The content of this post is on the shared RBC blog.
This is my last post on the shared blog, because Rationality Mega-Camp / Boot Camp is done after this weekend! Further posts by me will be here.
posted at: 18:16 | path: /rationality/bootcamp | permanent link to this entry
The content of this post is on the shared RBC blog.
posted at: 22:34 | path: /rationality/bootcamp | permanent link to this entry
The content of this post is on the shared RBC blog. I wrote this on Jul 31, and forgot to link it here last week -- whoops.
posted at: 22:00 | path: /rationality/bootcamp | permanent link to this entry
The content of this post is on the shared RBC blog.
posted at: 19:42 | path: /rationality/bootcamp | permanent link to this entry
The content of this post is on the shared RBC blog.
posted at: 06:46 | path: /rationality/bootcamp | permanent link to this entry
The content of this post is on the shared RBC blog.
posted at: 01:35 | path: /rationality/bootcamp | permanent link to this entry
posted at: 16:00 | path: /rationality/bootcamp | permanent link to this entry
The content of this post is at the shared RBC blog, so that there can be comments.
posted at: 23:57 | path: /rationality/bootcamp | permanent link to this entry
posted at: 16:00 | path: /rationality/bootcamp | permanent link to this entry
One of the defining philosophies of the local Singularity Institute community here is "Munchkining at life". Where I come from we would call it min/maxing your life, or hacking your life. It refers to winning (making money or happiness) by unconventional or unexpected means. You're following the rules (laws and ethics) but not obeying societal expectations.
Eliezer Yudkowsky and Michael Vassar proclaim (loudly and to anyone who will listen) that there are a huge number of munchkining opportunities for anyone to take. Low-hanging fruit, in other words. They believe this strongly enough that they've told us that one of the goals of Rationality Boot Camp is to produce millionaires, by teaching us the skills we need to notice and take advantage of these munchkin opportunities.
If Michael and Eliezer are correct about the low-hanging fruit, and if they can successfully teach the appropriate skills, then it makes sense that SingInst is paying for us to be here: they expect to get a lot of big donations, either from us directly in a few years, or through us because we end up in powerful positions.
Anyway, we had a session about munchkining today, called "The World is Mad." Eliezer first polled us about how much weight we assigned to the implication that "nobody else is doing X" should mean that X is a bad idea. I gave the weight four out of ten, and the answers were quite varied. Eliezer then gave lots of examples of things people weren't doing which were obviously a good idea (and things which people were doing which were shown to be retarded):
For each of these we rated how much this caused us to believe the "world is mad" hypothesis. For me the most impressive one was the hospital checklist one, followed by prison rape. How much do each of these cause you to update?
I guess the experiment is to determine whether people can really internalize that "the world is mad". Eliezer and Michael were saying that they present people with all this data, and yet people still look at what others are doing to determine whether something is a good idea. I can see what they're saying, because despite all this, I haven't updated more than a couple points on that scale. I initially answered 4 out of 10; I would now put my estimation of "the world is mad" at about 6 or 7. I am still not fully updated to that view; it's hard for me to internalize.
Apparently one science fiction author, Jack Vance, treats this view as a skill, "zs'hanh", which is defined as "contemptuous indifference to the activity of others". Eliezer tweaked this definition to "contemptuous indifference to the inactivity of others. I expect to make a lot of progress learning this skill this summer.
posted at: 03:00 | path: /rationality/bootcamp | permanent link to this entry
Here's a metaphor we've been using a lot at Rationality Boot Camp: People's behavior is controlled by an elephant and a human rider -- the rider represents your rational, cognitive, high-level thought processes, whereas the elephant represents your emotions, habits, and unconscious thoughts. The metaphor works well: the proportions are about right, and when the elephant and the rider conflict, guess who wins.
One practical technique we've been studying is Nonviolent Communication (NVC), a buzzword-driven methodology for communicating with people, which focuses on ways to defuse upsetting situations. Someone called it "elephant whispering". The idea is that if someone is angry with you, or frustrated, or whatever, their elephant is going crazy, and you can help tame it so their rider can get back in control.
The main technique we're learning through NVC is empathy. You can derive utterances which will reliably calm an upset person down and engage their cognition. The elephant apparently responds very well to this sort of empathy. This is as opposed to a number of common ways people try to talk to upset people, which don't usually work as well. Don't do these:
Instead, the technique says to talk only about: "observations, feelings, needs, and requests." Observations are factual statements, "You left the light on last night" (what not to say: generalizations or evaluations, like "you always leave the light on").
Feelings are observations about emotions. Things like "angry", "confused", "disappointed", "frustrated", "sad", "embarrassed", etc. There are also positive feelings: "hopeful", "intrigued", "proud" and so on.
"Needs" refer to basic human needs which our elephant wants: self-worth, food, rest, love, creativity, order, respect, etc.
And requests are from one person to another, reasonable things to ask of another person, like "turn off the light before you sleep" or "knock before you enter".
You can leave off several of observations, feelings, needs, or requests if they don't apply. You can also take guesses as to the cause. If you successfully signal curiosity, the other person won't get more upset, even if you guessed wrong -- instead, it will trigger them to correct your guess, but in the process it will engage their rider and naturally defuse their upsetness.
For example, if your roommate stomps into the kitchen and won't talk to you: "Are you upset because the kitchen isn't clean?" Even if you're wrong, more likely than not, your roommate will figure out her reason for being upset, and should become less upset and more willing to talk.
There's a lot of good examples and information on wikihow's article.
My evaluation? When I imagine some of the confrontations I've had, I can easily imagine NVC working. The instructor, Divia, says that her interpersonal relationships have massively improved with the use of NVC. For these reasons I am inclined to believe that it works, with fairly high uncertainty until I have put some of this stuff into practice. In any case, let me know if you've had any experiences which would show that it works or doesn't work.
posted at: 03:58 | path: /rationality/bootcamp | permanent link to this entry
Apparently I've fallen behind slightly on my weekly update schedule. I planned to update yesterday, at the last minute (of course) and fell prey to the planning fallacy.
Eliezer showed up at the Tortuga Less Wrong meetup yesterday, and it was a blast.
Tortuga is a communal living group in Mountain View. The Boot Camp is in Berkeley, which isn't exactly close to Mountain View; it's extremely inconvenient and expensive to get there via public transport. Fortunately some people had cars we could pile into.
There were probably 30 or 40 people there. We did an exercise called "The Strangest Thing an AI Could Tell You", where we came up with the strangest thing that an AI could tell you where you believed it, rather than believing that the AI was broken or you were going crazy.
One of the interesting questions we came up with is: "why does the AI output this particular statement?" Did you ask it "do all humans have a tail that we just can't see" and it said yes? Or did you ask it "tell me something surprising" and it produced this gem? The question doesn't specify, and it seems that the answers are quite different depending on the process by which the AI would have generated the question.
We spent a while talking about this, and then moved onto random other topics, and it was really a blast. More than half the attendees stuck around for at least four hours. Eventually some people went to hang out in the naked hot tub in the back. I went, of course, and Eliezer also went, and now I have had the privilege and bragging rights of playing Truth or Dare in a hot tub naked with Eliezer Yudkowsky.
posted at: 17:53 | path: /rationality/bootcamp | permanent link to this entry
I am learning how to memorize things!
Our instructor Geoff today taught us memorization techniques which really work and have a huge and obvious effect. The two techniques are called "pegs" and "memory palace".
I'd heard of memory palace before. I hadn't heard of pegs. They're both fairly ancient techniques, so nothing groundbreaking here, but they are highly effective and easy to learn.
Today, we first did a pre-test to measure memorization before trying to learn any techniques. Geoff read us a list of twenty random objects and we had to try to recall the objects in the order given. I did remember eight objects correctly -- the first four, the last two, and a couple in the middle which stood out.
After the pretest, we started working on memorization techniques. I chose pegs.
The idea with pegs is that you use normal flash-card style memorization techniques to bind numbers to objects. For me, one is "hat" and two is "hen" and three is "ham". These objects can be reused many times, so you only have to memorize the list once. When you're actually faced with a list of things to memorize, you just "hang" each of the memorizees on their associated pegs. So if Geoff reads "1. Zipper," you add a zipper to your hat. Later you just think of the number 1, which leads you to think of a hat, which leads you to remember the zipper on the hat.
The peg objects can be anything you want, but I followed the Major System, which assigns major consonants to each digit. So 1 is 'd' or 't', 2 is 'n', 3 is 'm'. You can look up the list on Wikipedia, but the idea is that you come up with a word you'll remember whose only major consonants are the digits of the number you're trying to remember. (Vowels and 'w','h','y' are not major consonants for this purpose). This appears to be a very scalable system and I expect to derive good use from it.
Anyway, I was able to correctly memorize nine things (instead of 8) in the post-test, but I was able to recall twelve things two hours later, and I can tell I would have done much better if I had been better at remembering my pegs. I'm going to spend time each day practicing my pegs until they're second nature. I would expect to get 19 or 20 out of 20 once I am quickly able to come up with the peg image for a given number.
posted at: 01:02 | path: /rationality/bootcamp | permanent link to this entry
I should point out that John Faben is also writing about Rationality Boot Camp.
posted at: 01:57 | path: /rationality/bootcamp | permanent link to this entry
There are a lot of people in our little house!
The house is at Shattuck and Essex in southern Berkeley. There's a commercial retail space on the first floor, as well as a tiny one-bedroom apartment. On each of the second and third floors there's a four-bedroom apartment.
We have rented everything except the second floor. The windows of the commercial space are papered over, so nobody thinks there's a business there, and the space is used for our morning and afternoon sessions, as well as Less Wrong meetups and such. But most social activity that isn't sessions goes on in the four-bedroom apartment, which is a little too small to really have a comfortable space for hanging-out-and-talking. That activity mostly happens in the kitchen and living room.
The kitchen is about the size of a typical bedroom, and it has two refrigerators and a large table in it, so it's a bit cramped. There are only a few chairs in the kitchen, so if people are talking there, most are standing up, and people are talking across the table.
The living room has a bay window and five doors off of it. It's easy to fit a bunch of people in here to chat. There are at least nine chairs and a mattress on the floor. It could probably fit four mattresses side-by-side if the chairs were moved. It's better for talking than the kitchen, but so far, more social activity has gone on in the kitchen, probably because it's easier to start conversations when preparing or eating food. It's hard to eat in the living room since there's no tables, so people usually eat standing around in the kitchen.
This group of folks is a bunch of exceptionally well-adjusted nerds. Many study computer science or math. Many love tabletop roleplaying, video games and/or board games. All are smart, learn quickly and want to learn faster. 90% are within the 20-30 age group. The average social ability and physical attractiveness of our group seems above average for the 20-30 age group.
Pretty much everyone here will participate in ad-hoc "calibration experiments": if a question of fact comes up in a conversation, before looking it up, the participants produce predictions. If it's a binary question of fact, then you just give your uncertainty level, expressed as a probability. If it's a numeric value, then a confidence interval is more appropriate. People usually use a 90% or 95% confidence interval. Finally, someone looks up the answer -- if it was outside the confidence interval, then your calibration takes a hit. You are on target if your calibration takes a hit only once every ten times you give a 90% confidence interval. So far I've been wrong about 3 or 4 times out of maybe 15 predictions, so I'm not doing so hot and I am trying to adjust my calibration.
Besides the people who live here, lots of Singularity Institute-affiliated people visit all the time. Michael Vassar is here right now; Eliezer Yudkowsky was here yesterday, and Anna Salamon the day before. There are two other SingInst communal houses nearby in Berkeley, called Slytherin and Outpost. I haven't visited these yet but I expect to this summer. Besides the 12 or so who live here regularly, during the day there are usually 5-10 SingInst affiliates who wander through the house and talk with people. These people rarely sleep at the apartment, though there are some who do on occasion.
People like to drink and smoke weed, though this has not happened with much intensity. We went through a decent amount of beer fairly quickly last week, but this week we haven't had any alcohol and our conversations have been of similar quality. 1:30am seems to be the median time by which everyone has crashed, and I'm one of the people who usually stays up until everyone is going to bed. I usually get up just before 9:30am, when we usually meditate and make breakfast.
Send me any other questions you have about the social atmosphere. I can probably answer them.
posted at: 20:35 | path: /rationality/bootcamp | permanent link to this entry
Yesterday for fun we did parkour lessons at "Athletic Playground," an indoor studio which has parkour, aerials and a bunch of other interesting-looking classes. I unexpectedly obtained status by being quite good at it, despite never having done any parkour training before. (Just my own enjoyment of jumping on things, hanging from things, and so on.)
posted at: 18:05 | path: /rationality/bootcamp | permanent link to this entry
This session was very long and split into several parts. It was given by a math researcher who practiced self-modification hacks for at least ten years. His stories were mostly about hacks which he had applied to himself, but he presented them in a generally applicable way.
For this lecture I highly recommend you go to Mega-Camp or Mini-Camp next year, where he will probably show up and teach these tricks. However, I'll give a brief overview. Basic rationality skills are needed: specifically recognizing rationalization vs. genuine curiosity.
The first example was that he found that grading was annoying and he never wanted to do it. To change this and enjoy it more, he constructed a desire to grade student homework in himself. This he achieved by breaking down this "want to desire to grade" into several steps. Basically, in order to desire to grade, he had to fan tiny sparks of random desire to grade into strong desires. He did this using reinforcement learning, by rewarding himself every time he noticed a desire to grade. In order to produce more sparks of desire, he also rewarded certain forms of thinking which could lead to desires to grade: thinking about grading, and noticing other desires (like desires for ice cream).
And yes, this process apparently worked. I'm going to apply this to taking actions which achieve my goals, and discourage actions which are procrastinatory.
The next thing which he presented which seemed useful is a "goal sheet": a worksheet you should fill out often, in order to review your current short, medium, and long-term goals, and evaluate whether you are working on them in a sensible and progress-making way.
Here are the questions on a goal sheet:
There was another especially useful exercise, the failure-mode inoculation exercise. Failure-mode inoculation involves placing yourself into a failure mode you anticipate running into for a given goal, which would typically make you abandon the goal.. For me it was running out of steam on my coding project. I run out of steam when I find that there's nothing "fun" left, only bugfixing, and there are no users who actually want to use the product.
I imagined myself into this failure mode, which wasn't very hard. Then I imagined how someone might try to talk me out of it. In this case it was imagining a lot of people using my product, and visualizing what kinds of people they were, and how they would find out about the product, and so on. It appears to be possible to talk oneself out of a hypothetical failure mode a lot more easily than with a real one. Now, the assertion that the speaker made was that you can actually practice getting in and out of failure modes by using Anki. Essentially, you create flashcards which prompt with a project and a failure mode, and when it comes up, you succeed at the card if you were able to think yourself into the failure mode, then think yourself out of it successfully. In theory, and if your success strategies are accurate, you will be able to quickly apply the strategy you figured out for getting rid of the failure mode.
The last part of the talk was where the speaker presented a method he executed in order to deal with a bunch of personal life issues: sick parents, breakups, mental health and so on. It would be a spoiler to present the method here, but it involved going away for three days without contact with anyone -- no family or friends -- and letting his unconscious mind do the work of identifying and learning to solve the problems. And no mind-altering drugs were used either! :)
posted at: 08:20 | path: /rationality/bootcamp | permanent link to this entry
Yesterday we did Rejection Therapy!
For us, what rejection therapy means is that you go downtown, wander around by yourself, and ask random strangers to do you favors.
Examples would be asking for their newspaper, or asking for a few dollars, or a phone number, or to try on their shoes. Or going into a store and asking for some free food.
Now, I'm not a very courageous person, especially when it comes to asking strangers for things. So when I heard we were doing rejection therapy, I was so nervous that I was shaking. But I knew that getting rid of this anxiety is exactly the point of rejection therapy. We walked for about 15 minutes before we made it downtown. I noticed that my anxiety made me very alert and I noticed a lot more things about a lot more people -- for example, whether they were carrying something that I could ask for, or wearing something I could try on. Of course, the exercise hadn't officially started yet, which is how I rationalized not approaching these opportunities I saw. During this walk I also brainstormed questions you could ask people.
Once we got downtown and the exercise officially began, I achieved slightly above my own expectations, which were fairly low. I approached maybe 12-15 people total, in an hour and a half. (Some of the others reported approaching two or three times that number.)
I asked for a number of newspapers. I asked to take the plastic flowers from a storefront. I asked to borrow board games from the board game store. I asked to climb a ladder that a construction worker was setting up -- and he let me!
I asked a couple eating lunch on the grass if they would meditate with me for 5 minutes. They didn't want to, but they said "after lunch maybe." I went back 20 minutes later and I was too shy to actually ask again, partly because they were cuddling and being rather private and I didn't want to disturb them. But that's mostly just a rationalization.
I asked a girl if I could try on her sunglasses and she said "what is this, some kind of psychology class experiment?" Apparently someone else in her group had recently been approached. I was a bit too flustered to ask her for anything else.
I noticed two good effects from the experiment. The first was that I was a LOT less shy once I got into the swing of things. I was still shy about talking to people for any length of time, but approach anxiety was greatly reduced. I was able to ask for something which I legitimately wanted: I smelled malt outside a brewpub, walked in, saw they were brewing, and asked the head brewer for a tour. (He didn't give me one for insurance reasons, but I got an invitation to come on a Saturday and he would show me around.) This is something I would never or rarely do before, and during Rejection Therapy, I felt a lot more comfortable doing it. I expect this level of comfort to decline somewhat, but just knowing that I can do it is a big step forward in terms of comfort.
Later that day, Anna Salamon gave a talk where she talked about one of the goals of rationality as "building affordances" in everyday life. This is an analogy to physical affordances, like a cup affords drinking and a knob affords turning and a button affords pressing. However, Anna is talking about mental affordances -- having flags go up in your brain when you notice that a situation matches a pattern, and knowing how to react in such a pattern. So the other effect from the rejection therapy experiment was that I noticed myself building the affordance that "strangers afford talking" -- that now I was noticing situations where I could ask strangers for things, and reasons to talk to them. I expect this effect will decline only a little bit.
posted at: 16:25 | path: /rationality/bootcamp | permanent link to this entry
First session today was led by Jasen. Meditation can refer to many things. So far, we've been talking about a specific skill which Jasen calls "noting," as in taking note of sensations. The first exercise was to focus on a single sensation -- in my case, the sensation of air moving past my nostrils while breathing. Continually note that sensation, and when you find yourself distracted, note the form of distraction (thinking, pain, etc.) and return to focusing on your nostrils.
The ostensible benefit of such an endeavor is that it will improve your ability to control what you are focusing on, and that controlling focus is incredibly important to all aspects of life.
One example: apparently, "suffering" is, according to some, just a failure to appropriately allocate your attention to the things causing you pain. So if you learn to allocate attention in this way, then you can greatly reduce your suffering and improve the quality of your life. Jasen directly claims this benefit: at some point after he started meditation practice, there was one meditation session after which about 80% of his personal suffering disappeared, and people commented that he seemed happier. Here is the article which states that suffering is an attention-allocational conflict. Jasen does not claim that this is "enlightenment" or anything like that (though he does talk about enlightenment, but has not claimed to achieve it).
Anyway, the benefits sound pretty good to me. These claimed benefits are the sorts which I would completely dismiss if they didn't come from someone who has extensively studied the subject, and whose rationality I trust. As is, I'm cautiously uncertain, but willing to try hard to achieve the benefits.
posted at: 02:59 | path: /rationality/bootcamp | permanent link to this entry
We haven't done any exercises yet; today was spent socializing. Brief update, but I'll give you some idea of what we've been chatting about: organizational stuff (schedules and such); what to do with the ground floor (it's nice retail space but right now the windows are just papered up, so I want a coffee shop); how each individual found Less Wrong and why they stayed and why they applied; our object-level goals; Bitcoin; mathematics; drugs (modafinil, marijuana, salvia...); what the world will be like after the Singularity; tech startups; what to rename Rationality Boot Camp to (we wanted a higher-status-sounding-thing for our resumes).
Tomorrow: intro games (icebreakers?); "meditation and decision-theoretic enlightenment" lesson; "emotional awareness" lesson.
If you're interested in making sure you don't miss updates on this please add my blog to your RSS reader.
posted at: 07:56 | path: /rationality/bootcamp | permanent link to this entry