A New Interpretation of the Marshmallow Test
I've begun to notice a pattern with experiments in behavioral economics. An experiment produces a result that's counter-intuitive and surprising, and demonstrates that people don't behave as rationally as expected. Then, as time passes, other researchers contrive different versions of the experiment that show the experiment may not have been about what we thought it was about in the first place. For example, in the dictator game, Jeffrey Winking and Nicholas Mizer changed the experiment so that the participants didn't know each other and the subjects didn't know they were in an experiment. With this simple adjustment that made the conditions of the game more realistic, the "dictators" switched from giving away a large portion of their unearned gains to giving away nothing. Now it's happened to the marshmallow test.
In the original Stanford marshmallow experiment, children were given one marshmallow. They could eat the marshmallow right away; or, if they waited fifteen minutes for the experimenter to return without eating the marshmallow, they'd get a second marshmallow. Even more interestingly, in follow-up studies two decades later, the children who waited longer for the second marshmallow, i.e. showed delayed gratification, had higher SAT scores, school performance, and even improved Body Mass Index. This is normally interpreted as indicating the importance of self-control and delayed gratification for life success.
Not so fast.
In a new variant of the experiment entitled (I kid you not) "Rational snacking", Celeste Kidd, Holly Palmeri, and Richard N. Aslin from the University of Rochester gave the children a similar test with an interesting twist.
They assigned 28 children to two groups asked to perform art projects. Children in the first group each received half a container of used crayons, and were told that if they could wait, the researcher would bring them more and better art supplies. However, after two and a half minutes, the adult returned and told the child they had made a mistake, and there were no more art supplies so they'd have to use the original crayons.
In part 2, the adult gave the child a single sticker and told the child that if they waited, the adult would bring them more stickers to use. Again the adult reneged.
Children in the second group went through the same routine except this time the adult fulfilled their promises, bringing the children more and better art supplies and several large stickers.
After these two events, the experimenters repeated the classic marshmallow test with both groups. The results demonstrated children were a lot more rational than we might have thought. Of the 14 children in group 1, who had been shown that the experimenters were unreliable adults, 13 of them ate the first marshmallow. 8 of the 14 children in the reliable adult group, waited out the fifteen minutes. On average children in unreliable group 1 waited only 3 minutes, and those in reliable group 2 waited 12 minutes.
So maybe what the longitudinal studies show is that children who come from an environment where they have learned to be more trusting have better life outcomes. I make absolutely no claims as to which direction the arrow of causality may run, or whether it's pure correlation with other factors. For instance, maybe breastfeeding increases both trust and academic performance. But any way you interpret these results, the case for the importance and even the existence of innate self-control is looking a lot weaker.
A Gamification Of Education: a modest proposal based on the Universal Decimal Classification and RPG skill trees
While making the inventory of my personal library and applying the Universal Decimal System to its classification, I found myself discovering a systematized classification of fields of knowledge, nested and organized and intricate, many of which I didn't even know existed. I couldn't help but compare how information was therein classified, and how it was imparted to me in engineering school. I also thought about how, often, software engineers and computer scientists were mostly self-thought, with even college mostly consisting of "here's a problem: go forth and figure out a way to solve it". This made me wonder whether another way of certified and certifiable education couldn't be achieved, and a couple of ideas sort of came to me.
It's pretty nebulous in my mind so far, but the crux of the concept would be a modular structure of education, where the academic institution essentially established what information precisely you need from each module, and lets you get on with the activity of learning, with periodic exams that you can sign up for, which will certify your level and area of proficiency in each module.
A recommended tree of learning can be established, but it should be possible to not take every intermediate test, if passing the final test proves that you've passed all the others behind it (this would allow people coming from different academic systems to certify their knowledge quickly and easily, thus avoiding the classic "Doctor in Physics from Former Soviet Union, current Taxi Driver in New York" scenario).
Thus, a universal standard of how much you have proven to know about what topics can be established.
Employers would then be free to request profiles in the format of such a tree. It need not be a binary "you need to have done all these courses and only these courses to work for us", they could be free to write their utility function for this or that job however they would see fit, with whichever weights and restrictions they would need.
Students and other learners would be free to advance in whichever tree they required, depending on what kind of profile they want to end up with at what age or point in time. One would determine what to learn based on statistical studies of what elements are, by and large, most desired by employers of/predictors of professional success in a certain field you want to work in.
One would find, for example, that mastering the peculiar field of railway engineering is great to be a proficient railway engineer, but also that having studied, say, things involved with people skills (from rhetoric to psychology to management), correlates positively with success in that field.
Conversely, a painter may find that learning about statistics, market predictions, web design, or cognitive biases correlates with a more successful career (whether it be on terms of income, or in terms of copies sold, or of public exposure... each one may optimize their own learning according to their own criteria).
One might even be able to calculate whether such complimentary education is actually worth their time, and which of them are the most cost-efficient.
I would predict that such a system would help society overall optimize how many people know what skills, and facilitate the learning of new skills and the updating of old ones for everyone, thus reducing structural unemployment, and preventing pigeonholing and other forms of professional arthritis.
I would even dare to predict that, given the vague, statistical, cluster-ish nature of this system, people would be encouraged to learn quite a lot more, and on a quite wider range of fields, than they do now, when one must jump through a great many hoops and endure a great many constraints in space and time and coin to get access to some types of educations (and to the acknowledgement of their acquisition thereof).
Acquiring access to the actual sources of knowledge, a library (virtual or otherwise), lectures (virtual or otherwise), and so on, would be a private matter, up to the learner:
- some of them already have the knowledge and just need to get it certified,
- others can actually buy the books they want/need, especially if keeping them around as reference will be useful to them in the future,
- others can subscribe to one or many libraries, of the on-site sort or by correspondence
- others can buy access to pre-recorded lectures, peruse lectures that are available for free, or enroll in academic institutions whose ostensible purpose is to give lectures and/or otherwise guide students through learning, more or less closely
- the same applies to finding study groups with whom you can work on a topic together: I can easily imagine dedicated social networks could be created for that purpose, helping people pair up with each other based on mutual distance, predicted personal affinity, mutual goals, backgrounds, and so on. Who knows what amazing research teams might be borne of the intellectual equivalent of OK!Cupid.
A thing that I would like very much about this system is that it would free up the strange conflicts of interest that hamper the function of traditional educational institutions.
When the ones who teach you are also the ones who grade you, the effort they invest in you can feel like a zero-sum game, especially if they are only allowed to let a percentage of you pass.
When the ones who teach you have priorities other than teach (usually research, but some teachers are also involved in administrative functions, or even private interests completely outside of the university's ivory tower1), this can and often does reduce the energy and dedication they can/will allocate to the actual function of teaching, as opposed to the others.
By separating these functions, and the contradictory incentives they provide, the organizations performing them are free to optimize for each:
- Testing is optimized for predicting current and future competence in a subject: the testers whose tests are the most reliable have more employers requiring their certificates, and thus more people requesting that they test them
- Teaching is optimized for getting the knowledge through whatever the heck the students want, whether it be to succeed at the tests or to simply master the subject (I don't know much game theory, but I'd naively guess that the spontaneous equilibrium between the teaching and testing institutions would lead to both goals becoming identical).
- Researching is optimized for research (researchers are not teachers. dang it, those are very different skill-sets!). However researchers and other experts get to have a pretty big say in what the tests test for and how, because their involvement makes the tests more trustworthy for employers, and because they, too, are employers.
- And of course entire meta-institutions can spring from this, whose role is to statistically verify, over the long term,
- how good a predictor of professional success in this or that field is passing the corresponding test, and
- how good a predictor of passing the test is to be taught by this or that teaching institution.
- how good a predictor of the test being reliable is the input of these or those researchers and experts
- It occurs to me now that, if one wished to be really nitpicky about who watches the watchmen, I suspect that there would be institutions testing the reliability of those meta-institutions, and so on and so forth... When does it stop? How to avoid vested interests and little cheats and manipulations pulling an academic equivalent of the AAA certification of sub-prime junk debt in 2008?
Another discrepancy I'd like to see solved is the difference between the official time it is supposed to take to obtain this or that degree, to learn this or that subject, and the actual statistical distribution of that time. Nowadays, a degree that's supposed to take you five years ends up taking up eight or ten years of your life. You find yourself having to go through the most difficult subjects again and again, because they are explained in an extremely rushed way, the materials crammed into a pre-formatted time. Other subjects are so exceedingly easy and thinly-spread that you find that going to class is a waste of time, and that you're better off preparing for it one week before finals. Now, after having written all of the above, my mind is quite spent, and I don't feel capable of either anticipating the effect of my proposed idea on this particular, nor of offering any solutions. Nevertheless, I wish to draw attention to this, so I'm leaving this paragraph in until I can amend it to something more useful/promising.
I hereby submit this idea to the LW community for screening and sound-boarding. I apologize in advance for your time, just in case this idea appears to be flawed enough to be unsalvageable. If you deem the concept good but flawed, we could perhaps work on ironing those kinks together. If, afterwards, this seems to you like a good enough idea to implement, know that good proposals are a dime a dozen; if there is any interest in seeing something like this happen, we can need to move on to proprely understanding the current state of secondary/superior/higher education, and figuring out of what incentives/powers/leverages are needed to actually get it implemented.
1By ivory tower I simply mean the protected environment where professors teach, researchers research, and students study, with multiple buffers between it and the ebb and flow of political, economical, and social turmoil. No value judgement is intended.
EDIT: And now I look upon the title of this article and realize that, though I had comparisons to games in mind, I never got around to writing them down. My inspirations here were mostly Civilization's Research Trees, RPG Skill Scores and Perks, and, in particular, Skyrim's skills and perks tree.
Basically, your level at whatever skill improves by studying and by practising it rather than merely by levelling up, and, when you need to perform a task that's outside your profile, you can go and learn it without having to commit to a class. Knowing the right combination of skills at the right level lets you unlock perks or access previously-unavailable skills and applications. What I like the most about it is that there's a lot of freedom to learn what you want and be who you want to be according to your own tastes and wishes, but, overall, it sounds sensible and is relatively well-balanced. And of course there's the fact that it allows you to keep a careful tally of how good you are at what things, and the sense of accomplishment is so motivating and encouraging!
Speaking of which, several netwroks and consoles' Achievement systems also strike me as motivators for keeping track of what one has achieved so far, to look back and be able to say "I've come a long way" (in an effect similar to that of gratitude journals), and also to accomplish a task and have this immediate and universal acknowledgement that you did it dammit (and, for those who care about that kind of thing, the chance to rub it the face of those who haven't).
I would think our educational systems could benefit from this kind of modularity and from this ability to keep track of things in a systematic way. What do you guys think?
How to Have Space Correctly
[NOTE: This post has undergone substantial revisions following feedback in the comments section. The basic complaint was that it was too airy and light on concrete examples and recommendations. So I've said oops, applied the virtue of narrowness, gotten specific, and hopefully made this what it should've been the first time.]
Take a moment and picture a master surgeon about to begin an operation. Visualize the room (white, bright overhead lights), his clothes (green scrubs, white mask and gloves), the patient, under anesthesia and awaiting the first incision. There are several other people, maybe three or four, strategically placed and preparing for the task ahead. Visualize his tools - it's okay if you don't actually know what tools a surgeon uses, but imagine how they might be arranged. Do you picture them in a giant heap which the surgeon must dig through every time he wants something, or would they be arranged neatly (possibly in the order they'll be used) and where they can be identified instantly by sight? Visualize their working area. Would it be conducive to have random machines and equipment all over the place, or would every single item within arms reach be put there on purpose because it is relevant, with nothing left over to distract the team from their job for even a moment?
Space is important. You are a spatially extended being interacting with spatially extended objects which can and must be arranged spatially. In the same way it may not have occurred to you that there is a correct way to have things, it may not have occurred to you that space is something you can use poorly or well. The stakes aren't always as high as they are for a surgeon, and I'm sure there are plenty of productive people who don't do a single one of the things I'm going to talk about. But there are also skinny people who eat lots of cheesecake, and that doesn't mean cheesecake is good for you. Improving how you use the scarce resource of space can reduce task completion time, help in getting organized, make you less error-prone and forgetful, and free up some internal computational resources, among other things.
What Does Using Space Well Mean?
It means consciously manipulating the arrangement, visibility, prominence, etc. of objects in your environment to change how they affect cognition (yours or other people's). The Intelligent Use of Space (Kirsh, "The Intelligent Use of Space", 1995) is a great place to start if you're skeptical that there is anything here worth considering. It's my primary source for this post because it is thorough but not overly technical, contains lots of clear examples, and many of the related papers I read were about deeper theoretical issues.
The abstract of the paper reads:
How we manage the spatial arrangement of items around us is not an afterthought: it is an integral part of the way we think, plan, and behave. The proposed classification has three main categories: spatial arrangements that simplify choice; spatial arrangements that simplify perception; and spatial dynamics that simplify internal computation. The data for such a classification is drawn from videos of cooking, assembly and packing, everyday observations in supermarkets, workshops and playrooms, and experimental studies of subjects playing Tetris, the computer game. This study, therefore, focuses on interactive processes in the medium and short term: on how agents set up their workplace for particular tasks, and how they continuously manage that workplace.
The 'three main categories' of simplifying choice, perception, and internal computation can be further subdivided:
simplifying choice
reducing or emphasizing options.
creating the potential for useful new choices.
simplifying perception
clustering like objects.
marking an object.
enhancing perceptual ability.
simplfying internal computation
doing more outside of your head.
These sub-categories are easier to picture and thus more useful when trying to apply the concept of using space correctly, and I've provided more illustrations below. It's worth pointing out that (Kirsh, "The Intelligent Use of Space", 1995) only considered the behavior of experts. Perhaps effective space management partially explains expert's ability to do more of their processing offline and without much conscious planning. An obvious follow up would be in examining how novices utilize space and looking for discrepancies.
What Does Using Space Well Look Like?
The paper walks the reader through a variety of examples of good utilization of space. Consider an expert cook going through the process of making a salad with many different ingredients, and ask how you would accomplish the same task differently:
...one subject we videotaped, cut each vegetable into thin slices and laid them out in tidy rows. There was a row of tomatoes, of mushrooms, and of red peppers, each of different length...To understand why lining up the ingredients in well ordered, neatly separated rows is clever, requires understanding a fact about human psychophysics: estimation of length is easier and more reliable than estimation of area or volume. By using length to encode number she created a cue or signal in the world which she could accurately track. Laying out slices in lines allows more precise judgment of the property relative number remaining than clustering the slices into groups, or piling them up into heaps. Hence because of the way the human perceptual system works, lining up the slices creates an observable property that facilitates execution.
Here, the cook used clustering and clever arrangement to make better use of her eyes and to reduce the load on her working memory, techniques I use myself in my day job. As of this writing (2013) I'm teaching English in Korea. I have a desk, a bunch of books, pencils, erasers, the works. All the folders are together, the books are separated by level, and all ungraded homework is kept in its own place. At the start of the work day I take out all the books and folders I'll need for that day and arrange them in the same order as my classes. When I get done with a class the book goes back on the day's pile but rotated 90 degrees so that I can tell it's been used. When I'm totally done with a book and I've entered homework scores and such, it goes back in the main book stack where all my books are. I can tell at a glance which classes I've had, which ones I'll have, what order I'm in, which classes are finished but unprocessed, and which ones are finished and processed. Cthulu only knows how much time I save and how many errors I prevent all by utilizing space well.
These examples show how space can help you keep track of temporal order and make quick, accurate estimates, but it may not be clear how space can simplify choice. Recall that simplifying choice usually breaks down into either taking some choices away or making good choices more obvious. Taking choices away may sound like a bad thing, but each choice requires you to spend time evaluating options, and if you are juggling many different tasks the chance of making the wrong choice goes up. Similarly, looking for good options soaks up time, unless you can find a way to make yourself trip over them.
An example of removing bad decisions is in factory workers placing a rag on hot pipes so they know not to touch them (Kirsh, "The Intelligent Use of Space", 1995). And here is how some carpenters structure their work space so that they can make good uses for odds and ends easier to see:
In the course of making a piece of furniture one periodically tidies up. But not completely. Small pieces of wood are pushed into a corner or left about; tools, screw drivers and mallets are kept nearby. The reason most often reported is that 'they come in handy'. Scraps of wood can serve to protect surfaces from marring when clamped, hammered or put under pressure. They can elevate a piece when being lacquered to prevent sticking. The list goes on.
By symbolically marking a dangerous object the engineers are shutting down the class of actions which involves touching the pipe. It is all too easy in the course of juggling multiple aspects of a task to forget something like this and injure yourself. The strategically placed and obvious visual marker means that the environment keeps track of the danger for you. Likewise poisonous substances have clear warning labels and are kept away from anything you might eat; both precautions count as good use of space.
My copy of Steven Johnson's Where Good Ideas Come From is on another continent, but the carpenter example reminded me of his recommendation to keep messy notebooks. Doing so makes it more likely you'll see unusual and interesting connections between things you're thinking about. He goes so far as to use a tool called DevonThink which speeds this process up for him.
And while I'm at it, this also points to one advantage of having physical books over PDFs. My books take up space and are easier to see than their equivalent 1's and 0's on a hard drive, so I'm always reminded of what I have left to read. More than once I've gone on a useful tangent because the book title or cover image caught my attention, and more than one interesting conversation got started when a visitor was looking over my book collection. Scanning the shelves at a good university library is even better, kind of like 17th-century StumbleUpon, and English-language libraries are something I've sorely missed while I've been in Asia.
All this usefulness derives from the spatial properties and arrangement of books, and I have no idea how it can be replicated with the Kindle.
Specific Recommendations
You can see from the list of examples I've provided that there are a billion ways of incorporating these insights into work, life, and recreation. By discussing the concept I hope to have drawn your attention to the ways in which space is a resource, and I suspect just doing this is enough to get a lot of people to see how they can improve their use of space. Here are some more ideas, in no particular order:
-I put my alarm clock far enough away from my bed so that I have to actually get up to turn it off. This is so amazingly effective at ensuring I get up in the morning that I often hate my previous-night's self. Most of the time I can't go back to sleep even when I try.
-There's reason to suspect that a few extra monitors or a bigger display will make your life easier [Thanks Qiaochu_Yuan].
-When doing research for an article like this one, open up all the tabs you'll need for the project in a separate window and close each tab as you're done with it. You'll be less distracted by something irrelevant and you won't have to remember what you did or didn't read.
-Having a separate space to do something seems to greatly increase the chances I'll get it done. I tried not going to the gym for a while and just doing push ups in my house, managing to keep that up for all of a week or so. Recently, I switched gyms, and despite now having to take a bus all the way across town I make it to the gym 3-5 times a week, pretty much without fail. If your studying/hacking/meditation isn't going well, try going somewhere which exists only to give people a place to do that thing.
-Put whatever you can't afford to forget when you leave the house right by the door.
-If something is really distracting you, completely remove it from the environment temporarily. During one particularly strenuous finals in college I not only turned off the xbox, I completely unplugged it and put it in a drawer. Problem. Solved.
-Alternatively, anything you're wanting to do more of should be out in the open. Put your guitar stand or chess board or whatever where you're going to see it frequently, and you'll engage with it more often. This doubles as a signal to other people, giving you an opportunity to manage their impression of you, learn more about them, and identify those with similar interests to yours.
-Make use of complementary strategies (Kirsh, "Complementary Strategies", 1995). If you're having trouble comprehending something, make a diagram, or write a list. The linked paper describes a simple pilot study which involved two groups tasked with counting coins, one which could use their hands and one which could not. The 'no hands' group was more likely to make errors and to take longer to complete the task. Granted, this was a pilot study with sample size = 5, and the difference wasn't that stark. But it's worth thinking about next time you're stuck on a problem.
-Complementary strategies can also include things you do with your body, which after all is just space you wear with you everywhere. Talk out loud to yourself if you're alone, give a mock presentation in which you summarize a position you're trying to understand, keep track of arguments and counterarguments with your fingers. I've always found the combination of explaining something out loud to an imaginary person while walking or pacing to be especially potent. Some of my best ideas come to me while I'm hiking.
-Try some of these embodied cognition hacks.
Summary and Conclusion
Space is a resource which, like all others, can be used effectively or not. When used effectively, it acts to simplify choices, simplify perception, and simplify internal computation. I've provided many examples of good space usage from all sorts of real-life domains in the hopes that you can apply some of these insights to live and work more effectively.
Further Reading
[In the original post these references contained no links. Sincere thanks to user Pablo_Stafforini for tracking them down]
Kirsh, D. (1995) The Intelligent Use of Space
Kirsh, D. (1999) Distributed Cognition, Coordination and Environment Design
Kirsh, D. (1998) Adaptive Rooms, Virtual Collaboration, and Cognitive Workflow
Kirsh, D. (1996) Adapting the Environment Instead of Oneself
Kirsh, D. (1995) Complementary Strategies: Why we use our hands when we think
Changing Systems is Different than Running Controlled Experiments - Don’t Choose How to Run Your Country That Way!
Trigger warning: Discussion of rape.
Example 1:
Say that each morning you tell yourself that you are lazy for not wanting to get out of bed to go to work, as a way to convince yourself to get up. Perhaps if the only variable you changed was to lower your level of guilt, you might not get out of bed to go to work, and would instead take the day off. So if you are running a motivation system that uses guilt, feeling guilt may well be something you do not want to get rid of. If you got rid of the guilt but stopped going to work, that would likely be a net negative for your life.
To contrast, with animal training, you reinforce behavior you want in the animal, and interrupt, redirect, or completely ignore (ie: no shaming or guilting) behavior you don't want. It's also a similar methodology that meditation uses. When you meditate, you are told to focus on a meditative object such as the breath. When your mind wanders from the meditative object, you are instructed to just return your attention to the meditative object, and to not in any way punish yourself for having wandered. Also, you are instructed to not punish yourself for punishing yourself for having your mind wander. Meditation does not use reward during the meditative process, although it's common to sound a beautiful chime which will give hedons at the end of a session, and people often perform a pleasant ritual before and/or after meditation that builds positive association with the activity of meditating. Example page of meditation instructions.
So, if you switch to a positive reinforcement motivational system, such as that which animal trainers use to train dogs, then guilt is counter-productive for motivation, because it is a form of punishment.
Example Summary:
If you only change one variable from a motivation system that uses guilt, then it may break the system, and be a net negative. However, there is likely a way to get a net utility gain by changing several variables of the system, such as by switching to a positive reinforcement based system where you add instant rewards that increase hedons and remove guilt and other punishments.
Example 2:
As it stands, there are many unreported rapes in American society. This excellent article debunks many myths about rape, including the classic myth that rapes are generally done by strangers using force:
A huge proportion of the women I know enough to talk with about it have survived an attempted or completed rape. None of them was raped by a stranger who attacked them from behind a bush, hid in the back of her car or any of the other scenarios that fit the social script of stranger rape. Anyone reading this post, in fact, is likely to know that six out of seven rapes are committed by someone the victim knows.
The author goes on to explain how most rapes are from repeat offenders who by a median age of 26.5, on average rape around 5-6 women each, and that it is almost always someone who was part of the woman's social circle, and intoxicants are usually used.
The suggestion of change of system that I got from this post is actually in the title of the blog: "Yes Means Yes."
If the social rules for consent are changed from "if a woman does not say no, then it may or may not be okay" to "it is only okay if a woman says yes," then the boundary becomes a lot more clear to both parties. It would be a pretty radical system change, that would make a lot of people uncomfortable.
To be more clear - with a "Yes Means Yes" system, you don't need to have "No Means No", because sex is only had when there is a Yes. If a woman is too drunk to say or enforce no, then she is also too drunk to say yes, and sex is not had unless there is explicit consent. Having a Yes Means Yes social policy would change the onus of responsibility for making sure that sex is consensual from the woman - who is obligated to say no if she doesn't want to - to both parties who must say yes to proceed. This would not stop all rape by any means, but if implemented in a system where people were taught good communication and assertiveness, it would cut down on it. For example, instead of feeling that it was her fault because she got drunk and didn't say no aggressively enough, a woman would realize quickly, "hey, I didn't say yes!" and a predatorial guy who was one of the small percentage of men who rape women would also realize that the woman would be less likely to just feel ashamed and keep quiet and would be more likely to take action to defend herself.
Perhaps some people would be afraid that they'd remain virgins for life in this system - some men might be afraid that they'd be too shy to ever ask, some women might not feel comfortable actually admitting that they want sex. And therefore, people of both genders might be resistant to switching systems because they would imagine the switch without a complete social system switch or training. And as it stands, perhaps a lot less sex would happen at first. A system like that would require retraining a lot of society to be more assertive.
Example Summary:
Just shifting one variable and telling men to say "I only have sex when women say yes" would be very weird. If a guy tried to implement that in the current system, some people might look at him like he was crazy or even get offended.
I think the "Yes Means Yes" system would work beautifully in a society that functioned based on a different system - where the social norm, which people were trained in, was to identify and state one's desires, and to not proceed without clarity. I do think it would cut down on rape, and unreported rape.
Overall Summary:
I've discovered that when talking to people about potential novel systems, that the most common response I get is for them to say why the alternative system won't work, based on what would happen if you changed one variable of the current system to be more like the novel system. Examples: "If I didn't feel guilty, I'd never get anything done," or "In a system where you always had to have a clear yes before having sex, people would feel really awkward and uncomfortable and opt out." (Alternatively I will often hear people justify alternative systems using similar arguments about single-variable changes.)
The examples above are a couple of the more simple examples of this general principle I've been observing quite a lot lately.
Consider how this applies to government systems, and other social systems. There are so many parts dependent on each other, that it is very hard to shift any single one without creating a domino effect of other shifts. So making any argument about how changing a single variable would fix or destroy a complex system like government is usually a huge oversimplification.
To quote Einstein:
It can scarcely be denied that the supreme goal of all theory is to make the irreducible basic elements as simple and as few as possible without having to surrender the adequate representation of a single datum of experience.
My thoughts on making large-scale change, are that you need to be thinking large scale. If you want to be a change maker, it is best to start small in your actions, study and experiment a lot. Focus your studies on success and failure scenarios as close as possible to what it is you want to effect, while as diverse as possible from each other.
Running single-variable experiments is important - it is just that it is only how you understand a little corner of the problem to be solved - that's not how you find the solution itself to a problem involving a complex system.
To give a biological analogy: Cancer is what happens when a single type of cell tries to become the whole system. Running a single-variable controlled experiment to determine what type of complex system you want to choose is like trying to determine the optimal form of cancer, as opposed to looking at an entire entity. Life is complicated.
Why economics is not a morality tale
Example nicked from this online Berkeley lecture.
Monopolies are bad (morality and economics agree here).
Firms that pollute are bad (morality and economics agree here).
What about monopolies that pollute?
What about strong monopolies that pollute and receive government subsidies?
Well...
Pollution, and other negative externalities, cause firms to produce too much of their product. That's because they don't pay the full cost of the product, including the impact of pollution.
The equilibrium behaviour for monopolies is to produce too little of their product, to keep prices and profits high.
So a monopoly that pollutes is subject to two opposite tendencies: the unpriced-pollution tendency to produce too much, and the monopolistic tendency to produce too little. If the effects are of comparable magnitude, then the monopoly might be much closer to social optimum than a free market would be (the social optimum, incidentally, will generally involve some pollution: we need to accept some pollution in the production of fertiliser, for instance, in order to have enough food to stop people starving).
In fact, if the monopolistic effect is too strong, then the firm may under-produce, even taken the pollution effect into account. In that case, we can approach closer to the social optimum by... subsidising the polluting monopoly to produce more!!
And that, my friends, is why economics is not a morality tale.
Are imaginary and complex numbers of decibans meaningful?
It's well-established that 0 decibans means 1:1 odds or 50% confidence; that 10 decibans means 10:1 odds; that -10 decibans means 1:10 odds; and that fractional numbers of decibans have similar meaning.
Does it make sense to talk about "i decibans", or "10 + 20i decibans"? If so, what does that actually mean?
Tiling Agents for Self-Modifying AI (OPFAI #2)
An early draft of publication #2 in the Open Problems in Friendly AI series is now available: Tiling Agents for Self-Modifying AI, and the Lobian Obstacle. ~20,000 words, aimed at mathematicians or the highly mathematically literate. The research reported on was conducted by Yudkowsky and Herreshoff, substantially refined at the November 2012 MIRI Workshop with Mihaly Barasz and Paul Christiano, and refined further at the April 2013 MIRI Workshop.
Abstract:
We model self-modication in AI by introducing 'tiling' agents whose decision systems will approve the construction of highly similar agents, creating a repeating pattern (including similarity of the offspring's goals). Constructing a formalism in the most straightforward way produces a Godelian difficulty, the Lobian obstacle. By technical methods we demonstrate the possibility of avoiding this obstacle, but the underlying puzzles of rational coherence are thus only partially addressed. We extend the formalism to partially unknown deterministic environments, and show a very crude extension to probabilistic environments and expected utility; but the problem of finding a fundamental decision criterion for self-modifying probabilistic agents remains open.
Commenting here is the preferred venue for discussion of the paper. This is an early draft and has not been reviewed, so it may contain mathematical errors, and reporting of these will be much appreciated.
The overall agenda of the paper is introduce the conceptual notion of a self-reproducing decision pattern which includes reproduction of the goal or utility function, by exposing a particular possible problem with a tiling logical decision pattern and coming up with some partial technical solutions. This then makes it conceptually much clearer to point out the even deeper problems with "We can't yet describe a probabilistic way to do this because of non-monotonicity" and "We don't have a good bounded way to do this because maximization is impossible, satisficing is too weak and Schmidhuber's swapping criterion is underspecified." The paper uses first-order logic (FOL) because FOL has a lot of useful standard machinery for reflection which we can then invoke; in real life, FOL is of course a poor representational fit to most real-world environments outside a human-constructed computer chip with thermodynamically expensive crisp variable states.
As further background, the idea that something-like-proof might be relevant to Friendly AI is not about achieving some chimera of absolute safety-feeling, but rather about the idea that the total probability of catastrophic failure should not have a significant conditionally independent component on each self-modification, and that self-modification will (at least in initial stages) take place within the highly deterministic environment of a computer chip. This means that statistical testing methods (e.g. an evolutionary algorithm's evaluation of average fitness on a set of test problems) are not suitable for self-modifications which can potentially induce catastrophic failure (e.g. of parts of code that can affect the representation or interpretation of the goals). Mathematical proofs have the property that they are as strong as their axioms and have no significant conditionally independent per-step failure probability if their axioms are semantically true, which suggests that something like mathematical reasoning may be appropriate for certain particular types of self-modification during some developmental stages.
Thus the content of the paper is very far off from how a realistic AI would work, but conversely, if you can't even answer the kinds of simple problems posed within the paper (both those we partially solve and those we only pose) then you must be very far off from being able to build a stable self-modifying AI. Being able to say how to build a theoretical device that would play perfect chess given infinite computing power, is very far off from the ability to build Deep Blue. However, if you can't even say how to play perfect chess given infinite computing power, you are confused about the rules of the chess or the structure of chess-playing computation in a way that would make it entirely hopeless for you to figure out how to build a bounded chess-player. Thus "In real life we're always bounded" is no excuse for not being able to solve the much simpler unbounded form of the problem, and being able to describe the infinite chess-player would be substantial and useful conceptual progress compared to not being able to do that. We can't be absolutely certain that an analogous situation holds between solving the challenges posed in the paper, and realistic self-modifying AIs with stable goal systems, but every line of investigation has to start somewhere.
Parts of the paper will be easier to understand if you've read Highly Advanced Epistemology 101 For Beginners including the parts on correspondence theories of truth (relevant to section 6) and model-theoretic semantics of logic (relevant to 3, 4, and 6), and there are footnotes intended to make the paper somewhat more accessible than usual, but the paper is still essentially aimed at mathematically sophisticated readers.
...so did we now get cold fusion to work or what?
Some of you may have heard about the following paper already:
http://arxiv.org/ftp/arxiv/papers/1305/1305.3913.pdf
Here's a news article wrapping up the main points:
http://atom-ecology.russgeorge.net/2013/05/20/an-italian-cold-fusion-tide-lifts-all-boats-arvix-independent-review-paper-confirms-rossi-fusion/
I'm way out of my depth here so I find it hard to judge, is this a pile of BS or are we finally getting somewhere for real?
Is burning coal (and using chemical reactions in general) for the purpose of producing energy coming to an end in the upcoming decades?
EDIT: Here's a review of the article, it should be read. http://scienceblogs.com/startswithabang/2013/05/21/the-e-cat-is-back-and-people-are-still-falling-for-it/
Privileging the Question
Related to: Privileging the Hypothesis
Remember the exercises in critical reading you did in school, where you had to look at a piece of writing and step back and ask whether the author was telling the whole truth? If you really want to be a critical reader, it turns out you have to step back one step further, and ask not just whether the author is telling the truth, but why he's writing about this subject at all.
-- Paul Graham
There's an old saying in the public opinion business: we can't tell people what to think, but we can tell them what to think about.
-- Doug Henwood
Many philosophers—particularly amateur philosophers, and ancient philosophers—share a dangerous instinct: If you give them a question, they try to answer it.
Here are some political questions that seem to commonly get discussed in US media: should gay marriage be legal? Should Congress pass stricter gun control laws? Should immigration policy be tightened or relaxed?
These are all examples of what I'll call privileged questions (if there's an existing term for this, let me know): questions that someone has unjustifiably brought to your attention in the same way that a privileged hypothesis unjustifiably gets brought to your attention. The questions above are probably not the most important questions we could be answering right now, even in politics (I'd guess that the economy is more important). Outside of politics, many LWers probably think "what can we do about existential risks?" is one of the most important questions to answer, or possibly "how do we optimize charity?"
Why has the media privileged these questions? I'd guess that the media is incentivized to ask whatever questions will get them the most views. That's a very different goal from asking the most important questions, and is one reason to stop paying attention to the media.
The problem with privileged questions is that you only have so much attention to spare. Attention paid to a question that has been privileged funges against attention you could be paying to better questions. Even worse, it may not feel from the inside like anything is wrong: you can apply all of the epistemic rationality in the world to answering a question like "should Congress pass stricter gun control laws?" and never once ask yourself where that question came from and whether there are better questions you could be answering instead.
I suspect this is a problem in academia too. Richard Hamming once gave a talk in which he related the following story:
Over on the other side of the dining hall was a chemistry table. I had worked with one of the fellows, Dave McCall; furthermore he was courting our secretary at the time. I went over and said, "Do you mind if I join you?" They can't say no, so I started eating with them for a while. And I started asking, "What are the important problems of your field?" And after a week or so, "What important problems are you working on?" And after some more time I came in one day and said, "If what you are doing is not important, and if you don't think it is going to lead to something important, why are you at Bell Labs working on it?" I wasn't welcomed after that; I had to find somebody else to eat with!
Academics answer questions that have been privileged in various ways: perhaps the questions their advisor was interested in, or the questions they'll most easily be able to publish papers on. Neither of these are necessarily well-correlated with the most important questions.
So far I've found one tool that helps combat the worst privileged questions, which is to ask the following counter-question:
What do I plan on doing with an answer to this question?
With the worst privileged questions I frequently find that the answer is "nothing," sometimes with the follow-up answer "signaling?" That's a bad sign. (Edit: but "nothing" is different from "I'm just curious," say in the context of an interesting mathematical or scientific question that isn't motivated by a practical concern. Intellectual curiosity can be a useful heuristic.)
(I've also found the above counter-question generally useful for dealing with questions. For example, it's one way to notice when a question should be dissolved, and asked of someone else it's one way to help both of you clarify what they actually want to know.)
Recent updates to gwern.net (2012-2013)
Previous: Recent updates to gwern.net (2011)
“But where shall wisdom be found? / And where is the place of understanding? / Man knoweth not the price thereof; neither is it found in the land of the living…for the price of wisdom is above rubies.”
As before, here is material I’ve worked on in the 477 days since my last update which LWers may find interesting. In roughly chronological & topical order, here are the major additions to gwern.net:
- I interviewed translator Michael House about his work in Japan as a translator
- finished data collection for my hafu anime statistics page and begun analysis. (I’ve achieved good coverage of characters, found an astonishingly consistent absence of Korean characters, and confirmed the blond-haired/blue-eyed stereotype; but my original thesis doesn’t seem to work and the data is too unevenly distributed to identify time trends.)
- judged the 2011 & 2012 results for the Haskell Summer of Codes and the accuracy of my predictions
- did a meta-analysis on whether dual n-back increases IQ, and examining possible biases and various claims about what makes the training work or not work
- did another meta-analysis on whether iodine increases IQ, etc
-
modafinil:
- checked for subjective effects of blinded modafinil
- updated my modafinil price-chart twice, and expanded with brand data and a new armodafinil table
- researched modafinil-related prosecutions & convictions in the USA
- and any connection with schizophrenia
- tried kratom
- did a nicotine gum/n-back experiment
- did 2 potassium experiments; neither improved my mood/productivity, and one damaged my sleep
- my Silk Road page has been expanded with a BBC interview, putting SR in a historical cypherpunk context, an updated account of all arrests & law enforcement actions, and application of basic statistics to ordering
- ran 2 sleep experiments on the timing of taking a vitamin D supplement: I found that taking vitamin D before bed substantially damaged my sleep, while taking vitamin D after waking up did not hurt & somewhat helped
- checked whether a walking desk (treadmill) damaged typing speed or accuracy
- I have run 3 Wikipedia experiments establishing that: Talk page edits are ignored by editors; random link deletions (and their restoration) are also ignored by editors; and external link suggestions on Talk pages are also ignored by readers. (I take the former 2 as indicative of the decline in edit activity and rise of deletionist beliefs on Wikipedia.)
- tried some economic/historical analysis: “Reasons of State: Why Didn’t Denmark Sell Greenland to the USA?”
- Defending sunk costs essay (LW discussion)
- “Slowing Moore’s Law: Why You Might Want To and How You Would Do It”
- “The Hyperbolic Time Chamber as Brain Emulation Analogy”
- tried estimating the bandwidth of a Death Note
- compiled predictions for Harry Potter and the Methods of Rationality
- looked into Conscientiousness and online education; studies so far are useless from a meta-analytic standpoint
- tripled length of appendix dealing with the reliability of mainstream science (methodological flaws, replication rates, etc)
- finished meta-ethics essay, “The Narrowing Circle”
- explained the philosophy saying “one man’s modus ponens is another man’s modus tollens”
- speculation about a restoration of the British monarchy
- clean up & exploratory data analysis of SDr’s lucid dreaming data
- Who wrote the Death Note script? (LW discussion)
- 2012 US election predictions: statistical comparison
- Turing-completeness in surprising places (inventory of particularly “weird machines”; relevant to computer and AI security)
Transcribed or translated:
- Nash’s letters on cryptography
- Douglas Hofstadter’s superrationality columns (from Metamagical Themas, 1985)
- “The Iron Law Of Evaluation And Other Metallic Rules”, Rossi 1987 (lessons from the large RCTs evaluating social & welfare interventions)
- “The Ups and Downs of the Hope Function In a Fruitless Search”, Falk et al 1994
- Gene Wolfe on writing
- “Shiny balls of Mud: William Gibson Looks at Japanese Pursuits of Perfection” (2002)
- “Otaku Talk”, Okada et al 2004
- “Earth in My Window”, Murakami 2005
- “On The Battlefield of ‘Superflat’”
- “Ero-Anime: Manga Comes Alive”, Sarrazin 2010
- 1996 NewType interview with Hideaki Anno (translated by me, with the help of an EGFer)
- 1997 Animeland interview with Hideaki Anno (bought, transcribed, and translated by me with the help of other LWers)
- 1997 Utena interviews
More technical:
- added edit history statistics/visualization for
gwern.netusing GitStats - site traffic updates: July-December 2011, January 2012-July 2012, July 2012-Jan 2013
- There’s also been a lot of backend changes: switching to Amazon S3+Cloudflare, adding error pages, metadata like tags, A/B testing, but no need to go into detail.
Personal:
- dumped my notes on my 2011 visit to San Francisco
- posted summaries of my personality & attitudes & my RSS feed collection
- enjoyed some mead; I still like tea better, though
- dumped notes on the 2012 SF convention ICON
Subscribe to RSS Feed
= f037147d6e6c911a85753b9abdedda8d)