SotW: Check Consequentialism
(The Exercise Prize series of posts is the Center for Applied Rationality asking for help inventing exercises that can teach cognitive skills. The difficulty is coming up with exercises interesting enough, with a high enough hedonic return, that people actually do them and remember them; this often involves standing up and performing actions, or interacting with other people, not just working alone with an exercise booklet and a pencil. We offer prizes of $50 for any suggestion we decide to test, and $500 for any suggestion we decide to adopt. This prize also extends to LW meetup activities and good ideas for verifying that a skill has been acquired. See here for details.)
Exercise Prize: Check Consequentialism
In philosophy, "consequentialism" is the belief that doing the right thing makes the world a better place, i.e., that actions should be chosen on the basis of their probable outcomes. It seems like the mental habit of checking consequentialism, asking "What positive future events does this action cause?", would catch numerous cognitive fallacies.
For example, the mental habit of consequentialism would counter the sunk cost fallacy - if a PhD wouldn't really lead to much in the way of desirable job opportunities or a higher income, and the only reason you're still pursuing your PhD is that otherwise all your previous years of work will have been wasted, you will find yourself encountering a blank screen at the point where you try to imagine a positive future outcome of spending another two years working toward your PhD - you will not be able to state what good future events happen as a result.
Or consider the problem of living in the should-universe; if you're thinking, I'm not going to talk to my boyfriend about X because he should know it already, you might be able to spot this as an instance of should-universe thinking (planning/choosing/acting/feeling as though within / by-comparison-to an image of an ideal perfect universe) by having done exercises specifically to sensitize you to should-ness. Or, if you've practiced the more general skill of Checking Consequentialism, you might notice a problem on asking "What happens if I talk / don't talk to my boyfriend?" - providing that you're sufficiently adept to constrain your consequentialist visualization to what actually happens as opposed to what should happen.
Discussion:
The skill of Checking Consequentialism isn't quite as simple as telling people to ask, "What positive result do I get?" By itself, this mental query is probably going to return any apparent justification - for example, in the sunk-cost-PhD example, asking "What good thing happens as a result?" will just return, "All my years of work won't have been wasted! That's good!" Any choice people are tempted by seems good for some reason, and executing a query about "good reasons" will just return this.
The novel part of Checking Consequentialism is the ability to discriminate "consequentialist reasons" from "non-consequentialist reasons" - being able to distinguish that "Because a PhD gets me a 50% higher salary" talks about future positive consequences, while "Because I don't want my years of work to have been wasted" doesn't.
It's possible that asking "At what time does the consequence occur and how long does it last?" would be useful for distinguishing future-consequences from non-future-consequences - if you take a bad-thing like "I don't want my work to have been wasted" and ask "When does it occur, where does it occur, and how long does it last?", you will with luck notice the error.
Learning to draw cause-and-effect directed graphs, a la Judea Pearl and Bayes nets, seems like it might be helpful - at least, Geoff was doing this while trying to teach strategicness and the class seemed to like it.
Sometimes non-consequentialist reasons can be rescued as consequentialist ones. "You shouldn't kill because it's the wrong thing to do" can be rescued as "Because then a person will transition from 'alive' to 'dead' in the future, and this is a bad event" or "Because the interval between Outcome A and Outcome B includes the interval from Fred alive to Fred dead."
On a five-second level, the skill would have to include:
- Being cued by some problem to try looking at the consequences;
- Either directly having a mental procedure that only turns up consequences, like trying to visualize events out into the future, or
- First asking 'Why am I doing this?' and then looking at the justifications to check if they're consequentialist, perhaps using techniques like asking 'How long does it last?', 'When does it happen?', or 'Where does it happen?'.
- Expending a small amount of effort to see if a non-consequentialist reason can easily translate into a consequentialist one in a realistic way.
- Making the decision whether or not to change your mind.
- If necessary, detaching from the thing you were doing for non-consequentialist reasons.
In practice, it may be obvious that you're making a mistake as soon as you think to check consequences. I have 'living in the should-universe' or 'sunk cost fallacy' cached to the point where as soon as I spot an error of that pattern, it's usually pretty obvious (without further deliberative thought) what the residual reasons are and whether I was doing it wrong.
Pain points & Pluses:
(When generating a candidate kata, almost the first question we ask - directly after the selection of a topic, like 'consequentialism' - is, "What are the pain points? Or pleasure points?" This can be errors you've made yourself and noticed afterward, or even cases where you've noticed someone else doing it wrong, but ideally cases where you use the skill in real life. Since a lot of rationality is in fact about not screwing up, there may not always be pleasure points where the skill is used in a non-error-correcting, strictly positive context; but it's still worth asking each time. We ask this question right at the beginning because it (a) checks to see how often the skill is actually important in real life and (b) provides concrete use-cases to focus discussion of the skill.)
Pain points:
Checking Consequentialism looks like it should be useful for countering:
- Living in the should-universe (taking actions because of the consequences they ought to have, rather than the consequences they probably will have). E.g., "I'm not going to talk to my girlfriend because she should already know X" or "I'm going to become a theoretical physicist because I ought to enjoy theoretical physics."
- The sunk cost fallacy (choosing to prevent previously expended, non-recoverable resources from having been wasted in retrospect - i.e., avoiding the mental pain of reclassifying a past investment as a loss - rather than acting for the sake of future considerations). E.g., "If I give up on my PhD, I'll have wasted the last three years."
- Cached thoughts and habits; "But I usually shop at Whole Foods" or "I don't know, I've never tried an electric toothbrush before." (These might have rescuable consequences, but as stated, they aren't talking about future events.)
- Acting-out an emotion - one of the most useful pieces of advice I got from Anna Salamon was to find other ways to act out an emotion than strategic choices. If you're feeling frustrated with a coworker, you might still want to Check Consequentialism on "Buy them dead flowers for their going-away party" even though it seems to express your frustration.
- Indignation / acting-out of morals - "Drugs are bad, so drug use ought to be illegal", where it's much harder to make the case that countries which decriminalized marijuana experienced worse net outcomes. (Though it should be noted that you also have to Use Empiricism to ask the question 'What happened to other countries that decriminalized marijuana?' instead of making up a gloomy consequentialist prediction to express your moral disapproval.)
- Identity - "I'm the sort of person who belongs in academia."
- "Trying to do things" for simply no reason at all, while your brain still generates activities and actions, because nobody ever told you that behaviors ought to have a purpose or that lack of purpose is a warning sign. This habit can be inculcated by schoolwork, wanting to put in 8 hours before going home, etc. E.g. you "try to write an essay", and you know that an essay has paragraphs; so you try to write a bunch of paragraphs but you don't have any functional role in mind for each paragraph. "What is the positive consequence of this paragraph?" might come in handy here.
(This list is not intended to be exhaustive.)
Pleasure points:
- Being able to state and then focus on a positive outcome seems like it should improve motivation, at least in cases where the positive outcome is realistically attainable to a non-frustrating degree and has not yet been subject to hedonic adaptation. E.g., a $600 job may be more motivating if you visualize the $600 laptop you're going to buy with the proceeds.
Also, consequentialism is the foundation of expected utility, which is the foundation of instrumental rationality - this is why we're considering it as an early unit. (This is not directly listed as a "pleasure point" because it is not directly a use-case.)
Constantly asking about consequences seems likely to improve overall strategicness - not just lead to the better of two choices being taken from a fixed decision-set, but also having goals in mind that can generate new perceived choices, i.e., improve the overall degree to which people do things for reasons, as opposed to not doing things or not having reasons. (But this is a hopeful eventual positive consequence of practicing the skill, not a use-case where the skill is directly being applied.)
Teaching & exercises:
This is the part that's being thrown open to Less Wrong generally. Hopefully I've described the skill in enough detail to convey what it is. Now, how would you practice it? How would you have an audience practice it, hopefully in activities carried out with each other?
The dumb thing I tried to do previously was to have exercises along the lines of, "Print up a booklet with little snippets of scenarios in them, and ask people to circle non-consequentialist reasoning, then try to either translate it to consequentialist reasons or say that no consequentialist reasons could be found." I didn't do that for this exact session, but if you look at what I did with the sunk cost fallacy, it's the same sort of silly thing I tried to do.
This didn't work very well - maybe the exercises were too easy, or maybe it was that people were doing it alone, or maybe we did something else wrong, but the audience appeared to experience insufficient hedonic return. They were, in lay terms, unenthusiastic.
At this point I should like to pause, and tell a recent and important story. On Saturday I taught an 80-minute unit on Bayes's Rule to an audience of non-Sequence-reading experimental subjects, who were mostly either programmers or in other technical subjects, so I could go through the math fairly fast. Afterward, though, I was worried that they hadn't really learned to apply Bayes's Rule and wished I had a small little pamphlet of practice problems to hand out. I still think this would've been a good idea, but...
On Wednesday, I attended Andrew Critch's course at Berkeley, which was roughly mostly-instrumental LW-style cognitive-improvement material aimed at math students; and in this particular session, Critch introduced Bayes's Theorem, not as advanced math, but with the aim of getting them to apply it to life.
Critch demonstrated using what he called the Really Getting Bayes game. He had Nisan (a local LWer) touch an object to the back of Critch's neck, a cellphone as it happened, while Critch faced in the other direction; this was "prior experience". Nisan said that the object was either a cellphone or a pen. Critch gave prior odds of 60% : 40% that the object was a cellphone vs. pen, based on his prior experience. Nisan then asked Critch how likely he thought it was that a cellphone or a pen would be RGB-colored, i.e., colored red, green, or blue. Critch didn't give exact numbers here, but said he thought a cellphone was more likely to be primary-colored, and drew some rectangles on the blackboard to illustrate the likelihood ratio. After being told that the object was in fact primary-colored (the cellphone was metallic blue), Critch gave posterior odds of 75% : 25% in favor of the cellphone, and then turned around to look.
Then Critch broke up the class into pairs and asked each pair to carry out a similar operation on each other: Pick two plausible objects and make sure you're holding at least one of them, touch it to the other person while they face the other way, prior odds, additional fact, likelihood ratio, posterior odds.
This is the sort of in-person, hands-on, real-life, and social exercise that didn't occur to me, or Anna, or anyone else helping, while we were trying to design the Bayes's Theorem unit. Our brains just didn't go in that direction, though we recognized it as embarrassingly obvious in retrospect.
So... how would you design an exercise to teach Checking Consequentialism?
Loading…
Subscribe to RSS Feed
= f037147d6e6c911a85753b9abdedda8d)
Comments (311)
Cleverness-related failure mode (that actually came up in the trial unit):
One shouldn't try too hard to rescue non-consequentialist reasons. This probably has to be emphasized especially with new audiences who associate "rationality" to Spock and university professors, or audiences who've studied pre-behavioral economics, and who think they score extra points if they come up with amazingly clever ways to rescue bad ideas.
Any decision-making algorithm, no matter how stupid, can be made to look like expected utility maximization through the transform "Assign infinite negative utility to departing from decision algorithm X". This in essence is what somebody is doing when they say, "Aha! But if I stop my PhD program now, I'll have the negative consequence of having abandoned a sunk cost!" (Sometimes I feel like hitting people with a wooden stick when they do this, but that act just expresses an emotion rather than having any discernible positive consequences.) This is Cleverly Failing to Get the Point if "not wanting to abandon a sunk cost", i.e., the counterintuitive feel of departing from the brain's previous decision algorithm, is treated as an overriding consideration, i.e., an infinite negative utility.
It's a legitimate future consequence only if the person says, "The sense of having abandoned a sunk cost will make me feel sick to my stomach for around three days, after which I would start to adjust and adapt a la the hedonic treadmill". In this case they have weighed the intensity and the duration of the future hedonic consequence, rather than treating it as an instantaneous infinite negative penalty, and are now ready to trade that off against other and probably larger considerations like the total amount of work required to get a PhD.
One of the other models people have for the rationalizing sort of "rationality" is that of lawyers.
Lawyers are very good at logic — the LSAT, the entrance examination for U.S. law schools, leans heavily on logic puzzles — but the whole point of being a trial or appeals lawyer is to come up with clever (and socially respectable) arguments for whatever position your client may have at the moment.
This extends past real-world lawyerhood. The tabletop role-playing game crowd have the expression "rules lawyer" for a person who comes up with clever arguments for why their character should get away with whatever they want to at the moment.
Rationalization is an important skill and should be rewarded, not punished. If you never try to rationalize others' decisions then you won't notice when they actually do have a good justification, and if you never practice rationalization then you'll never get good enough at it to find their justifications when they exist. The result is gross overconfidence in the stupidity of the opposing side and thus gross overconfidence in one's own rationality. That leads to tragedies and atrocities, both personal and societal.
Perspective-taking is a separate "skill" from rationalizing one's own behavior.
Hm, is perspective-taking the same skill that I was talking about? I can't tell. Also I thought that Eliezer's examples were phrased in the hypothetical, and thus it'd be rationalizing others' beliefs/behavior, not one's own. I'm not sure to what extent rationalizing a conclusion and rationalizing one's own behavior are related. Introspectively, the defensiveness and self-justifying-ness inherent to the latter makes it a rather different animal.
"Coming up with explanations" is a good skill.
"Coming up with a single, stupid explanation, failing to realize it is stupid, and then using it as an excuse to cease all further thought" is a very, very bad skill.
Thinking "well, but abandoning a sunk cost actually IS a negative future event" is smart IFF you then go "I'd be miserable for three days. How does that weigh against years spent in the program?"
It's very, very bad, however, if you stop there and continue to spend 2 years on a PhD just because you don't want to even THINK about those three days of misery.
I think understanding this dichotomy is critical. If you stop even thinking "well, but abandoning a sunk cost IS a negative future event" because you're afraid of falling in to the trap of then avoiding all sunk costs, then you're ignoring real negative consequences to your decisions.
My normal response is, "so what's bad about that?" and go a few rounds until the person has to struggle for an answer... the teachable moment where I can say, "you see what you're doing? you're just making stuff up. What's actually going to happen?"
(That being said, it would definitely have been helpful for me in the past if I had thought to confine questions of consequences to things happening at a point-in-time. I eventually figured out that I needed to ask that for things people were thinking about or remembering, but there was a long time where I also had the hit-them-with-a-stick frustration to this kind of response.)
The only suggestion I have for exercises is to make people write down their own thinking (or state their thinking out loud), and then read it back as a kind of grammar-checking exercise. Are these abstract nouns or concrete nouns? Do they describe a point in time or some sort of vague non-timey thing?
I've done some similar things with small groups, though, and one thing that becomes quickly apparent is that everybody already knows when somebody else is doing it wrong. The part of the exercise that's hard, is learning to apply it to your own thoughts or utterances, and for that, it helps to externalize them first, then treat them as input.
To put it another way, the prerequisite 5-second skill for consequence checking is reflecting on what you just said or thought. If people don't reflect on their utterances, no further debiasing skills can be applied.
I wouldn't even allow that. I much prefer to treat such a sense as a (misguided) signal about the map, rather than a piece of territory that I intrinsically care about. Seeing things with this framing allows you to explore the signals with less distortion, and allows them to go away more easily once you take them into account. If you start treating them as things to worry about, then you get sadness about sadness, fear about fear, and other information cascades that can be quite destructive.
Additionally, on the cases where the irrational discomfort actually sways your decision over the threshold, you're training yourself to listen to things that should not exist in the first place, which just reinforces the problem.
This strikes me as a perfect lead-in to Spock style "Bah, my emotions SHOULDN'T exist, therefor I will just IGNORE them". This does not work well.
If we ignore a REAL negative consequence in our planning, we're going to get frustrated when the consequence happens anyway, because now it's an UNEXPECTED negative consequence of our decision. If we further decide that we're not REALLY having that negative consequence, then it will get further exacerbated by our unwillingness to accept the situation, and therefor our inability to actually do anything to fix the situation. It's entirely possible that we're now miserable for two weeks instead of three days.
Heck, It's entirely possible the whole thing could have been fixed by thinking about it and saying "I would normally feel bad, but since I'm aware of this, I can instead just remind myself of the awesome rational decision I'm making, and how cool my life is because of this Rationality thing!", possibly supplemented by a celebratory slice of cake to reinforce that this is a positive, not a negative, event. (And cake makes everyone happy!)
No no no, not that. That's terrible!
"listen" is ambiguous - oops. You want to acknowledge the feeling, but not act on it. Once you can acknowledge it, you can realize that it doesn't make sense, and then release that feeling and be done with it.
I think it's important to try to convert the reason to a consequentialist reason every time actually; it's just that one isn't done at that point, you have to step back and decide if the reason is enough. Like the murder example one needs to avoid dismissing reasons for being in the wrong format.
"I don't want to tell my boyfriend because he should already know" translates to: in the universe in which I tell my boyfriend he learns to rely on me to tell him these things a little more and his chance of doing this sort of thing without my asking decreases in the future. You then have to ask if this supposed effect is really true and if the negative consequence is strong enough, which depends on things like the chances that he'll eventually figure it out. But converting the reason gets you answering the right questions.
Sunk cost fallacy could be a sign that you don't trust your present judgement compared to when you made the original decision to put the resources in. The right question is to ask why you changed your mind so strongly that the degree isn't worth it even at significantly less additional cost. It is because of new information, new values, new rationality skills or just being in a bad mood right now.
An advantage is that you feel just as clever for coming up with the right questions whatever you decide, which out to make this a bit easy to motivate yourself to implement.
Definitely useful. I personally find the two have a very different emotional/internal "flavor" - I can tell when I want to avoid a sunk cost vs when I'm in a bad mood and just don't want to deal with a cost - but that's not necessarily always true of me, much less anyone else.
This is perhaps ironic because I have been going through precisely this PhD sunk-cost problem for the past few months, but regret bias is a serious part of behavior psychology. I've been dissatisfied with the direction that publication standards are moving in my current field (computer vision) for a while, and as a result have had a tough time finding an adviser/project match that would let me do things at a more abstract mathematical level. No one is very interested in those papers. Ultimately, over a two-year period, I reasoned that it was better for me to leave the PhD program, find a job that allowed me to pursue certain goals, and to leave research ideas to my own spare time. The single most difficult hurdle in reaching this decision was feeling very worried that I would regret leaving my institution (Harvard) because everyone tells me that a PhD from Harvard "opens lots of doors" and lots of people who I trust and think are non-trivially intelligent have insisted that unpleasantly sticking it out in the PhD program just to obtain the credential is absolutely the best thing.
My own assessment is that I will do just fine without that particular credential and that being able to use personal time to pursue the research I care about, even if I ultimately am not talented enough to publish any of it on my own, will be more fulfilling. But this was a damn hard conclusion to come by. I felt stressed and nervous, concerned that I will hate my future job's working conditions and beat myself up over not sticking it out at Harvard. I largely made it into Harvard through sheer, stupid ability to work unreasonably long hours to self-teach. That is, by stubbornly never quitting; it's not easy, however rational I wish to be, to feel free of these kinds of self-identity stigmas (e.g. don't be a quitter).
I guess what I'm trying to say is that perceived future pain of regretting a decision is a legitimate consequence to consider. And sometimes that is absolutely a consequence that one should wish to avoid. To offer another example from my own life, a family member was in a position where she became pregnant unexpectedly while she was an unmarried 19-year-old college student. After many talks about the situation in general, I was asked what my own opinion was about the option of getting an abortion. I said it seemed like a reasonable option and might ultimately be the best thing, obviously modulo the person's personal beliefs. Ultimately, however, this family member chose not to get the abortion because of the counterfactual regret of having terminated a potential life.
The person said, "if I get an abortion, then in the future I will remember that I did that thing and (as far as I can tell right now) I will always feel visceral pain about that." That is a legitimate future consequence.
I think the problem that you want to isolate is different than just regret bias. I think the problem you want to address is the fact that a person's current self is usually slavishly in the service of the remembering self as Kahneman puts it. We buy things because we think they will provide lots of utility, but then a few months later we don't even use them any more. We prefer to keep our hands in painfully cold water for a longer total time as long as the last bit of the time includes warmer water (and thus fosters a more pleasant transitional memory). And we think we will regret something a lot more than we really will.
You want to design exercises that bring about a stark comparison between how you think you will remember something vs. the actual facts of the matter. And then focus on situations where the first component (how you think you will remember something) actually should matter (perhaps an abortion is a good example of that), and how the cognitive machinery that applies to problems like that one is completely inappropriate for problems like "should I upgrade to the new iPad2 because of the shinier screen", yet we use the same cognitive software for both problems.
This sort of thing has been looked at w.r.t. dietary decisions before. I believe the results showed that when you are under a cognitive load, you'll make consistently poorer snack choices than when you are not being asked to answer hard questions. Imagine how much more this would be influenced by the stresses of a situation like anguishing about whether to leave a PhD program.
I'm not optimistic that there is an easy way to address this. It seems to fit in with Hanson's near/far mode ideas as well. When in near mode, we'll be more capable of isolating practical constraints and consequences of a decision. But if a question immediately puts us in far mode, it's much harder.
Consider the difference between "Should I leave my PhD program?" and "Should Jeff leave Jeff's PhD program?" As much as people fail to pick out the consequences in their own decisions, I would suspect it's far worse when migrated to another person's issue. We tend to give advice in far mode, but always expect to receive advice in near mode.
There's just a lot to disentangle about this. My opinion is that it would be better to break up the problem of "Why don't people make sound consequentialist decisions?" into a bunch of smaller domain-specific sub-problems, and then to build the small tests around recognizing those sub-issues. Once people are good at dealing with any given sub-issue conditioned on the event that they recognize that they are in that sub-issue, then move on to exercises that teach you how to recognize potential sub-issues.
It occurs to me that games with some significant strategic component might be useful for priming the "but what consequences does it have?" response. I'm thinking of games like Magic: the Gathering, Settlers of Catan, Risk, etc. (I'm sure the board game aficionados will have better examples than I). I say this because of personal experience with Magic players - as they get better at magic, they tend to get better at life. Well, some of them do. The others perhaps compartmentalize too much, so maybe this won't help with everyone.
In any case, my model for what would work is a relatively easy social game that allows a non-trivial number of actions with unclear consequences... unless you stop to think about them. Magic would be perfect... if it wasn't so complicated and if fantasy tropes didn't turn off a large segment of the population. Ideally the game would be something you create instead of something your subjects/clients may have played before.
I have no ideas for the actual game, but maybe this sparks someone else's imagination.
Really? I sure haven't noticed this. If anything from my own circle of acquaintances it looks like those who got better at life were the ones who stopped putting so much of their time and attention into card games.
Roughly, there's two populations - those who apply what they learned in magic (microeconomics, essentially) to life and those that don't. The latter tend to spend way to much time on card games. The former start saying things like "this event is pretty low EV for me, i think I better study/write that paper/work on that project/etc. instead."
In any case, as people get better at Magic, they get better at thinking about the consequences of their actions within the game. This seems like a natural stepping stone to thinking about consequences in all situations, though the trick is getting people to generalize it.
If I was sensible, I probably should stop playing Magic, or at least paying money for cards... but I have too much of my self-esteem wrapped up in that stupid game. It's like trying to quit smoking. :P
This is why I tend to have an immediate aversion to using Magic as a rationality teacher. The whole game is set up on a business model that incentivizes constantly shelling out money for new cards to keep your deck from becoming obsolete. Wizards Of The Coast's goal is to make sure that their players cannot continue to be competitive without providing a constant revenue flow. If you want to teach people good rationality skills, don't start by encouraging them to get into something like that.
I've always been turned off my MtG on the grounds that I should just be able to print up any cards I like and use them as long as they form a valid deck, rather than having to follow WotC's anti-"counterfeiting" policy. Do any Magic players actually do this?
I have friends who did so, but they only used them to compose special print decks to play with the few other friends who were also using print decks, and I think they used their "real" decks more even among each other than the print decks.
People create "proxy" decks all the time. It's one of the dominant ways of testing for big tournaments (when you don't know what cards you'll need until you settle on a decklist, but you don't want to buy every potential card). However, for some reason the casual community doesn't seem to do this as much. This is somewhat ironic because sanctioned tournaments are the only place you have to use real cards.
I don't know if it's a consequentialism issue, but "if I was sensible" seems like a way of locking a problem in place.
Maybe there should be a separate category for noticing identity issues.
How would one distinguish between the scenario in which they begin to apply Magic-like thinking to their regular life and begin optimizing there, and the scenario in which ordinary diminishing marginal returns to playing Magic causes them to switch to the other activities?
If they're actually optimizing, you should be able to see the results, though measuring them is another problem in itself.
If I'm thinking of games to reinforce consequentialism, my first thought is to use games with actual story involved; you don't lose points, or regions, or so on, you lose the lives of characters you're attached to, or their trust, or maybe you fail to prevent a genocide, etc. Things which people will be more likely to associate "this is a bad game outcome" with "this would have been a bad choice in real life."
The first solution that comes to mind for this is a video game, perhaps some kind of visual novel that features a large number of choices and forces the players to choose consequentially on pain of causing Bad Things to happen in the game. But I don't think this is actually a very good solution considering how much effort it takes to make a visual novel, which can be played in its entirety and will no longer offer a single new choice afterwards, and how many people are simply not interested in playing visual novels.
Maybe some sort of roleplay would be more feasible, at least you wouldn't be designing a whole video game for each scenario, but it still sounds like an awful lot of work.
I have two interpretations of your idea, so I'll just say what I think of both.
1) Underlying, known, game mechanics with a story behind them involving role playing.
I like this because it gives the players something they can easily point to and say "look, consequences!" in the game mechanics while making the situation feel closer to reality. However, reality doesn't give you the mechanics by which it works, so this may not translate into real-life decision making as well. On the upside, this is easy to make into a social game - think DnD but with less magic and dice.
2) No game mechanics, just a "choose your own adventure" game.
The consequences are more nebulous in this version, which is both a positive and a negative. It's a positive because it forces more brainstorming of actual consequences, but it's a negative because that makes it harder to initially start thinking about the consequences of actions. It's also difficult to make this type of game vary from playthrough to playthrough.
Starting with a type 1) game and then moving to a type 2) game seems like it might take advantage of both types' strengths. Alternatively, there's really a continuum between the two types, so maybe somewhere closer to the middle is best.
I sometimes try to get myself to make better decisions by pretending I'm a character in a Choose Your Own Adventure book. (E.g. "If you decide to stay on the couch because you're too lazy to work, turn to page 30.") Unfortunately, in the real books it's rare that enough information is given for you to make a really good decision, and the authors also appear to like messing with you by having good decisions blow up in your face.
So, maybe a similar book that actually gave you enough information to make a good decision and rewarded good decisions and punished bad ones?
This sounds like a more useful, more intuitive, much more widely applicable reification of my own method of "What Would Your TV Tropes Page Say?"
The problem with TV Tropes is that they've been heavily primed with fictional evidence.
If you are influenced by the fictional evidence, your TV Tropes page will say Wrong Genre Savvy.
I don't know how many people have this issue, but I can't read Choose Your Own Adventure books without marking several past pages so I can rewind time, or try multiple branches, or safely find out what was hidden behind the venomous Venusian potted plant. Really, the only bound on it is that I eventually run out of fingers to mark my place, which constrains my time travel abilities to about four save-states. (In visual novels it's even worse, since there are enough actual save states that saving at anything that looks like a potentially significant branching point becomes viable. I've actually started using walkthroughs from GameFAQs to find out where I don't need to save, so I can stop fretting about making an irreversible decision. Trivial time travel is surprisingly addictive! What would the world be like if everyone could do it, I wonder?)
I really, really wish that this were a useful approach to life, but if it's possible to save and restore universe states, I have not been made aware of this. And obviously I haven't noticed anybody else doing it.
IMHO that's a really important point. You get a better grasp about consequences of your choice after trying several options and seeing how the consequences of different actions differ.
The best laboratory example of this is playing go on a computer. Typical go software records your games, and then lets you replay, play different variants, analyze when things went really bad after a silly move, etc. After a while you get a tree of diverging game records. In some you won, in others you lost. It's a good learning experience.
(disclaimer: I'm not sure how to un-compartmentalize this learning to be applicable in real life, not just in a game of go)
I do the same thing. I found that I needed far fewer save states when I routinely took the BAD choice first, since they usually lead to the shortest further decision tree. I'd also occasionally use physical bookmarks, for the few rare books that just would NOT kill you off until the very end (even though you were quite possibly stuck on a guaranteed-negative branch of the decision tree)
As to applying it to real life, I will sometimes think about the decision tree involved. Playing Chess is a good example of this: If I make THIS move, my opponent could do X, Y, or Z. If she goes with X, I can do X-a, X-b, or X-c... and then weighing all this based on probability ("She hates doing X!" "Y is her best move!") and expected value (if she does X, I'll lose. If she does Y, I go up a pawn.) Fortunate for me that she hates doing X :)
reminds me of http://www.epicsplosion.com/epicsploitation/adventures , maybe you'll be able to find something there?
Lost Cities might work, if you took your time and tried to make the optimal play every move. I think you can make it work with playing cards.
Careful, there: some vindictiveness ("if you attack me in Africa despite our pact, I will go totally apeshit on you for the rest of the game") is an essential part of playing e.g. Risk well (in our group) - naive consequentialism ("looks like I lost Africa, taking Australia from (unrelated player) seems best now") does not work very well on intelligent and adversarial agents.
Of course, most of the world is not an intelligent and adversarial agent - pre-committing to going totally apeshit on an unthinking animal is just stupid. The easiest and biggest wins for consequentialism are there, not in games of Risk.
(Non-naive consequentialism works fine. Naive consequentialism probably works fine in many games, e.g. two-player games like Magic.)
Totally agree. I'm ruthlessly vindictive but perfectly trustworthy (meaning I refrain from making promises I do not keep) when it comes to strategic situations like that. It looks superficially like being completely unsophisticated but it works.
"Intense, long, certain, speedy, fruitful, pure—
Such marks in pleasures and in pains endure.
— Jeremy Bentham's mnemonic for the signs of "consequentialist reasons"
EDIT: It occurs to me that I should explain this more. Bentham was trying to popularize consequentialism and remind his readers of what sorts of things count as consequentialist reasons to prefer a particular outcome. Eliezer suggests that we should ask about the closeness in time and the duration of a consequence. Bentham mentions these ("speedy" and "long") but also includes some others:
I haven't had a chance to test it much, but I think I have an idea. Frame the question as what will happen if I do x that won't happen if I do y. At this point it seems like it should be possible to reuse the mental processes for making beliefs pay rent. Basically typecast "I do x" and "I do y" to beliefs. Then see what experiences I anticipate as a result of the beliefs "I do x" and "I do y". Then determine which set of anticipated experiences has higher utility.
This could also be caused be confusing correlation with causation.
What, we're not even allowed to have identities now?
We are the Borg. Lower your shields and surrender your ships.
To become a true rationalist one must shed the trappings of personhood. The rationalist's mind has no goal except rationality itself; no thought except the Bayesian update.. Only once you are free of worldly concerns and the concept of autonomy may you see the light of Bayes.
edit: Sorry, I was joking. I thought I was being ridiculous enough for it to be obvious.
Incorrect indeed.
I thought it had the goal of maximizing expected utility.
Um, no.
Sorry, I was joking. I thought I was being ridiculous enough for it to be obvious.
I should have remembered that you've been around for a while, but bear in mind that the joke is just the sort of Straw Vulcan reasoning that some new people think Less Wrong obviously must subscribe to.
'Twas completely obvious to me. I mean seriously, "light of Bayes".
laughs The username was a pretty obvious give-away, IMO :)
Short answer? No, you shouldn't.
Depends what the consequences of asserting one to yourself are.
Identity shouldn't act as a normative consideration. "He's going to do X because he belongs to a reference class Y" may be a valid outside view observation, a way of predicting behavior based on identity. On the other hand, "I'm going to do X because I belong to a reference class Y" is an antipattern, it's a descriptive explanation, fatalist decision rule, one that may be used to predict, but not to decide. An exception is where you might want to preserve your descriptive identity, but then the reason you do that is not identity-based.
So you can have an identity, the way you can have a pair of gloves or a Quirrell, just don't consider it part of morality.
Identity shouldn't act as a normative consideration for an angel, maybe. For a human, "identity" is a pragmatic reification of cached complexes of moral conclusions that aren't immediately accessible for individual analysis. "Normative" is a misleading word here.
Still shouldn't for a human, even if does. It's a normative consideration, not a descriptive one.
...Is there a word for "normative given bounded rationality"?
Bounded rationality is like the mass of the Sun, difficulty of the problem, not a kind of goal.
I don't understand.
If you're trying to dam a river, and you only have 100,000 bricks, then there is a normative solution, i.e., the solution that has the greatest chance of successfully damming the river. Talking about solutions that require one million bricks is talking about a different problem that is only relevant to people with millions of bricks. So when you say, "identity shouldn't act as a normative consideration", that sounds to me like, "you should already have one million bricks, there is no normative solution if you only have 100,000 bricks". Using 100,000 bricks to dam a river isn't using an approximation of the solution you would use if you had a million bricks. That's why I say "normative" is a misleading word here. It implies that you should try to approximate the million-brick solution even when you know you don't have enough bricks to do that: a tenth of a great million-brick dam is one millionth as useful as a complete 100,000-brick dam. Why not just renormalize such that your constraints are part of your environment and thus part of the problem, and find a normative solution given your constraints? Otherwise the normative solution is always to have already solved the problem. "What would Jesus do? Jesus would have had the foresight not to get into this situation in the first place." "Normative" is always relative to some set of constraints, so I don't see why normative-given-boundedness isn't a useful concept. I'm reminded of Nick Tarleton's intuition that decision theory needs to at some point start taking boundedness into account.
It's useful to take the limitations of decision-making setup into account, but that is not fundamentally different from taking the number of bricks into account. The idealized criteria for comparing the desirability of alternatives don't normally depend on which alternatives are available. People shouldn't die even if it's impossible to keep them from dying.
Prescriptive.
I'm not sure this is responsive to Will's point... at least, it seems plausible that the moral considerations he considers identity to imperfectly encapsulate are also normative, which is why he refers to them as moral in the first place. That is, I think he means to challenge the idea that identity shouldn't be/isn't a normative consideration.
I agree but.... purposely self-identifying with a reference class that has supposed-skills that you are trying to acquire does seem to have benefits in actually becoming more likely to have those skills. eg "I'm a hard-working person and hard-working people wouldn't just give up" is a way of convincing (/tricking) yourself into actually being a hard-working person.
EDIT: that being said - it certainly wouldn't be consequentialist. :)
I previously wrote a comment that seems relevant here:
An agent that lets identity influence its decisions probably deviates from ideal rationality, but how to fix that? If we just excise the identity-based parts of its decision procedure without any compensation, that could easily make it worse off if for example it's CEV depends on its identity.
We did the Really Getting Bayes game at the Mountain View meetup this week. My impression of it was that the explanation was at first a little unclear, but that once we had gotten the sense of it, it was worthwhile.
One thing that I realized during the game was the degree to which I was using the availability heuristic to provide my likelihood ratios. For instance, one object I was given was either an electrical extension cord or an audio cable. In coming up with RGB likelihoods, I thought, "Electrical extension cords are usually black or white, hence not RGB; whereas audio cables are often RGB" — by using instances I could bring readily to mind to come up with this thought. My partner then told me the item was RGB, and I correctly predicted it as an audio cable. However, the electrical cord in his other hand was in fact orange, not black or white. While my availability heuristic worked for the purpose at hand, it clearly would have been wrong for other purposes.
In case this wasn't done, a physical demonstration of a game like this at first is important, with a concurrent verbal description to tag it for indexing: "Step 1: we do this", "Step 2: we do this." Showing beats telling alone. Verbal or written instructions are a low bandwidth form of communication that are better used for tagging/clarification of demonstration data (i.e. naming the steps while you do them) or error-correcting after a demonstration (i.e. people can look stuff up if they get confused).
I remember reading about an experiment performed by behavioral economists where person A divides some cash and person B either accepts their division and gets their allocated share of the money or rejects it and neither party gets their allocated share. You could say the consequentialist solution is to always accept the division of money, which most folks don't do, so this could make a good trial exercise. On the other hand, if person A is someone person B is going to have repeated interactions with, one could argue that the social capital of training person A to divide things fairly might be worth forgoing cash... So maybe it wouldn't work in a scenario where the class meets again and again? (Unless things were anonymized somehow...)
There is also the Newcomb's Problem aspect to this, where having taught the class about consequentialism will make it appear as though you have made everyone who is Person B worse off.
Reading up on experiments behavioral economists have done in general seems like it could be a good source of ideas.
I can't help but think that the best way to actually get people to be consequentialist is similar to the way to actually get people to be atheists: convince them that all the cool kids are consequentialist. This probably contributed to me becoming more consequentialist, in the form of reading about behavioral economics studies where people did silly and irrational things and wanting to not be one of the silly and irrational ones.
I would strongly recommend against going this direction. Consequentialism is about methodology, not particular results. As soon as you say "the consequentialist always accepts" the clever students will get a funny look on their face, as they try to cost out and compare the immediate gain and long-term loss.
Consider Kohlberg's stages of moral development, which doesn't care about the conclusion drawn but does care about the stated justification for the conclusion.
http://en.wikipedia.org/wiki/Ultimatum_game
Right, this article appears on the surface to endorse causal decision theory, which we know Eliezer doesn't in fact endorse. Mostly that's fine, but there are occasions where CDT will make the wrong call, such as the examples you point out.
I predict that if a stranger tried a one-shot Ultimatum game against Eliezer with a 99-1 split in the stranger's favor, EY would refuse it on TDT grounds. Thus any person who knows Eliezer subscribes to TDT wouldn't offer a manifestly unfair split to him.
You could structure the game so that the person making the offer and the person receiving it were paired randomly after the offer was specified.
By the way, if you don't mind my asking, once you've come up with your rationality curriculum what do you plan to do with it? Are you making inroads with whoever you would need to talk to to get this in a school curriculum, for instance?
I think they plan on running workshops or seminars, likely targeted at startup founders or business/consultant-type people handling large decisions (both from a capability-to-pay and a convinced-of-the-value point of view, this makes far more sense than school curriculums).
Initially. School curriculum would be harder to develop, so the plan is for that to happen later.
What is the closest existing thing to this? How can we make friends with someone who is good at it? Are there any books about them doing it that we could read?
Brainstorm:
* Sales seminars
* Continuing education programs for businesspeople, e.g. evening MBAs
* Getting things done seminars, for organizations and the public
* Lean thinking, kaizen seminars
* Six Sigma seminars
* http://www.richdadeducation.com/ (kind of scammy)
* The Landmark Forum (even scammier)
* http://danariely.com/speaking/
* http://personalmba.com/ (note that ways to pay Josh other than buying his book are not immediately apparent, and that he is a Less Wrong user).
* http://thinkingthingsdone.com/ (also a less wrong user)
* Tony Robbins, firewalking, etc.
I would guess that large organizations are more willing to pay for live instruction than startup founders are. On the other hand, you wouldn't be able to suggest that they do anything that wasn't in the best interest of their employer like quit their job.
If organizational seminars are going to be a goal, it might not be a bad idea to start talking to relevant organizational folks to make sure you're making a product they actually want to buy. Jane Street could be an ideal first client, since they've got prestige you can use to sell other clients, EY has a pre-existing relationship with them, and they seem genuinely interested in improving their rationality. On the other hand, their rationality may be at a level where they don't think they could benefit from this sort of workshop, or targeting the workshop at them would mean developing a different set of materials. (But these "advanced" materials might appeal to clients who had already purchased and enjoyed the "basic" materials.)
The standard way to do this sort of B2B sale is to graduate to more and more important clients, since a lot of businesses will not buy a novel product unless some other businesses bought it and were happy with it. That's why getting and pleasing the first few clients is so important.
Robin Hanson has had a hard time selling prediction markets to businesses. Should we expect this to be more successful? I would guess yes, since it's not as explicitly targeted at replacing the people who might choose to implement it.
Another inferential path: it may be valuable to differentiate them as attitudes and events. If my motivation for getting a PhD is "I will feel terrible if I do not get a PhD", that's an attitude motivation, which in theory I have internal control over. If my motivation for getting a PhD is that I will get hired for a different sort of job, that's an event motivation, for which control is primarily external. I don't have control over whether or not a variety of job requires a PhD, but I do have control over whether or not my attitude will be negative if I don't have a PhD.
The obvious social mirror of the Really Getting Bayes game is to have people pair up and dissect motivations for something, trying to identify what's event-based and what's attitude-based (or however you're presenting it). It will help for them to be motivations they're feeling, rather than descriptions they're reading.
Asking them to provide examples from their life is potentially useful and will promote bonding between participants, but requires heavy investment by participants. Another approach is to offer them a choice between gambles, where all potential consequences are as apples and oranges as you can make them. Some more trickery could be used to provide non-consequential reasons to pick one gamble over another- perhaps give participant's names to the gambits, but don't require them to pick the gambit with their name.
Obligatory link: Urges vs. Goals: The analogy to anticipation and belief.
Well, this might be a completely trivial suggestion but, if the point is to get people using consequentialist thinking in their lives, why not have them each pick some big important decision in their lives (either an upcoming one or an ongoing one), preferably one they aren't entirely set in already or are uneasy with, so they would be more open to changing their mind, then get into groups and each take turns to discuss their decisions and options (hopefully without applying any judgement to them at this stage), then the other members trying to come up with as many new options as possible. Then they discuss the outcomes and benefits of each option and try to find the best one, not moving on until the entire group feels that the last person's problem was solved adequately. That way not only do they practice consequentialism in a practical way, they might get immediate help with a current problem which could end up being useful on its own and make consequentialist thinking seem more useful.
I think of this as "looking effectward", one of the basic directions in concept space (opposite causeward, making it the inverse operation of asking "why").
Here's a long-form exercise:
The other participant(s) ask questions to clarify the problem/decision.
The problem/decision can be either real or imaginary, but if imaginary the presenter must come up with appropriately detailed answers to questions.
This will probably take a fair amount of time.
The shorter-form version starts with steps 1-3 already done as example problems.
It probably helps if step 6 is a surprise.
Incidentally, I prefer "Compare Likely Consequences" to "Check Consequentialism" as a label for the skill.
The day-to-day cognitive skills I've mastered most completely (I will not say "rationalist skills," because this is true of my countless irrational skills too) are the ones which I learned during a moment of strong emotion — any emotion, excitement or curiosity or joy or surprise or depression or fear or betrayal.
In the case of this particular skill, it was betrayal. I brought it on myself — the details aren't important; suffice it that I spent two weeks living in the "should-universe" (I like this term) before a rude reminder of reality — but the emotion, the physical neurendocrine experience of betrayal, was quite real. And I've been able to return to it ever since, and if I'm ever in a situation where I might be working from a cached plan, I can relive a hint of it and ask myself, "Now, you don't want to feel that again, do you?"
Unfortunately, this experience strongly ties the five-second skill of "check consequentialism" to the emotion of betrayal in my mind. It is very easy for me to construct social experiments in which the teacher radically betrays her students, and then turns around and says, "Don't let anyone do that to you again!" But that is horrible teaching. It's a lot more difficult for me to imagine what "check consequentialism" would feel like if it carried a strictly positive emotional association, and then extrapolate outward to what kind of social situation would provide that emotional/cognitive link.
Students must abandon a cached plan, and evaluate the real-world consequences of their actions instead, at precisely the moment they get a strong positive emotional charge. Preferably "fun." Preferably in the sense of a party game, not a strategy game: both because people who have learned to win without disrupting social bonds (or who care more about winning than about socialization) have often already learned this skill, and because the moment I construct "winning" as a state which disrupts social bonds, I've set up a false dilemma which misleads my students about what rational thought actually is.
But what's the chain of causation? A dispassionate experimenter times the payoff to correlate with the decision? That seems awfully Pavlovian. Leaving the plan causes a reward which provides an emotional payoff? Maybe, but if a student only leaves the plan in expectation of reward, they haven't actually learned anything beyond the latest professorial password. The excitement of getting the right answer to a puzzle inspires leaving the plan? I suspect this is the way to go. But then what sort of puzzle?
I'm going to press the "Comment" button now, even though I don't think I've contributed much beyond a restatement of your original dilemma. Perhaps having done so, I'll think of some specific scenarios overnight.
Practicing this could be fun in pairs, dissecting an acted out scenario. Two instructors act out previously conceived scenarios, with a Influencer and a Reactor. At some point, 'twill be implied the Reactor wishes to act on the scenario itself or the knowledge presented therein; the scenario will then halt, and the students put in pairs to brainstorm the beneficience and maleficience of possible actions. Each student will take turns (which can be timed) being the brainstormer and the consequentialist (utilitarian?); of course the pairs can have different functions, like as suggested. These just serve to outline the general idea.
For example:
INFLUENCER: Good day, sir! On your way to the place we are going?
REACTOR: Why yes, I am! However odd you too shall be going there; I wish we shall fall upon their fancy!
INFLUENCER: Oh dear! The Gods are weeping once more!
REACTOR: Dear me! I prefer not to be wet, and so I always carry an umbrella upon my person!
INFLUENCER: Indeed, I see it now grasped in your hand! Whatever shall you do? Poor me, if only I were so prepared....
Halt
BRAINSTORMER: He opens his umbrella, and uses it himself.
CONSEQUENTIALIST: The umbrella protects him, but not his companion to any significant degree. (The companion must dodge the edges so as not to be poked in the eye, and may be offended.) The umbrella may wetten things once inside(, and earn him the ire of some people by whom he'd rather not be thought ill.)
Both would then consider the merits and, as outlined by the parentheticals, disadvantages of these outcomes, moving on to the brainstormer's next suggestion afterward.
After each has taken a turn, the instructors would go around the room asking each pair their brainstormed actions, their potential consequences, and the positive and negative aspects of each; as Vladimir suggests, these aspects can ("should") be relativized against each other - if they do relativize, they would state the pair's preferred action and its predicted consequence. The instructors could reinforce correct applications, and constructively criticize incorrect applications, with care taken to not put any pairs down too much (using softeners, etc.: "That's quite creative! We're glad you thought of that, this is an excellent example of how even the best consequentialists can go wrong...").
Kahneman suggests such an exercise for groups after pointing out that organizations generally act more rationally than individuals. The devil's advocate role and thinking at the worst possible outcome. We don't always have the luxury of having others near us for checking our thoughts. But we often have imaginary conversations with friends or parents. So it shouldn't be very difficult to assign a devil's advocate position to a imaginary voice. That should put in perspective the way we feel about the subject. It is a basic mean of delaying the strong coherence of the first good narrative.
Maybe it would be great to have an imaginary Bayesian friend...
This seems like a comparatively reliable procedure: imagine a collection of possible worlds generated by possible actions; explain what future events distinguish between these worlds in a way that makes one of them preferable to the others; then choose the action that leads there.
Paying attention to events that distinguish these possible futures from each other guards against errors such as comparing to status quo or should-world (neither of which is among the possible futures), or worse comparing an arbitrarily picked-out event (in one of the possible futures) to an arbitrary anchor (i.e. a cost that feels excessive in some absolute way, and not by comparison to alternatives). Focusing on future events and not past events or actions themselves guards against deontological and identity-based arguments, such as "this is a proper action", "he shouted first" or "because I'm a human being".
Saying "positive consequence" sounds like a bad habit to me: positive compared to what? The comparison should be among the alternatives, without anchoring on some baseline of neutrality such that some consequences are more "positive" than that.
So... how would you design an exercise to teach Checking Consequentialism?
I would check to see if such a thing already exists or if there are people who have experience designing such things. I know of a Belgian non-profit 'Center for Informative Games' that not only rents games designed to teach certain skills but will also help you create your own.
From their site: On request C.I.S. develops games for others. The applicant provides the content of the game, while C.I.S. develops the conceptual and game technical part to perfection. The applicant has the opportunity to attend some game tests and to redirect when necessary.
They also offer coaching if you want to work on your own: Do you want to create a game concept on your own, but you don't know where to start? No worries C.I.S. can give you a hand. During a number of concrete working sessions C.I.S. facilitates your development. In between sessions the applicant continues the work to, finally, end up with a solid game.
I have enjoyed their games in the past and can attest to their quality. The obvious problem is that it's a purely Belgian organization and the 'search' function only works with Dutch words. However if you want to check them out I'd be happy to act as a go-between. Since a couple of months there is even a Brussels LW meetup, I'm certain I could get a couple of members to help in the production process (again, if this seems interesting)
Maybe the easiest way to teach it is to teach how it applies to others. That is, train people to catch nonconsequential reasoning in arguments that others make, and then hope that they apply that to themselves. The easiest way to do that is by reflexively asking, "so what?"
Nice. Many people are much better at criticising others than finding the same flaws in themselves.
So... how would I design an exercise to teach Checking Consequentialism?
Divide the group into pairs. One is the decider, the other is the environment. Let them play some game repeatedly, prisoners dilemma might be appropriate, but maybe it should be a little bit more complex. The algorithm of the environment is predetermined by the teacher and known to both of the players.
The decider tries to maximize utilitiy over the repeated rounds, the environment tries to minimise the winnigs of the decider, by using social interaction between the evaluated game rounds, e.g. by trying to invoke all the fancy fallacies you outlined in the post or convincing the decider that the environment algorithm actually results in a different decision. By incorporating randomness into the environment algorithm, this might even be used to train consequentialism under uncertainty.
You could use mazes where your score is -(total distance traveled) . First, give a simple maze with two obvious paths, A and B but Path A is much shorter. Then give a second maze identical to the first but you are taking over from another player who has already gone down Path B but the shortest way to the exit is to double back and then go down Path A. Then give the same maze but now there is an obstacle on Path A that you must go around if you take this path and so it's now optimal to go on Path B. The obstacle was placed there for some unfair reason (Perhaps only women will face the obstacle.) Next, the same situation as before but you have the ability to erase the obstacle at no cost. Finally the same as before but now you are told that the maze is only a map and you can't erase the obstacle on the map. You should have the people play for real money to make the game more emotionally "meaningful." Also, you could plant someone who makes arguments against the consequentialist outcome using analogies to real life situations (i.e. women should never give in to discrimination.)
And then you collapse in to short vs long-term gains. Giving in to discrimination is a net long-term loss, since then you'll face it in the future...
Also, for initial teaching, you want to present the SIMPLEST possible version. Rationality 102 is when you start introducing enemy agents, and then eventually combine that idea with this one after BOTH "sabotaging agents" and "consequentialism" are understood.
An important question to ask that you are leaving out is "What are my alternatives to this course of action?"
The comparison of consequences requires an alternative set of consequences to compare to. Considering the question "Should I be in graduate school?" The answer may well be different if the alternative is getting a decent job than if the alternative is starving unemployment.
The listing of alternatives also helps catch cheating. If the alternative is implausible and disastrous (Stay in grad school or be a hobo) then it is likely that Checking Consequentialism isn't being done seriously.. The alternative compared needs to be a serious answer to the question "What would I do if I couldn't/wouldn't do this?"
I think an "Improv to Rationality" class would be very fun and interesting. I don't know how WELL the games would work to teach the desired skills, but I definitely think it would make the lessons more memorable, AND you'd be able to add the phrase "Team-Building" to your class description. You can sell anything if you promote it as "Team-Building." :P
A game that would teach how to look for alternative options (see parent) could be "New Choice". In this game, a scene is played out, with the moderator occasionally demanding a new choice(s) from the players. (Video Example (includes 15 sec commercial)).
Not only would it teach about not becoming too set on a current course of action, but could have the lesson of: "When thinking up alternative actions, don't just stop at one or two. Instead imagine a moderator yelling "New Choice!" at you. Keep thinking up new alternatives until you get to crazy-land (i.e. "Here's a cat from my pants")
Another sub-skill mentioned, was the ability to recognize bad rationalizations (i.e. "I will feel bad if I quit something I put so much effort into."). Perhaps one way to learn to recognize these, is to do bad rationalization ON PURPOSE, to see what it feels like. A good game for that is "Challenge in a Minute".
In this game, a two-sided silly debate is chosen, such as pirates v. ninjas, or Coke v. Pepsi. All the players line up and challenge each other's arguments. Arguments are supposed to start somewhat seriously ("Ninjas are sneakier") and devolve over the course of the game ("I don't like his pants"). (It's easier to just watch a video, than explain the rules.)
Participants could then brainstorm what it felt like to come up with the bad rationalizations. I would expect answers like: Grasping at straws, Searching your brain for things to support your position, Being proud of clever retorts, etc. Participants could then ACTUALLY try to answer the debate question (ninja v. pirates, or whatever) in teams, and then discuss what it felt like to actually try to find the answer. I would expect answers like: Defining the problem, Not knowing the answer, Looking for sources, Being willing to change my mind (debate question must be sufficiently silly that most people would be willing to change their mind.)
Note: I'll admit that I worked backwards for this. Instead of thinking "What's the best way to teach consequentialism?", I thought "It would be awesome to do an Improv Rationality class. How can I relate some improv games to the lessons we're trying to teach?" So this solution probably isn't optimal.,..But it IS fun!
Short exercise. Does anyone actually think pirates stand a chance against professionally trained assassins? I thought the only reason people defend pirates is because it's a way to say both of them (or their identity-memes) are just so damn cool.
In equal numbers? At sea? Are the ninjas manning a ninja-vessel?
Why would the ninjas try to fight pirates at sea? Are they intellectually disabled ninjas? Clearly they would wait until the pirates land, get ridiculously drunk and start carousing (and raping and pillaging if they are into that sort of thing). Then poison them or stab them in the back.
Maybe they're just sailing on their way to a ninja convention, when suddenly - bam, pirates!
(Laughing.)
And the pirates board the ship trying to take the loot when suddenly the ninjas casually slaughter all the pirates, board their ship (the pirates probably crippled the ninja ship before boarding) and sail the pirate ship away.
Mind you I'm still giving an idiot ball to all the ninjas for sailing in a ship that isn't heavily armed.
I think the group with practice winning naval engagements would more likely win the naval engagement.
And by this you mean the Navy vessel that is transporting the ninjas would more likely win the naval engagement? (The engagement that the pirates would never initiate because they are experienced in not picking fights that they will lose).
Or are we still talking about the landlubber ninjas who have never learned to sail that are for some reason trying to sail themselves across a sea for conference purposes? I'm imagining them up in the rigging comically trying to sail and needing to use all their ninja agility to dodge the yardarms that are swinging around wildly as they try to figure out which way they are supposed to pull the ropes...
Basically, pirates can beat ninjas if they load up their cannons with idiot balls for ammunition and the ninjas eagerly run out to catch them.
International ninja conferences would be fun, though. (Once the transportation difficulties were ironed out.)
Most people are landlubbers. Most landlubbers going on a voyage would be on a passenger ship, not a navy vessel. I see no reason either of these wouldn't apply to ninja. In this particular situation - one or several ninja on a passenger vessel raided by pirates - I expect the pirates to win. Of the circumstances where ninja and pirates would be in the same place at the same time, this seems to be the most likely where the pirates would have the edge.
Well, you see what you've done here is you've put me in a situation where I'm in an argument I don't care about, I agree with you, and yet some part of my brain is still playing Pirate's Advocate. Thanks very much.
Sacrilege! Next you will be telling me you don't care about astronauts vs cavemen either! (If they are fighting in space the cavemen lose. Just sayin'.)
A self-taught dirty fighter/swashbuckler against a professional assassin with quality weapons and poinsions isn't much of a fight.
Exactly!
Upvoted for delicious ironic ambiguity.
I believe Dr McNinja has something to say about this.
Are we talking about real pirates and ninjas, or fantasy pirates and ninjas?
Amongst other advantages, fictional pirates have the devil's own luck.
OTOH, fictional ninjas have more Rule of Cool mojo than fictional pirates.
Fictional ninjas are handicapped by the fact that they attack one at a time. Fictional pirates, meanwhile, in addition to not suffering from scurvy like you'd expect, have improbably awesome fencing powers.
Well, how else can they dispose of their stolen loot?
Or keep their dogs in their yards?
Well, there you have it, then. They can hide behind their magical fences.
Real pirates and ninjas. Fantasy pirates and ninjas are of course equally awesome, with the pirate persona perhaps having more potential for injecting individual charisma. Because a ninja who is complaining that the rum is gone just seems incompetent.
It is my position that any debate about ninjas vs pirates that is maintained must be about declaring the coolness of fictional pirate and ninjas. (Which is a perfectly respectable passtime!) Anyone who actually thinks real pirates aren't just ridiculously worse than ninjas is not thinking very well at all.
If we are being "real" then it only makes sense that the historical ninja would be fighting wokou, which are Japanese pirates of the period.
Yes, ninjas can beat Japanese pirates of that period - if for some reason the ninja's masters decide that having a bunch of pirates that mostly attack neighboring countries is a bad thing. And, if you allot enough time to mount a campaign, give them a map and an even more highly unlikely set of orders they could beat any other pirates of their time period too.
Why would real ninjas fight real pirates ? Isn't that the job for the Coast Guard, or the ancient Japanese equivalent thereof ?
Goatee envy. Pirates have much better goatees.
And from what I understand the Japanese pirates were mostly a problem for China, not Japan.
What are we talking about here? Ninjas didn't normally kill in open combat (nor did pirates if they could help it) and pirates didn't generally fight in set piece battles on land. In order to give a serious answer to such a question, you would first need to specify a combat situation: 17th century-style naval combat (of which the outcome would be largely dependent on the state of Japanese versus Portuguese maritime technology), open field daylight desert island team combat, single combat (with or without firearms and at what distance?), coastline castle siege, covert assassination of captain/quartermaster aboard a ship, etc...
Not sure I agree.
If I'm a trained covert assassin and I find myself in the middle of a 17th century-style naval combat involving Japanese vs. Portuguese marine technology, it seems what I ought to do is attempt to covertly board the enemy vessel and assassinate its captain. If I'm a trained covert assassin and I find myself in the middle of open field daylight desert island team combat against trained shock troops, it seems what I ought to do is run away and come back in about twelve hours and try again.
More generally, part of what a sufficiently trained combatant is taught to do is control the combat situation, rather than take it as a given.
I am talking about ninjas and pirates in any situation which doesn't involve giving all the ninjas idiot balls. So, for example, the ninjas don't swim out to sea and start throwing shurikens at pirate ships.
The best pirates could do if they somehow got advance notice that the ninjas were out to get them is to hide and stay away from any land the ninjas have access to (and can maintain reasonable intelligence on). Meanwhile they are almost no threat at all to the ninjas.
Suppose for some reason a stalemate with pirates hiding from the ninjas and restricting themselves from any activities that would give the ninjas a chance to catch them doesn't count as "ninjas are clearly better". Even then it would take very little time for the ninjas to gain sea based dominance too. They are a paramilitary organisation with government support. They'll quickly get better ships, easily get sailing training or just conscript sailors and then go hunt down the pirates.
If ninjas and pirates get into conflict the pirates just lose. Fortunately for pirates they aren't out to win - they are out to get booty. Pirates prey on the weak or undefended and avoid fighting the powerful - be it ninjas, armies, navies, heavily defended settlements or sith lords. They also make sure they don't do anything that pushes them across the threshold of being a nuisance to the powerful and a serious threat that needs to be dealt with by elite forces. (Like ninjas.)
Let's say the pirates have a secret, semi-fortified island base. Ninja objective is to eliminate the threat of pirates to local shipping; pirate objective is to escape with as much of the loot as possible, or failing that, a workable ship and their lives. Ideally the pirates would also like to have a base from which to operate, but such a location is only really useful if it hasn't already been located by enemy forces.
This also applies to my paradigm of "pirate".
You might be thinking of privateers.
Yes. There was a very blurry distinction between the two while England wanted to encourage piracy against France.
I don't think that word means what you think it means.
Pirates were often privateers, which fit that definition very well.
Furthermore, the overlap between piracy and warfare meant that many pirates served in paramilitary forces before, after, or during their pirate days. For an illustration, note this description of the Battle of New Orleans, including a prominent pirate acting as an irregular military advisor / commander, and many of his compatriots serving even in land-based artillery companies.
I sort of assumed half the pirate crew would just BE ninjas in disguise, and then the other half would just be dead.
I was mistaken. I'm amazed how much debate the question prompted here even with this framing. I really thought it was just a closed question.
I'm amazed at your amazement.
I'd have expected at least this much out of any such silly comparison, even here. Try:
Or any other conflict between trochees.
Most of your amazement can be explained by you thinking that that 'pirates vs ninjas' belongs in the same reference class as:
That seems utterly ridiculous. Are you being disingenuous or are you serious?
I'm being serious. See http://xkcd.com/856/
I think the relevant reference class is "trochees with a 'vs.' between them", and you can find many such engaging debates on the Internet. I'm skeptical that anyone would care to compare pirates and ninjas if they were called something else.
I don't think the trochee pattern is so critical to these debates. Cavemen vs. astronauts is a notorious counterexample.
Oddly, I don't think I've encountered anything vs. astronauts before.
I retract my statement. I've seen the entirety of Angel.
Cavemen is a trochee. Astronauts is a dactyl, which is why it's less funny, but still close enough to a trochee.
Well cavemen are well-known in the literature for their pterodactyl-defeating skills, so I suppose that would generalize to other dactyls.
Like thomblake, I'm amazed at your amazement, but on a different track. Unless you're an expert in the histories of navies, espionage, and other miscellany, why would you expect your intuition to both identify closed questions and the correct answers?
The identity-memes are the only reason the question even exists. Historical pirates were mostly desperate or ambitious but otherwise ordinary sailors, and usually had pretty short careers. Historical ninja were usually dirt-poor burakumin without much in the way of reliable support, and -- in common with a lot of other historical assassins -- were individually used more as ammunition than as soldiers. I'm having a hard time coming up with a reasonable scenario in which they (as opposed to the pop-cultural image of either one) would have any incentive to fight, and if you stretch to create one the outcome would be almost completely determined by the circumstances.
Nornangest!History seems to be different to Wikipedia!History. My limited familiarity only extends to the latter.
Presuming you're talking about my take on ninja I'm informed mainly by martial arts lore, which (outside the schools calling themselves ninjutsu, which incidentally are almost all modern inventions) is probably a little more interested in demystifying the tradition than the Wikipedia authors are. The wiki definitely puts a more glamorous spin on it, but reading between the lines I don't see too much that's actually incompatible with my take -- note the emphasis on infiltration and sabotage rather than combat, and that the field agents were mostly drawn from the lower class.
Well, piracy was a huge thing in Japan and China and so on, so if there were any conflicts between them, they could have been recorded. But I don't see why they would have been- typically, there was no significant benefit to killing a pirate captain (rather than sinking his boat or hanging him and all of his crew), and so assassins would only be employed against people whose deaths were meaningful enough (generals, title-holders, etc.). Similarly, I doubt ninja would transport valuable cargo all that frequency, and typically it was Japanese pirates preying on Chinese vessels anyway.
I mean, pirate just means "sea bandit" but evokes the image of mostly European sea bandits at a time when navies were based far away, coastal land was essentially free, and cargoes were really valuable, because that's when there was a veneer of excitement over lowlifes murdering for fun and profit.
Even that's pretty time-period dependent; check out e.g. Jean Lafitte's utterly ridiculous career.
Well, and now the question "How can we teach the skill Check Consequentialism?" has degenerated into an erudite debate on pirates vs. ninjas.
I have never, ever been tempted to say this before on LW but WELCOME TO REDDIT.
Edit: The conversation seems much more intelligent than average Reddit, but I still think we're solving the wrong problem here.
Edit 2: And now, no longer feeling as encouraged by the 150 comments I saw when I checked back in.
On the plus side, I'll concede we've demonstrated beyond a reasonable doubt that "pirates vs. ninjas" can be an argument -generator for all audiences.
puh-lease. There were pirates vs. ninjas debates on the Internet long before Reddit existed.
You happen to have carved out a small portion of the Internet, a medium that aside from porn is primarily for pirates vs. ninjas debates, and declared it's for some other purpose. That doesn't mean you're allowed to be surprised when pirates vs. ninjas debates happen.
And this site's software based on Reddit. Is "WELCOME TO REDDIT" even worth saying when there's a "powered by reddit" icon on the bottom-right of the site?
Now, back to the local version of pirates-vs-ninjas, which sometimes looks deceptively like discussion of rationality...
Is he allowed to be surprised when lesswrong porn happens?
I think porn itself has somehow managed to stay off Less Wrong long enough to warrant surprise.
But no, it's not warranted to believe that porn about Less Wrong does not exist. Rule 34.
It's not porn, but did you see Yvain's drawing?
Thanks, I hadn't seen that. Though for maximal thread derailment, the image really should've appeared right in your comment.
I've joked in the past about writing YudkowskyxBostrom slash fics; I swear to Bayes, if people keep annoyingly discussing LW porn, I will write it!
Corollary to Rule 35. You have to now.
I'd write it, but I'm too busy working on that fic where Harry has to marry Draco to save someone from Azkaban.
Please include a threesome with Vassar.
Besides, it's not like we're having really ridiculous thread-derailing discussions. It's not like anyone's tried to claim something insane, like that Twilight Sparkle is the best pony, or a plane on a treadmill will be able to take off, or that billy goats are not delicious.
Reading this prompted me to ask myself a similar question:
Could wind powered plane on the ground traveling directly downwind take off? My answer is yes, and science is awesome! but I expect I'd get into arguments about it with even my educated friends who would say "no".
Perhaps even better, a wind-powered car can travel faster than the wind downwind! link
This seems to imply that you have some other mechanism for the plane to take off than by harnessing that very mechanism with enough efficiency and elegance that it can generate lift to take off engine free, powered by wind, with the direct force of the wind actually counting against it. Either that or you evaluate engineering coolness very differently.
I know you are mostly just building up to a "Twilight Sparkle" joke but I'm going to express agreement with this anyway. The main way that this thread has gone off topic is in as much as the skill that daenyrs is testing and training isn't 'consequentialism' it is a different rationalist skill.
Technically, both pirates and ninjas, as they are depicted in popular folklore, are fictional entities. While people named "pirates" and "ninjas" did exist (and do exist to this day), they bear little resemblance to grog-swilling scallywags or black-clad wire-fu artists or what have you. Thus, before we can answer your question, we need to nail down exactly what you mean by "pirates" and "ninjas"; what capabilities you expect these fictional combatants to possess, and then go from there.
I can't believe you missed the chance to say, "Taboo pirates and ninjas."
Oooh yes. Upvoted for awesomeness.
Taboo pirate ninja badger mushroom narwhal!
Doesn't work - the accent is on the second syllable.
See, in context I was tricked into reading that Taboo, which is totally not natural.
"Pirates versus Ninjas is the Mind-Killer"
If the world were going to end right after I took an action, which action would I choose? (Alt: If everybody saw what choice I was about to make, but then circumstances changed and my decision turned out not to matter, what choice would I want to have made?)
Did answering that question feel the same as answering the actual question? If so, I'm not really thinking about consequences.
I think you're onto something good here. Given any question, there are probably lots of hypothetical variations, like the world-ending or the exposure to everyone's judgement which you mention, which shouldn't make a difference but do, or should make a difference but don't. Maybe list a few more such circumstances and get the class to decide whether and why the variations make a difference.
In a group, with a leader who knows the exercise:
Get a volunteer to act as a judge (or a few to act as a jury, in a large group). Have her leave the room. The leader presents the rest with a short set of Contrived Hypothetical Situations, each with finite options and either clearly-defined outcomes for each option, or a probabilistic distribution of outcomes for each option. The leader says, "Please write down your choice for each problem, sign your paper, and turn it in to me. Then I'll call in the judge, and have her decide on each problem. You get a point wherever her decision agrees with yours. The winner is the one with the most points." When the judge is called in, however, the leader doesn't tell them the actual problems. Rather, the leader just reports the outcomes (or distributions), and asks them to choose which outcome or distribution is best. The winners are announced based on that.
Example: One of the situations given is some variant of the trolley problem. When the judge comes in, she is just asked whether she'd prefer one person to get hit by a trolley, or five. Everybody laughs as she replies "...one?"
Example: The problem given to the group is "You drive 45 minutes away from home to go to a new restaurant for dinner. When you get there, you discover that you dislike the ambience and the selection is poor. You remember that you have decent leftovers at home. You're mildly hungry. Do you try the restaurant anyway (25-minute wait, 10% very enjoyable meal, 10% decent meal, 80% unenjoyable meal) or just head back home (5-minute-prep once you get home, 100% chance decent meal)?" The problem given to the judge is "You're mildly hungry. In 25 minutes, you can have a meal that is (10% very enjoyable, 10% decent, 80% unenjoyable). Or, in 50 minutes, you can have a guaranteed decent meal."
Two details the judge isn't told about are 1) you would have to pay for the former meal, but not for the latter, and 2) if you stay in the restaurant, you gain useful information you'll be able to take in account the next time you might want to eat there.
1) is patchable by specifying that the leftovers are non-perishable, so eating them is equivalent to buying a meal.
2) Either the judge is told that the variable meal is repeatable if it's good, or we specify in the group problem that you're not going back there no matter what.
If you stay, you gain information about the restaurant. There's the dollar cost of dining out. It's actually not as easy as it looks to generate a "clean" example.
How much need we worry about excluding consequences we can't consciously list and/or quantify?
I patched this example by saying "you're on vacation in another city", so the value of information is mostly negligible.
But yeah, it's still pretty hard. Also, ideally not all of our examples end up being instances of sunk-cost-fallacy.
Beware of motivated stopping. If someone wants to do A, because B will happen, that is only the beginning. There are several directions it's worth exploring further, with one person exploring and another prompting them with questions such as these:
Will B actually happen (or be more likely to), given A?
What makes B a desired consequence? Some further consequence that it leads to? Some larger purpose for which B is a means? Or is B terminally desirable?
At some point one has to stop, but the very first consequences one thinks of may not be that point.
To me, this comes down to what I am trying to learn as my anti-akrasia front kick: I cache the question "Why am I doing what I am doing?". While I lose some amount of focus to the question itself, I have gained key insights into many of my worst habits. For instance, my employer provides free soft drinks- I found that I would end up with multiple, open drinks at my desk. The cached question revealed I was using the action of getting a drink whenever I felt the need to stretch and leave my desk. Browsing reddit too much at work- cached question can catch it. Eventually, when I have affirmative answers for the question, it no longer even draws focus away from the task at hand- it is simply an itch that is easily scratched, as I know I am doing something that accomplishes a larger goal.
I don't have suggestions on the main question, but I strongly recommend that you design the curricula to be consumable by corporate executives. If you can convince companies that they need to send their execs to week-long rationality seminars, to help them improve their decision-making skills, you are going to make megabucks.
Not attempting to answer the question, but I've been nursing a thought about the rationality org that might be useful: The nearby Stanford University has a world-renown program in "decision sciences" http://decision.stanford.edu/ which is basically "how to make decisions scientifically"; they virtually invented influence diagrams, they teach biases as a part of the program, etc. The head of the program, Ronald Howard, also co-founded http://www.decisioneducation.org/ , his teen-oriented "rationality org".
there is probably things to learn from both
if "rationality org" has a value proposition to these organizations they can be useful in teaching opportunities and for credibility building
These do indeed sound like people to talk to, and local too - thanks!
Please tell me how that goes. Also, you planning to look into this, or Anna et al?
The reading list (books) from decisioneducation.org.
A comma broke one of the links in your post.
Your check consequentialism sounds a lot like risk management. Risk is the effect of uncertainty on objectives (ISO 31 000). The risk management process involves indentifying risks, analysing how significant they are, and then treating the big ones so that they don't prevent you from attaining your objective. This is fairly straightforward to do. The difficult part is building a risk management culture where the risks are considered before making a decison, embarking on a project, etc. Just identifying the risks is often the big deal. Once you are aware that a risk exists you will probably deal with it. Sorry that I have not given you an activity, but perhaps I have given you a useful keyword to help your search.
What kinds of exercises you use to teach a skill like "checking consequentialism" should probably be placed in the greater context of a rationality curriculum. You have to know where the students are coming from at each step.
That said-- making the assumption that the students are already familiar with the theory of heuristics and biases, and just need to learn how to apply them-- I think most of these can be taught with similar kinds of hypotheticals and problems.
For checking consequentialism, you might want to focus on problems involving sunk costs. To illustrate how sunk costs can affect how someone automatically approaches a problem, split students up into groups (they will probably produce worse results as a group, which is good for this part) and give them all similar problems, with the initial conditions modified slightly. Example: "John has been working on his PhD for X years, and expects to finish in Y. He knows W and Z facts about how his degree will benefit him." Modify parts of the problem, X and Y especially, to try and prime their System 1 for a different result. Have them make a decision quickly. Reconvene, discuss the problem, point out the issues with sunk costs and why the groups did or didn't reach a different result.
This is just a starting activity; it could be followed by having students do more hypotheticals individually. The instructor needs to give a lot of feedback on the problems as they go, asking students key questions that they might not have thought of, so they'd ideally be well trained in a) rationality and b) drawing out student thinking.
Ideally the exercises wouldn't require too much instructor skill, though, so I'll think about this some more.
I think an important pre-skill is to decide what your goals are BEFORE checking the consequences. Otherwise, you can easily retroactively change your goals to fit whatever action you want.
For example, in the sunk costs PhD scenario, it would be easy for someone to say something like "If I pursue my PhD, it will make my family proud. This is very important to me, so I will pursue my PhD." However if you asked them what their goals are BEFORE they listed the consequences of each action they probably would not list "Make family proud", but would list things like "Job I enjoy" and "Make more money in my career".
The trap is that it can be very difficult to admit to your ACTUAL goals. For example, it took a while for me to realize that I don't ACTUALLY care about making money once I have enough to survive (unless phrased in a way such as "If I do Job B, I only have to work 12 hours/week" or somesuch).
So I think figuring out what your actual goals are (noit the cached thoughts of your goals) is a set of exercises itself, but that they should probably be done before the Consequentialism exercises.
Conversely, you might not recognize your true rejection until you go "well, but wait, I don't want to abandon my PhD - my family wouldn't be proud of me." If you get stuck on that, it's possible that "make your family proud" really IS important to you.
I do agree that it's a useful idea, but new insights aren't necessarily ex post facto rationalizations.
Why is teaching people to think like consequentialists a good idea again? Serious question.
If they're (relatively successful) mathematicians and programmers I don't see how it could go wrong but I'm awfully worried about some of the rest of the population. Specifically people not being able to sustain it without letting other things slip.
second edit: I should clarify. It's teaching the habit that I'm thinking about. Everyone should be able to think like a consequentialist but is instilling these reflexes gonna be a net positive?
Umm, are you under the impression that (the non-mathematical-ish part of the population/anyone) is constantly operating near their sustainable cognitive maxima? So near that adding a nearly-automatic reflex would push them over?
Neither looking around nor introspection suggests that that is true.
Indeed. That would imply that our shared goal of raising the sanity waterline would cause most of the population to drown :)
Mind you, I like that the OP is asking what the consequences would be. However my guess is: more people making slightly better decisions some of the time, and with no obvious mechanism for "letting other things slip", I don't see a downside.
Devil's Advocating Here:
I do think we need to not forget that most people's minds do NOT operate like the typical LWian!
I know this personally, in that I tend to make intuitive-based decisions (and by intuitive, I mean things like waking up one morning thinking "I should eat less meat", and so becoming a vegetarian for the next 8 (so far) years.)
Decisions I have made intuitively like this include: atheism, vegetarianism, not drinking alcohol (that one only lasted 7 years), quitting grad school, not having children, polyamory, pretty much every career decision, liking rationality.
The social situations were different enough for each of these for me to think that social concerns were not the main trigger, but I somehow feel like I've ended up making rational-style choices by following my intuition (and I recognize that I probably just lost all credibility I might have had here by writing this post :P).
The upside of this, is that since I am always doing what I feel like, I rarely feel like I am having to fight myself. For example, giving up meat was amazingly easy for me, because it was like the decision had already been made.
From someone who really enjoys learning about rationality, I can still see how it wouldn't mesh with many people's methods of living, without a complete lifestyle overhaul (which is an unlikely result of a single class).
But I do not AT ALL think that this means that we shouldn't teach people about consequentialism or other rationality topics (I am all about spreading rationality), I just think we need to make sure that we do so in a way that can encompass a wide range of people. First figure out what percentage of the overall population you want to be accessible to, (say the top 60% intellectually, MINUS the 20% most intuitive types) and make sure that your presentations and materials are able to reach whatever your target is.
Your intuition appears to like LW-approved things, and you are on LW. Don't you think that learning about consequentialism might beneficially rewire one's intuition?
If you are implying that learning about rationality on LW made my intuitions more rational, then you should know that I made all those decisions long before joining LW about 5 months ago.
However, I wouldn't be surprised with the fact the LW conforms to most of my intuitions (except the whole anti-deathism and singularity stuff) as one of the reasons I joined the site. I remember thinking "OMGWTF there are people who are OPENLY POLY on here, outside of a poly-specific group!!!" I'm sure it doesn't hurt that I find rationality and psychology to be amazingly interesting.
Well, when teaching non-perfect people about consequentialism you should teach them about ethical injunctions as well. I don't think teaching both will be a net negative.
Many of the pain points listed have a common trait: the decision would seem easier with less information. For example, the PhD decision is easier if you didn't know about the costs which have been sunk, the identity decisions are easier if you're not sure of your own identity, cached thought problems are easier without having that thought cached, etc...
But we know that information should never have negative value. So why not highlight that dissonance? Imagine the following exercise:
Handout: "You spent the last 3 years working toward a PhD. You passed up a $90k job to stay in the program. Now you have 2 years left, and another $90k job offer has come your way. Do you take it?" (I don't know much about PhD programs, so feel free to imagine more plausible numbers here and add narrative).
Exercise 1: Is there any information you would prefer to not know? Exercise 2: How much would you pay to not know it?
If you really want to have fun, give people monopoly money and let them bid to remove information from a range of scenarios. Note that we're not offering to change the facts, just to not know them.
Personally, I think this would be a lot easier if I could just forget about all that time spent in the PhD program.
At least in this case, the exercise highlights the difference between consequentialist and non-consequentialist reasons/excuses for doing things. The "how much would you pay to not know it" is especially handy, since it puts a number on that mental pain. Then we can ask whether the mental pain is worth all the money you'd lose by, in this example, staying in the PhD program.
I'm not sure this works well - last time "I" made a decision, "I" preferred five years of work for a PhD title to a $90k job now. It would seem unlikely that I'd prefer a $90k job now over two years of work for a PhD title, especially given that I'm now more sure that there are good jobs waiting for me.
Why the fancy words? This just seems like a complicated way of saying: "Because the person would then be dead. And that is bad".
People being dead is a bad outcome. Killing people is a bad action. Consequentialism does not recognize bad actions, only actions that lead to bad outcomes.
A human corpse poofing into existence from nowhere wouldn't be in itself a bad outcome. So we need to specify that the human was once alive.
An alternate phrasing might be "Because this would cause the person to die." But the word "die" is historically imprecise. Open-heart surgery stops a beating heart. Destructive uploading would cause brain death.
Free food! (It doesn't count as cannibalism if the corpse has never been a member of your species!)
Nobody said "free." The operational costs of corpse-poofing might be prohibitive.
"the person would then be dead" seems to pretty clearly imply that there was a person involved. In the case where a corpse poofs into existence from nowhere, there doesn't seem to have ever been a person involved. I conclude that "Because the person would then be dead" doesn't apply to the case where a corpse poofs into existence from nowhere. So I'm not sure why we would need to further specify anything here.
All of that said, the whole approach of counting deaths as negative utility seems to me to be rescuing the wrong part of the original nonconsequentialist claim in the first place.
It's clear that one consequence of increasing the human population from 1 billion people to 7 billion people is that many more people die per unit time, but it doesn't follow from that fact that we should reject increasing human population on consequentialist grounds. (It might be true that we should so reject it, but even if true it doesn't follow from that fact.)
It seems that the part we would want to rescue from a consequentialist POV is the idea that more life-years is good, so any act that reduces expected net lifeyears is bad... and also, perhaps, the idea that more life-years/person is good, so any act that reduces expected net lifeyears/person is bad.
This would also render all concerns about how we define "death" moot.