Category Archives: Almost

What you can’t say to a sympathetic ear

Suppose we live in a society where it is strongly frowned upon to believe that an onion is a fruit. It is ok to disagree about what defines ‘fruit’, or what Allium varieties are onions. But none of this will get you off the hook—you had just better not suggest that an onions is a fruit.

You don’t think about the issue much yourself. If you had to, you would probably agree with the consensus view that the onion is not a fruit, given a few clarifications of the question. If you were allowed, you would probably admit that you don’t care much about the question. And that you would kind of prefer that it was possible to discuss the issue calmly and without accusations of transcendent evil.

However none of these things is relevant in the real world, because you daren’t even advocate for calmer consideration of the onion classification issue. People would infer that (in a sense) you don’t want to punish people who say that an onion is a fruit. And (in a different sense) not punishing it is much like endorsing it. Punishing non-punishers is an important part of cooperation.

Now suppose you and I are chatting over lunch in a work cafeteria, and I glance furtively around and then lean over to you with gleaming eyes, and whisper that I am making fruit soup tonight, and ahem, there are many people who would cry if they watched me cutting up the fruit for it.

You see what I’m saying. Nobody else seems to have heard.  Are you annoyed? Do you think worse of me?

My guess is yes, at least quite plausibly.

And you are not annoyed because you find my comments troubling in their own right. You disagree with them, but don’t find them intrinsically offensive. Outside the context of our society, you wouldn’t mind.

You might be offended that I am willing to suggest an onion is a fruit in your presence in spite of knowing that most people would be unhappy about this. But suppose that we know each other well, and you know I know you are hard to hurt, even with grievous categorization errors.

I think you still have a strong reason to be annoyed. Which is that I am intentionally taking an action that the rest of the world thinks you are strongly obliged to punish—for instance, by threatening to stop associating with me unless I have an amazing excuse for what kind of seizure took over my mouth. Which means you must decide on the spot whether to punish me (at a cost to our relationship) or implicitly collude a bit with my renegade controversial-thing-saying faction. At a cost to your relationship with the world, because if they learned of this, they would hate you.

This makes my implied classification of onions as fruit into an ultimatum: ‘Me or the rest of society?’ If it is intentional, then it is a test of our friendship, at your expense. It’s like randomly saying ‘ok, if you really care about our friendship then steal $10 from your grandmother to prove it’.

Saying that onions are fruit quietly to you is holding our friendship hostage unless you shift your alliances away from the rest of the world and toward me. Or, more likely, it is an accident that still puts you in this position.

And it is very annoying to have your valuables taken hostage, and even more annoying to be threatened on short notice, with a deadline, so that you can’t just put it at the bottom of your to-do list and deal with it another time.

I hadn’t explicitly noticed that this kind of dynamic existed before (and it may not), but I think it might play a large part in my own feelings, on both sides of situations that are a bit like this.

I am sometimes annoyed when people reveal disagreeable views to me, even if I don’t especially disagree with them. Which is a bit surprising, on the face of it. And other times, I find myself in the position of wanting to say things that may sound controversial, and feeling hesitant, in part for the other person’s sake. So I got to thinking about the possible ways that could harm someone. And imagining myself in their shoes, this is the kind of harm I expected. I have not much idea if others feel the same way in these circumstances, or would construe the situation similarly in terms of game theory.

This might all seem pretty unimportant, being as it is a speculative and hand-wavey analysis of an already obscure social situation. But the existence of multiple reasons to be offended by officially offensive statements—even if you are sympathetic to them—means that social bans on views should be more stable than you might think. It’s one of those things where even if everyone comes to privately believe that onions are indeed fruit, and also thinks that nobody should be punished for saying this, and they can all talk to each other, everyone might still end up saying that onions aren’t fruit forever.

This means sanctions on speech aren’t just costly because they make it hard for individuals to hear ideas that might turn out to be true. They are more costly than that, because even if every individual manages to hear the ideas, and they are good, still they might not be able to update their behavior or the social consensus. And if everyone has to talk and behave as if a claim is false, we have lost a lot of the value of knowing that it is true.

To successfully condemn a view socially is to lock that view in place with a coordination problem. We could all freely identify onions as we wanted, but we have to all at once decide to change the norms, or else we get punished. And changing the norms would be hard to arrange at the best of times, but is harder when trying to arrange it warrants punishment.

If this analysis is correct, I think this situation should raise the bar for condemning views, because it makes it even harder for future people to undo our mistakes where we have erred. Condemnation is more permanent.

ETA Sept 3 2017: I am on reflection happy for people to tell me their controversial views if they are interesting—I bring up my slight feeling of annoyance about it as evidence that it is imposing some cost. But I am usually willing to pay the cost.

Research immersiveness

Suppose you want someone to calculate the current price of cheap computing hardware.

Alice knows roughly how much computing hardware cost a year ago, and know what the different kinds of computing hardware are, and how prices vary by time and market, and as she does the calculation she thinks things like ‘ooh, that is cheap’.

Bob doesn’t know anything about hardware and also isn’t trying to think about it. He mechanically follows an abstract research process he devised without reference to the subject matter, and makes no attempt to form an intuitive model. He reasons that in order to find the price of X, it will be necessary to locate the cheapest readily available version of X. He searches in various ways for cheap X. He writes down the best X prices he can find. He doesn’t notice if the prices are similar, or if there is a pattern behind the different X and prices associated with them. He determines the lowest price on his list, and returns it as the cheapest X. He writes words like ‘hardware’ instead of X, but it doesn’t really matter for his process.

I expect Alice to produce better research. Here are some advantages I expect her to have over Bob:

  • She will know to look for cheap prices for common kinds of computers such as supercomputers, CPUs, and GPUs.
  • If her searches only return CPUs, she will notice that maybe supercomputers aren’t coming up for some reason other than their not being cheap, and look them up directly.
  • If Alice finds numbers that are very different from the number a year ago, she will be surprised and check more carefully.
  • Alice will have an idea of whether the prices of GPUs she has found so far are relatively representative cheap GPUs, based on  things like how much they vary, what brands they are from.
  • If any numbers or patterns of numbers are surprising or interesting, Alice will notice and be able to do further research on a relevant sub-question, such as ‘is Moore’s law speeding up for computing cost?’
  • She will remember much more of what she has found so far, because it is meaningful to her.
  • Basically, Alice is getting feedback, while Bob may as well have set up the whole research procedure to run mechanically at the start, because he is not getting any information from what he finds. And often feedback is good.

You can do research without engaging with the topic, carrying out abstract calculations while ignoring the numbers you find, not updating your views, or feeling curiosity about the subject; feeling nothing. Or you can know what you are studying like you know your lover’s face: understand what each number means, and see what it implies for everything else. Usually you will do something in the middle of this spectrum, but it is good to consider what the ends are like. This is an axis of research I often think about, because sometimes I look more like Alice and sometimes I look more like Bob. I’ll tentatively call Alice’s research style ‘high immersement’, and Bob’s ‘low immersement’. I think of Bob’s as ‘Chinese room style research’.

I have assumed that high immersement is just overwhelmingly better, perhaps because I have some unreasonable bias in favor of knowing what one is doing. Or because of the list of reasons above, and ones like them. (Or because it is more impressive?)

But I expect one advantage of low immersement is that it should be less biased. Bob basically doesn’t know what his data means, so it is harder for him to push the results one way or another. He is like a blinded researcher, and it is often considered highly valuable to have everyone involved in research be blinded. And because Bob doesn’t notice errors, he avoids the kind of problem where he mostly corrects errors that disagree with his own intuitions, biasing the findings toward his own intuitions. I can’t think of obvious reasons he would be more biased in a particular direction. Though maybe in some topics, looking into a thing in more detail reliably leads to higher or lower numbers, and Bob might look into things in less detail. But overall Bob’s research seems likely to be more wrong and less biased.

Most research probably happens somewhere in the middle of the spectrum, and maybe there are non-linear effects to having some idea of what you are doing, on the potential for bias. Even so, I think I expect less bias from people who have less idea what they are doing.

My guess is still that high immersement research is generally much better, but to the extent you can control it, I wonder if it is worth being in low immersement mode to do some research tasks. In particular, tasks where you don’t really want your own interpretation of things to influence your behavior. For instance, in the survey I’ve been working on lately it might have been good to be in high immersement mode for looking into how to ask questions, and for interviewing people to inform question writing, and low immersement mode for turning the results into graphs and statistics, and then high immersement mode for looking at the graphs and statistics and speculating about what they mean.

Is science the worst field for learning science in?

I wrote that science classes usually demonstrate the art of not significantly changing your beliefs when evidence conflicts with an answer you are sure of, at the expense of the central idea of science: experimenting for the purpose of changing your mind.

Chris Chang asked me how it could be taught better:

I can imagine the occasional chemistry assignment that gets around this by having mystery reagants whose identities you need to discover, and the occasional physics assignment where you have anonymous materials and need to figure out some of their properties. But it seems hard to me to extend this to an entire course worth of labs; any more ideas?

I doubt it’s hard to keep children in the dark about the entities in a whole course’s worth of labs, but it may conflict with wanting to teach them more reliably discovered information on the same topics, or at least not wanting to mislead them. Fortunately though, science is applicable to any subject matter within the realm of reality. The topics on which we have collected a lot of knowledge via science are also called ‘science’, which makes it easy to forget they are not the only possible, or even necessarily the best, subjects of science.

We want to teach our children some of this knowledge we have gained from science. But there seems little reason to combine understanding what we already know, about say chemistry, with learning about the process of doing science. If you want to learn to do science, with all the thrills of actually discovering anything, you are probably best to pick an area where people don’t already know all of the cheap answers (those not requiring monkeys, brain scanners or large chunks of subterranean Europe).

The most obvious place to find undiscovered cheap answers available for scientific investigation is in topics that are not very important to people who pay for science. A notable example is the many small scale questions that are relevant to a given student but not to society as a whole. Does decreasing the length of my skirt increase the propensity of the cool students to talk to me? Does learning the piano as a child really make people happier later in life? Does Father Christmas exist? Do the other children hate me or are they just indifferent? What factors best cause my brothers to leave me alone? How much do my grades change if I do half an hour more or less homework each night? Does eating sugar all evening really keep me awake? How often will I really be approached by potential kidnappers if I hang out at the mall by myself after school?

You can probably think of better questions. As far as I know most children and teenagers disagree with their parents, teachers and other adults on a large number of issues. Investigating those issues scientifically might have the added benefit of getting students in the habit of keeping their opinions related to reality. Another plus may be engaging the students’ interest without having to explode progressively larger things.

Science fair projects and the like seem to move in this direction, though many seem to manage the ‘not especially interesting to the rest of the world’ criterion without managing the ‘especially interesting to the student’ one, beyond the student’s inherent interest in doing a project at all.

Roger Shank proposes something similar. He thinks people don’t consider their own activities as experimentation, and so miss valuable information from them:

If school taught basic cognitive concepts such as experimentation in the context of everyday experience, and taught people how to carefully conduct experiments in their own lives instead of concentrating on using algebra as a way of teaching people how to reason, then people would be much more effective at thinking about politics, child raising, personal relationships, business, and every other aspect of daily life.

I don’t know if this is true, but it’s probably worth testing.