind your lips, and you couldn't get it out. You almost wished it would tumble out of your mouth accidentally. "Just say it," you thought to yourself, and remained silent. For some reason, you were terrified of the decision, and inaction felt more like not deciding.

When I heard this story from Anna, I had social anxiety. I didn't have way more decisions than I knew how to handle, but I did find certain decisions terrifying, and was often paralyzed by them. For example, this always happened if someone I liked, respected, and wanted to interact with more asked to meet with them. It was pretty obvious to me that it was a good idea to say yes, but I'd agonize over the email endlessly instead of simply typing "yes" and hitting "send".

So here's what it looked like when I applied the technique. I'd be invited to a party. I'd feel paralyzing fear, and a sense of impending doom as I noticed that I likely believed going to the party was the right decision. Then, as soon as I felt that doom, I'd take a mental step backward and not try to force myself to decide. Instead, I'd imagine a version of myself who wasn't scared, and I'd predict what she'd do. If the party really wasn't a great idea, either because she didn't consider it worth my time or because she didn't actually anticipate me having any fun, she'd decide not to go. Otherwise, she'd decide to go. I would not decide. I'd just run my simulation of her, and see what she had to say. It was easy for her to think clearly about the decision, because she wasn't scared. And then I'd just defer to her.

Recently, I've noticed that there are all sorts of circumstances under which it helps to predict the decisions of a version of myself who doesn't have my current obstacle to rational decision making. Whenever I'm having a hard time thinking clearly about something because I'm angry, or tired, or scared, I can call upon imaginary Rational Brienne to see if she can do any better.

Example: I get depressed when I don't get enough sunlight. I was working inside where it was dark, and Eliezer noticed that I'd seemed depressed lately. So he told me he thought I should work outside instead. I was indeed a bit down and irritable, so my immediate response was to feel angry--that I'd been interrupted, that he was nagging me about getting sunlight again, and that I have this sunlight problem in the first place. 

I started to argue with him, but then I stopped. I stopped because I'd noticed something. In addition to anger, I felt something like confusion. More complicated and specific than confusion, though. It's the feeling I get when I'm playing through familiar motions that have tended to lead to disutility. Like when you're watching a horror movie and the main character says, "Let's split up!" and you feel like, "Ugh, not this again. Listen, you're in a horror movie. If you split up, you will die. It happens every time." A familiar twinge of something being not quite right.

But even though I noticed the feeling, I couldn't get a handle on it. Recognizing that I really should make the decision to go outside instead of arguing--it was just too much for me. I was angry, and that severely impedes my introspective vision. And I knew that. I knew that familiar not-quite-right feeling meant something was preventing me from applying some of my rationality skills. 

So, as I'd previously decided to do in situations like this, I called upon my simulation of non-angry Brienne. 

She immediately got up and went outside.

To her, it was extremely obviously the right thing to do. So I just deferred to her (which I'd also previously decided to do in situations like this, and I knew it would only work in the future if I did it now too, ain't timeless decision theory great). I stopped arguing, got up, and went outside. 

I was still pissed, mind you. I even felt myself rationalizing that I was doing it because going outside despite Eliezer being wrong wrong wrong is easier than arguing with him, and arguing with him isn't worth the effort. And then I told him as much over chat. (But not the "rationalizing" part; I wasn't fully conscious of that yet.)

But I went outside, right away, instead of wasting a bunch of time and effort first. My internal state was still in disarray, but I took the correct external actions. 

This has happened a few times now. I'm still getting the hang of it, but it's working.

Imaginary Rational Brienne isn't magic. Her only available skills are the ones I have in fact picked up, so anything I've not learned, she can't implement. She still makes mistakes. 

Her special strength is constancy

In real life, all kinds of things limit my access to my own skills. In fact, the times when I most need a skill will very likely be the times when I find it hardest to access. For example, it's more important to consider the opposite when I'm really invested in believing something than when I'm not invested at all, but it's much harder to actually carry out the mental motion of "considering the opposite" when all the cognitive momentum is moving toward arguing single-mindedly for my favored belief.

The advantage of Rational Brienne (or, really, the Rational Briennes, because so far I've always ended up simulating a version of myself that's exactly the same except lacking whatever particular obstacle is relevant at the time) is that her access doesn't vary by situation. She can always use all of my tools all of the time.

I've been trying to figure out this constancy thing for quite a while. What do I do when I call upon my art as a rationalist, and just get a 404 Not Found? Turns out, "trying harder" doesn't do the trick. "No, really, I don't care that I'm scared, I'm going to think clearly about this. Here I go. I mean it this time." It seldom works.

I hope that it will one day. I would rather not have to rely on tricks like this. I hope I'll eventually just be able to go straight from noticing dissonance to re-orienting my whole mind so it's in line with the truth and with whatever I need to reach my goals. Or, you know, not experiencing the dissonance in the first place because I'm already doing everything right.

In the mean time, this trick seems pretty powerful.

View more: Next