So, the Center for Applied Rationality just ran another workshop, which Anna kindly invited me to. Below I've written down some thoughts on it, both to organize those thoughts and because it seems other LWers might want to read them. I'll also invite other participants to write down their thoughts in the comments. Apologies if what follows isn't particularly well-organized. 

Feelings and other squishy things

The workshop was totally awesome. This is admittedly not strong evidence that it accomplished its goals (cf. Yvain's comment here), but being around people motivated to improve themselves and the world was totally awesome, and learning with and from them was also totally awesome, and that seems like a good thing. 

Also, the venue was fantastic. CFAR instructors reported that this workshop was more awesome than most, and while I don't want to discount improvements in CFAR's curriculum and its selection process for participants, I think the venue counted for a lot. It was uniformly beautiful and there were a lot of soft things to sit down or take naps on, and I think that helped everybody be more comfortable with and relaxed around each other. 

Main takeaways

Here are some general insights I took away from the workshop. Some of them I had already been aware of on some abstract intellectual level but hadn't fully processed and/or gotten drilled into my head and/or seen the implications of. 

  1. Epistemic rationality doesn't have to be about big things like scientific facts or the existence of God, but can be about much smaller things like the details of how your particular mind works. For example, it's quite valuable to understand what your actual motivations for doing things are. 
  2. Introspection is unreliable. Consequently, you don't have direct access to information like your actual motivations for doing things. However, it's possible to access this information through less direct means. For example, if you believe that your primary motivation for doing X is that it brings about Y, you can perform a thought experiment: imagine a world in which Y has already been brought about. In that world, would you still feel motivated to do X? If so, then there may be reasons other than Y that you do X. 
  3. The mind is embodied. If you consistently model your mind as separate from your body (I have in retrospect been doing this for a long time without explicitly realizing it), you're probably underestimating the powerful influence of your mind on your body and vice versa. For example, dominance of the sympathetic nervous system (which governs the fight-or-flight response) over the parasympathetic nervous system is unpleasant, unhealthy, and can prevent you from explicitly modeling other people. If you can notice and control it, you'll probably be happier, and if you get really good, you can develop aikido-related superpowers
  4. You are a social animal. Just as your mind should be modeled as a part of your body, you should be modeled as a part of human society. For example, if you don't think you care about social approval, you are probably wrong, and thinking that will cause you to have incorrect beliefs about things like your actual motivations for doing things. 
  5. Emotions are data. Your emotional responses to stimuli give you information about what's going on in your mind that you can use. For example, if you learn that a certain stimulus reliably makes you angry and you don't want to be angry, you can remove that stimulus from your environment. (This point should be understood in combination with point 2 so that it doesn't sound trivial: you don't have direct access to information like what stimuli make you angry.) 
  6. Emotions are tools. You can trick your mind into having specific emotions, and you can trick your mind into having specific emotions in response to specific stimuli. This can be very useful; for example, tricking your mind into being more curious is a great way to motivate yourself to find stuff out, and tricking your mind into being happy in response to doing certain things is a great way to condition yourself to do certain things. Reward your inner pigeon.

Here are some specific actions I am going to take / have already taken because of what I learned at the workshop. 

  1. Write a lot more stuff down. What I can think about in my head is limited by the size of my working memory, but a piece of paper or a WorkFlowy document don't have this limitation. 
  2. Start using a better GTD system. I was previously using RTM, but badly. I was using it exclusively from my iPhone, and when adding something to RTM from an iPhone the due date defaults to "today." When adding something to RTM from a browser the due date defaults to "never." I had never done this, so I didn't even realize that "never" was an option. That resulted in having due dates attached to RTM items that didn't actually have due dates, and it also made me reluctant to add items to RTM that really didn't look like they had due dates (e.g. "look at this interesting thing sometime"), which was bad because that meant RTM wasn't collecting a lot of things and I stopped trusting my own due dates. 
  3. Start using Boomerang to send timed email reminders to future versions of myself. I think this might work better than using, say, calendar alerts because it should help me conceptualize past versions of myself as people I don't want to break commitments to. 

I'm also planning to take various actions that I'm not writing above but instead putting into my GTD system, such as practicing specific rationality techniques (the workshop included many useful worksheets for doing this) and investigating specific topics like speed-reading and meditation. 

The arc word (TVTropes warning) of this workshop was "agentiness." ("Agentiness" is more funtacular than "agency.") The CFAR curriculum as a whole could be summarized as teaching a collection of techniques to be more agenty. 


A distinguishing feature the people I met at the workshop seemed to have in common was the ability to go meta. This is not a skill which was explicitly mentioned or taught (although it was frequently implicit in the kind of jokes people told), but it strikes me as an important foundation for rationality: it seems hard to progress with rationality unless the thought of using your brain to improve how you use your brain, and also to improve how you improve how you use your brain, is both understandable and appealing to you. This probably eliminates most people as candidates for rationality training unless it's paired with or maybe preceded by meta training, whatever that looks like.

One problem with the workshop was lack of sleep, which seemed to wear out both participants and instructors by the last day (classes started early in the day and conversations often continued late into the night because they were unusually fun / high-value). Offering everyone modafinil or something at the beginning of future workshops might help with this.


Overall, while it's too soon to tell how big an impact the workshop will have on my life, I anticipate a big impact, and I strongly recommend that aspiring rationalists attend future workshops. 

New Comment
73 comments, sorted by Click to highlight new comments since:
Some comments are truncated due to high volume. (⌘F to expand all)Change truncation settings

[Note: I also attended the workshop and am writing this before reading the main post to avoid being biased.]

The out of class discussions alone were more than worth my while. However this is partially due to my circumstances, I am about to graduate college with a double major in CS and psychology. Deciding which (if either) field to pursue, and how to best pursue it is obviously quite important to me. Before going to the workshop, I had already done most of the obvious “due diligence” research on the matter, but still updated significantly during the workshop. Talking to several of the other CfAR participants, the instructors, and several LW people who stopped by (Lukeprog, Yvain ect.) uncovered many important points that I had missed. For example, I assumed the median staring salary for computer scientists was a reasonable estimate for what my starting salary would be. It turns out that I can expect to make about twice that much money if I use certain job hunting techniques I learned at the workshop and optimize for money (instead of, say, cool sounding problems). I don't expect most people to be in similar positions, but I do think asking questions at a CfAR event is a ver... (read more)

Upvoted for writing this before reading the main post. Is that the sort of technique they taught you?

To repeat my request to Qiaochu: would you be willing to follow up in a year or two and tell us whether the effects lasted? If you commit now, and use Boomerang to remind yourself, I would be grateful.

It's related to a point that Julia made in her "Evaluating Advice" unit about contamination of evidence. The idea is to avoid anchoring people to a particular answer when you ask them questions. See also this xkcd comic.
Upvoted out of interest. I try to do this myself, but it never seems to work as well as I expect it should; without an anchor, people frequently just try to guess what I want to hear instead of outputting their own opinion. Or at least I think they do. It is inherently difficult to verify that sort of thing.
For boring reasons I don't feel like using Boomerang to remind myself, but if you message me in a year (and I still have an active LW, and insert other obvious caveats here) I'll write the post and/or post significant comments on the retrospectives of other attendees. ETA Short answer yes. Long answer, it is the obvious converse of some of Julia's techniques for soliciting useful advice on the other hand they did not directly suggest it for giving good advice, and I suspect I would have thought to do it even if I had missed that session. Its a pretty obvious idea (at least to me).
I have set a Boomerang reminder to myself. I will ask you in a year. Thanks.
What changed your expectation of your starting salary?
Talking to a large number of computer scientists at the workshop, and deciding I'd be fine with, among other things, working in finance in the general San Francisco /Silicon valley area.
Is that a better idea than trying to get a market-rate salary with a significant chunk of equity at an early-stage startup that your unusually rational brain has suggested to you has an unusually large chance of making it big?
I'm not sure. I just assign very high probability to it being better than going into academic psychology or the sort of CS jobs I had been considering in my hometown. Do you have some reason to believe I should try the start-up route?
I would point out that a psych/CS person could be really useful to a startup. Having more than just coding skills can be huge, and if anything you know about psych is related to behaviour, then you're golden. I'd like to point to this article about instagram by Nir Eyal: "But at its core, Instagram is the latest example of an enterprising team, conversant in psychology as much as technology, that unleashed an addictive product on users who made it part of their daily routines."
Do you mean as a founder, an early hire, or both? I don't really have many start up concepts on hand.
Either/both. Someone who can just code is limited to that range of experience. Someone who can code and knows psych can not just code, but also use their knowledge of psych to shape product design decisions (e.g. Instagram) or even design software to solve problems in the field of psychology (either psychological problems people have or problems psychologists have). You get way more options by having not just one niche.
Well, there's this. The startup thing is probably better for someone like me who has spent a ton of time thinking about what sort of startup is likely to be successful. I was just wondering if you had any thoughts on it. BTW, it's worth noting that there are probably also ways to strategically maximize your income as a psychologist.
A standard way to make more money as a psychology professor is to get a job in a business school instead of a psychology department. It's worth maybe a factor of 2. * Dan (beoShaffer, we talked some about jobs at the workshop, but the business school option didn't come up. PM/email me if you want to talk more.)

Thank you for writing up your thoughts, Qiaochu!

I feel the need to offer one very minor correction:

For example, sympathetic nervous system activity, which governs the fight-or-flight response, is unpleasant, unhealthy, and can prevent you from explicitly modeling other people.

It's actually sympathetic dominance over the parasympathetic side that does this. Both the sympathetic and parasympathetic systems are running all the time, and that's really quite essential. You cannot stand up without the sympathetic system, for instance, nor can you lie down without going into a panic without the parasympathetic side doing its job.

But as long as you replace "activity" with "dominance," I think we're good!

Thanks! Corrected.
I think you may now have the definitions mixed up. It looks like you're saying the parasympathetic system governs fight or flight, when it's the sympathetic system that runs fight or flight.
The phrase is "sympathetic nervous system dominance over the parasympathetic nervous system."
In its current form, I couldn't figure out what the sentence was trying to say either. Would it still be correct to say, "For example, dominance of the sympathetic nervous system (which governs the fight-or-flight response) over the parasympathetic, is unpleasant, ..."?
That's probably less confusing. Thanks! Edited.
Based on this, and your linked comment, and some quick reading of the Wikipedia articles, I'm wondering if I have an issue with my sympathetic nervous system dominating my parasympathetic nervous system. Possibly in a fairly serious way. Is there more reading I can do on this? Techniques or tests to see what's going on or try to change it? My suspicion is that stress/fight or fight reaction is preventing my body from properly going into maintenance and repair mode.

I was waiting for Val to answer this, but I'll give it a shot. The relevant CFAR unit is called "againstness." You can think of sympathetic dominance as related to (being?) a sensation of "againstness," e.g. when you get angry during a heated argument your feelings are directed against the person you're arguing with. Val gave us both mental and physical techniques for releasing againstness ("fighting againstness" is kind of againsty). The mental techniques (which I'm just going to quote verbatim from the worksheet; hopefully Val won't mind):

  • Gratitude: appreciate the opportunity to practice releasing againstness in adverse situations. (Also, appreciate the pleasure of PNS activation to help stabilize it.)
  • When againstness is related to others:
    • Empathy: model the other person in enough detail to appreciate likely reasons why what they're doing is, to them, the most sensible thing they could be doing in that moment.
    • Connection: intentionally increase your sense of tribal togetherness; see the other person as a fellow human being and wish to help them realize their full potential.

That might sound a little woo but the above is intended to be a descri... (read more)

Well said! I'll just add that the book What Every Body Is Saying is quite good at illustrating what various ways of addressing SNS dominance look like. The author frames it in terms of "limbic activity," but it's basically the same beast. There are a few details in that book I'm not convinced of (e.g., I haven't been able to use people's feet as an indicator of their honest intentions), but the majority is quite good and can make it easier for you to look for important cues that others are sending you about their states.
Thanks for taking the time to put together such a complete response. Lots for me to think about and read up on here.
The statement as he made it is correct, you are not correcting it but making a narrower, also correct, version of his statement. (I think I need to gzip that sentence.)
Really? I'm not so sure. Sympathetic nervous system activity isn't necessarily unpleasant or unhealthy. It's what makes you sweat and have a fast heart rate when exercising. It's very unhealthy not to have that! But if others interpret Qiaochu's comment in a way that accurately reflects reality, then the fact that it's different than how I read it is not terribly relevant. I was only concerned that some here might come away with a mistaken idea of what the message of that particular CFAR unit was. If there are no such mistakes, then all is good!

Also see The Power of Agency.

And for those who want to pronounce Qiaochu's name correctly in their heads when reading his username, I think it's "Chow-chew."

Or as I like to introduce it sometimes, "chow like food and chew also like food."

Is it Mandarin? I would say it sort of like "Chee-ow choo". We also need some tones & the 汉字 for good measure.

Yep. "Qiao" is second tone and "chu" is third tone. The characters are 翘楚.

Well said.

We CFAR workshop participants got workbooks containing materials for the various sessions (and some other things, like the rationality habits checklist). I took my notes right in the workbook. Sometime today or tomorrow I'm going to start going through the workbook and transferring the stuff I want to capture to the computer. I've started a list of things to think about while doing so. Below, "this" means "this tool, concept, exercise, worked example of an exercise, personal failure mode, etc.".

  • Do I want to put this into my spaced repetition system?
  • Do I want to do this in the future? If so, how can I make that happen? (E.g., schedule a time or put it in my to-do list system.)
  • How much can I improve at this, and how valuable would improving at this be?

I've made a couple other comments about the workshop: thank-you to my hosts (I showed up a day before and stayed a few days after); impressions of the workshop written for my family and friends.

Offering everyone modafinil or something at the beginning of future workshops might help with this.

It would help, but would inevitably offend people and not at all worth the consequences.

We couldn't afford them this year for cost reasons, but by next year we'll hopefully be able to supply Time-Turners for all workshop participants.

That should have some interesting effects on the prediction markets.
The meta ones will destroy time/space.
Maybe they already did and that's why we don't have Time-Turners anymore.
I already had to hide one market with all of you operating in only one direction in time!
What market did you have to hide?
P(The last bet on this market will be >50%) Big incentive to be the last one to bet, and, since I didn't want to make it a dangerous scrum, I hid it 3 hours before closing and made it a scavenger hunt instead.
This only solves the problem if you also provide modafinil.
What other reasons are there for not being able to afford things?

I was at the workshop, brought my own (legally obtained) modafinil but never used nor needed it.

I loved the workshop. I intend on teaching my game theory students (an 80 person class at Smith College) some of what I learned. Today we did 15 minutes on goal factoring.

I found that pre-comiting to go to bed at certain time (I offered to remind others on my way out), and setting a timer on my phone solved the sleep problem.

So, I wouldn't call this a solution to the sleep problem. The sleep problem as I conceive it is the following: staying up during a CFAR workshop has unusually high value because of the people you can talk to, but on the other hand sleep is important for learning things. It's not obvious to me how to optimally handle this tradeoff and I'd like a way of handling it that involves decreasing the need for sleep rather than increasing the quantity of sleep if this is feasible.

My solution was running on adrenaline at the workshop (bed at 2-3am, up at 7:15) and then getting sick when I got home. Same way I did hell week for every piece of theatre I was in. So, uh, better suggestions welcome.
I do pretty much that every single week.
Yes, it was an opportunity cost problem - at what point did the cost of being cogent in the morning outweigh the cost of missing great late night conversations. I can't think of any optimal solution that doesn't involve loads of caffeine or bilocation, time turner induced or otherwise.
Shortly after the July workshop last year, I adopted the polyphasic every man 3 schedule (3 hour core sleep with 3 20 minute naps). It seems to be working for me. Though others who tried it have given it up. I suspect the every man 1 schedule(~6 hour core with 1 20-30 minute nap) might be more generally achievable. During the workshop itself, I just got less sleep (like 4 to 6 hours per night), and this was OK because, as near as I can tell, I was only deprived of slow wave sleep, which takes a week to cause problems. Though I think the only person to get less sleep than me was Matt, who was already polyphasic.
What about my previous suggestions to measure something, anything, like spaced repetition or dual n-back?
I'm curious as to why caffiene wasn't sufficient, but also why modafinil would offend people? What about trying bright lighting?:

I'm curious as to why caffiene wasn't sufficient, but also why modafinil would offend people?

As a schedule IV drug, it's surely some sort of crime to offer or accept. Some people will not want to associate with such people or organizations on moral grounds, risk-aversion grounds, or fear of other people's disapproval on either ground etc.

Ah, I thought it was an over the counter drug.

It is, some places. Just not the USA where CFAR is operating now and the foreseeable future. I'm a big fan of modafinil as you might guess, but if CFAR were even idly considering providing or condoning modafinil use, I'd smack them silly (metaphorically); organizations must obey different standards than individuals.

I agree that they should uphold strict standards for numerous reasons. That doesn't prevent CFAR from discussing potential benefits (and side effects) of different drugs (caffeine, aspirin, modafinil, etc.). They could also recommend discussing such things with a person's doctor as well as what criteria are used to prescribe such drugs (they might already for all I know).

My current stance, which I'll push for quite strongly unless and until I encounter enough evidence against to update significantly, is that CFAR would do very poorly to talk explicitly about any drugs that the USA has a neurosis about. We can talk at a layer of abstraction above: "How might you go about determining what kinds of effects a given substance has on you?" But I am pretty solidly against CFAR listing potential benefits and drawbacks of any drugs that have become rallying cries for law enforcement or political careers.

CFAR folk, please consider/research NYC as a location.

Logistics might be different of course, but there is a lot of people here who wouldn't have to travel or need sleeping arrangements, lowering the total time+money cost for them without taking away CFAR revenue (since they are not a hotel/travel agency).

Also, these workshops are entrepreneur-oriented and there is an active and geographically concentrated start-up scene here, in which I'm somewhat active and would be happy to help with promotion and maybe some of the logistics.

We really are considering it. Some much more key things have emerged that will keep our focus mostly in the Bay area for the next six or so months, but New York City is definitely on the table for a CFAR event this year. No promises, of course, but we very much care about the LW community and know that there's a rather huge core of it in NYC.

Without making concrete plans, I'd like to offer some ideas off-line, PM me?
Better yet, could you email me? I'd love to talk about this, but I find I check Less Wrong infrequently enough that PMs here are not a reliable way for me to keep up with communications. My email is valentine at appliedrationality dot org. Thanks!

This is useful to me as I'll be attending the March workshop. If I successfully digest any of the insights presented here then I'll have a better platform to start from. (Two particular points are the stuff about the parasympathetic nervous system, which I'd basically never heard of before, and the connection between the concepts of "epistemic rationality" and "knowing about myself" which is more obvious-in-retrospect).

Thanks for the write-up!

And yes, I'll stick up at least a brief write-up of my own after I'm done. Does LW have an anti-publication-bias registry somewhere?

Not that I know of, but that does sound quite awesome. I look forward to meeting you, Giles!
Any advice on how to set one up? In particular how to add entries to it retrospectively - I was thinking about searching the comments database for things like "I intend to", "guard against", "publication bias" etc. and manually find the relevant ones. This is somewhat laborious, but the effect I want to avoid is "oh I've just finished my write-up (or am just about to), now I'll go and add the original comment to the anti-publication bias registry". On the other hand it seems like anyone can safely add anyone else's comment to the registry as long as it's close enough in time to when the comment was written. Any advice? (I figured if you're involved at CFAR you might know a bit about this stuff).
A reasonable assumption, but alas, false! I don't think I have anything useful to add to this. Sorry!

while it's too soon to tell how big an impact the workshop will have on my life

Would you be willing to follow up in, say, two years and tell us how well these changes lasted? If you commit now, and use Boomerang to remind yourself, I would be grateful.

I anticipate a big impact within a year and have used Boomerang to remind myself to do this accordingly. (Edit: it's worth mentioning that part of why I anticipate a big impact is that I'm local to Berkeley and have more direct access to CFAR than most.)
I am grateful!

For the GTD stuff, I use emacs + org-mode + .emacs based on this configuration + mobile org.

Since I try to work exclusively in emacs, I can quickly capture notes and "things that need to get done" in their proper context, all of which is aggregated under an Agenda window. The Agenda window manages a collection of ".org" files which store the specific details of everything. MobileOrg syncs all these .org files to my phone. Combined with the GTD philosophy of never having anything uncategorized bouncing around in my mind, this system work... (read more)

I'm glad to hear it is working well and is well received!

Once there has been some experience running these workshops I really hope there is something that CFAR can design for meetup groups to try / implement and/or an online version.

Is there a CFAR webpage that covers this particular workshop and how it went?

This is definitely on our horizon. Not yet. I'm not sure putting it on our website is the right thing to do either. We might send out a summary in our newsletter, though. You can subscribe to it by clicking the letter icon at the top of our website.

tricking your mind into being more curious

that sounds very useful. How exactly does this technique work?

Stolen from the relevant CFAR worksheet: avoid environments or experiences that limit your ability to feel curious (for me these include time limits, extreme temperatures, and bad smells). Think back to a specific and recent time when you felt curious, try to remember what it felt like, and focus on that feeling. Think of a person (possibly fictional) who is frequently or intensely curious (for me this is MoR!Harry) and pretend to be them. Think about the kind of problems that you feel curious about and try to make the problem you're working on feel more like those problems (e.g. by identifying some conceptual feature they have in common).

Offering everyone modafinil or something at the beginning of future workshops might help with this.

Probably better to offer 1/4 of the participants modafinil or something to see if affected more than just wakefulness. E.g. a friend with lots of modafinil experience reports that modafinil may "make it too easy to work enthusiastically on slightly the wrong thing".