Multidimensional signaling

What do you infer about a person who has ugly clothing? Probably that they have poor taste (in clothes or subcultures). But it could also be that they are too poor to improve their wardrobe. Or can’t be bothered.

What about someone with poor grades? The obvious inference is that they aren’t so capable at the subject, but it may again be that they can’t be bothered, or that they have more urgent things to do with their time.

And someone who makes clever jokes? Probably that they are smart and naturally funny, but if they had more time or effort to spend on this, it probably helped.

For all kinds of traits that people might try to signal with their behavior, someone can send a better signal if they have more money or time or self-control. Even when the main signal being sent is not usually thought to be about any of those things.

The reason that this interests me: if signals often divide the population into ‘better or richer’ vs. ‘worse or poorer’, I wonder if this would cause us to imagine that being rich is associated with being better, even if the two were entirely independent. (And similarly for wealth in other general-use resources, like self-control and time).

In a simple case, suppose there are just people with pretty clothes (who have good taste, and also the wealth and industriousness to show it) and people with ugly clothes (who either have bad taste, or lack resources or will). Then do observers come to think of ‘rich, go-getter good taste’ type and a ‘poor, lazy, bad taste’ type? Or do they pay more attention to the actual structure of the space, and know for instance that learning that someone really has bad taste does not actually means they are more likely to be lazy or poor?

New Doc 2017-10-15 (1)

A simple example. X marks everyone who don’t signal.

Note that I’m not merely suggesting that a person with more wealth can send signals to look like they are better—that much is clear. I’m suggesting that at a population level, if the wealthier people can’t be distinguished from the better people on some axis, then observers may come to think that the two are associated in general, even if they are not at all.

If so, this would be important, because it would apply in a huge range of cases of signaling. So that the properties of poverty and weak-willedness and such would appear to us to be much worse than they really were.

Signal seeding

What does it say about a person if they never get up before noon?

If they are the first person to exist, it probably says that morning was for some reason a convenient time to sleep.

If they live in parts of modern society, it might say that they are lazy and weak willed.

How did getting up at one time or another come to signal laziness? You still have to get up once per day.

One story I can imagine is that originally there was some weaker reason to get up early. For instance if your work benefited by sunlight, you could get a bit more in. And then since that was a reasonable thing to do, people who didn’t do it looked like they were less good at getting up. Which made getting up early even better thing to do, so that everyone knows that you can.

And then people who had been on the fence before about whether to bother getting up early started to find it worth their while.  Making the remaining noon-sleepers even more weak willed on average. And so it continues, until sleeping until the afternoon strongly suggests laziness.

In general, if an action is a tiny bit good, not doing it can look a tiny bit bad (or stupid, or lazy, or incapable). Which makes it better to do, which makes it look worse to not do it, and so on. And maybe in the end the speck of good that started this disappears, but the value of sending the signal if you can is enough that the equilibrium is stable.

Does this actually happen?

Prosocial manipulation

There is an axis of social calculativeness: whether your speech and social actions were carefully designed for particular outcomes, versus being instinctive responses to the situation.

This is related to an axis of honesty: whether your words represent your actual state. I suppose because the words most likely to produce the best response naively are often not true. Though I’m not sure if this is reliably true: feelings in the moment are often misleading, and honesty is often prudent.

Another axis is selfishness versus pro-socialness: whether your actions are meant to produce good outcomes for you (potentially at the expense of others) or a larger group such as the world.

The calculativeness axis seems widely expected match the selfishness axis well. Manipulative people are bad. I don’t see why they should go together though, in theory. You can say what you feel like in conversation, or say things calculated to achieve goals. Shouldn’t people saying things to achieve goals do so for all kinds of goals, many venerable? In about the same distribution as people doing other things to achieve goals?

A natural question is whether calculated behavior really is reliably selfish, or whether people just feel like it is for some reason. I can think of cases where it isn’t selfish. For instance, a diplomat trying to arrange peace is probably choosing their words very carefully, and with regard to consequences. But it is hard to say how rare those are.

Perhaps we just don’t think of that as being calculative? Or I wonder if we do, and while we like it if peace is arranged, we would still be somewhat wary of a very good diplomat in our own dealings with them. Because even if they are acting for the good of the world, we suspect that it won’t be for our good, if we are the one being calculated about.

After all, we are presumably being led away from whatever our default choice would have been after hearing the person just represent their internal state as came naturally. And moving away from that sounds probably worse, so more likely that manipulation means to exploit us somehow than to secretly help us get an even better outcome. This is closely related to the honesty axis, and would mean ‘manipulative’ doesn’t really imply ‘globally consequentially bad’ so much as ‘dangerous to deal with’.

I am speculating. Are there common positive connotation terms for ‘socially manipulative’ or ‘calculating’? Is that a thing people do?

For signaling? (Part I)


Your T-shirt is embarrassing. Have you considered wearing a less embarrassing T-shirt?

You are suggesting I spend my precious time trying to look good. Well I am good, and so I’m not going to do that. Because signaling is bad. You can tell something is bad when the whole point of it is to have costs. Signaling is showing off. Signaling benefits me at someone else’s equal expense. I won’t wear a less embarrassing T-shirt because to Hell with signaling.

Hmm. That seems wrong. Signaling is about honest communication when the stakes are high—which is often important! And just because it’s called ‘costly’ doesn’t mean it is meant to have costs. It only has to be too costly for liars, and if it’s working then they won’t be doing any signaling anyway. ‘Costly signals’ can be very cheap for those who use them. I think signaling is often wonderful for society.

Give me three examples where it is ‘wonderful’.

Driver’s licences. Showing a driver’s licence is a costly signal of being a decent driver, which communicates something useful honestly, is cheap for the people who are actually good drivers, and lets the rest of society distinguish people who are likely to drive safely from people who are not, which is amazingly great.

Driving tests don’t seem that cheap to me, but I’ll grant that they are probably worth it. Still, this seems like a strange corner case of ‘signaling’ that was explicitly designed by humans. It fits the economic definition of ‘costly signaling’ but if you have to go that far from the central examples to find something socially beneficial, that doesn’t increase my regard for signaling. Next?

One of the most famous examples of signaling is in the job market. Potential candidates show a hirer their qualifications, which allows the hirer to employ more appropriate candidates. You might disagree about whether all of the signals that people use are socially optimal—for instance if education is mostly for signaling, it seems fairly destructive, because it is so expensive. But you must agree that companies do a lot better hiring the people they choose than they would hiring random people they would get if good candidates couldn’t signal their quality. And at least many aspects of the interview process are cheap enough to be totally worth it. For instance, being able to have a polite and friendly conversation about the subject matter.

Of course companies are better off—companies aren’t the people destroying years of their productive lives on deliberately arduous fake work. Or learning a lot of irrelevant but testable skills. Or degrading themselves and society with faux friendliness. And you ignore some other key details, like what the actual alternative would realistically look like. But let’s not go into it—I’ll grant you that hiring probably goes better overall than it would with zero signaling and no replacement, even though the signaling is awful. And more importantly, that the the whole of society on net is probably best off with some kind of signaling there. I don’t know of a good replacement.

Ok, great. So, third—T-shirts. T-shirts signal personality traits. It is free to wear any T-shirt you want, but T-shirts are still costly signals in a sense, because if you aren’t a punk you won’t  know which T-shirt to wear to look like a genuine punk. And if you don’t like ABBA it is more costly for you to wear an ABBA t-shirt than it is for someone who does like them, because you’ll be embarrassed or unhappy at the association. And if you have bad taste, it is hard to know which T-shirt would indicate good taste. This all seems good, because it lets people cheaply find other people with similar interests, and also to learn facts about the people around them, regardless of similarity. Which is why it is socially destructive for you to wear that T-shirt— your taste can’t be that bad, so you are basically lying.

Ok, a fourth: how about when a friend is sick, and you make them tea and soup and put on a movie for them. This is a costly signal that you care about them, or at least  about your continuing friendship with them. Because it is effort for you with no reward if you don’t care much, and are looking to scale down the relationship soon. But aside from the signaling, this is probably a net social benefit—your friend gets soup and tea and a movie at a time when they could especially use them. Plus, feeling cared for instead of uncared for is a real benefit.

Ok, I concede that costly signaling can be honest, cheap, and on net socially beneficial. But I still think it usually isn’t! And I’m not sure how far we can get thinking about specific examples, since there are so many.

Ok, what do you propose?

Talking about our overall impressions. The big picture. Here is mine: the world is full of people pouring real wealth into things whose only use is to be rubbed in the face of those who can’t afford to destroy so much value. Where it isn’t even good for society to be able to distinguish the signalers from the rest.  Letting everyone see who is rich and who is poor, who is socially competent and who is not, who is beautiful, who is smart, who can win at things that only exist to be won at—does this really lead to a great world?

There is much signaling that the world would be better off without. I admit I don’t really know what the balance of good and bad is like. But I disagree that we should be talking about signaling overall. Or even what is best for the world in this particular case. You are not the world. Even signaling that is terrible for the world is often good for you. If you are in a zero-sum game, and you are more worthy than the opponent, then do your best to win! And if you aren’t, then be more worthy!

What if I want what is best for society?

Even then, you don’t serve society by failing at signaling. Just because people fighting to look good is costly for society doesn’t mean that society gains anything by you intentionally losing that fight. If you are directing your resources to society, then it is better for society if you win. Often better enough to warrant the costs of playing. Serve society by winning at signaling and donating the proceeds to society. Wear a well ironed suit. Don’t talk about your erotic porcelain dinosaur collection. Go to university. Try to exercise good taste…

I agree, at least often. But I think you believe in a heuristic that says you should signal about as much and in similar ways as if you were selfish. Because you are on the side of good, so protecting yourself is protecting the good. You see people looking weird and embarrassing themselves in the name of caring about something, and you think they are failing at signaling. And that’s wrong.

Yeah, I guess you should signal a tiny bit less on the margin, in cases where signaling is socially destructive. But it’s such a small thing, I’m not sure it is worth thinking about.

I don’t mean that. Your selfish interests can come apart from society’s interests almost entirely, in signaling. As an extreme case, imagine that you became confident that by far the best cause for improving the world was promoting incest. From a selfish perspective, you probably don’t want to look like you are promoting incest, because there are few worse ways to look in modern society. But from an altruistic perspective, supposing that you were right about incest, it may well be best for you to promote it, because it would do so much for making incest look better, at just the cost of your own reputation.

You should distinguish between wearing a clean shirt—good for your cause—and wearing a shirt that is more respectable because it is not about your cause—which is often bad for your cause. You can’t just use ‘looking good’ as a heuristic, even though it is generally good for your cause when its proponents look good.

That’s an interesting point, and I hadn’t really thought about it. But surely that’s pretty rare. There are systematic reasons that it’s unlikely that there is some cause which is radically more important than any other, and is completely politically unpalateable.

I agree that’s unlikely—just brought it up as a clear example of it being not worth looking good. I think this issue is maybe ubiquitous though, in less clear and extreme cases. For instance, everywhere sophisticated people play it cool, withholding enthusiasm from ideas until they no longer lack enthusiasm, polishing their own image at the expense of the very projects they are most excited about, or would be if they deigned to experience excitement.

A bold claim—I am curious to hear two more examples, but I have a lot of signaling to get done this evening. Same time next week?

Most likely. I hope you are correctly identified as the superior type in all of your endeavors.

Impression track records

It is good to separate impressions from beliefs.

It is good to keep track records.

Is it good to keep separate impression and belief track records?

My default guess would be ‘a bit, but probably too much effort, since we hardly manage to keep any track records.’

But it seems maybe more than a bit good, for these reasons:

  1. Having good first impressions, and being good at turning everyone’s impressions into a good overall judgment might be fairly different skills, so that some people are good at one and some are good at the other, and you get a clearer signal if you separate them.
  2. We probably by default mostly learn about beliefs and not impressions, because by assumption if I have both and they are different, I suspect the impression is wrong, and so will make me look worse if I advertise that I hold it.
  3. Impressions are probably better than beliefs to have track records for, because the point of the track records is to know how much to weight to give different sources when constructing beliefs, and it is more straightforward to know directly which sources are good than to know which aggregations of sources are good (especially if they are mostly bad, because nobody has track records).

As in, perhaps we mostly keep belief track records when we keep track records, but would do better with impression track records. What would we do if we wanted to keep impression track records instead? (Do we already?)