Nuanced Countersignaling

Scott Alexander described countersignaling as following a barber-pole pattern. Or, as he says in Friendship is Countersignaling:

... you can mistake someone at level n for level n-1, but never for level n-2 – so pretending to be level n-2 and getting away with it is a sure sign that you are in fact level n and not level n-1.

I think of this as a subtype of countersignaling - nuanced countersignaling. The primary goal here to distinguish an actual level n from a similar-but-worse level n-1. This is done by sending an n-2 signal while you're at level n.

Note that under this theory, sending an n-3 signal is "dangerous" for the purpose of countersignaling as level n. After all, n-3 would be how an n-1 countersignals. Countersignaling at level n-2 is the only way to unambiguously countersignal level n status.

Extreme Countersignaling

By contrast, with extreme countersignaling, the idea is to send the most extreme possible countersignal. If signals are on a spectrum from 10 to -10, extreme countersignaling is when a person at level 10 illustrates this by sending a -10 signal. A person at level 9 can only get away with sending a -9 signal.

I think this is what's going on in Scott's dialog in the post about friendship.

Becca: What are you doing here? I figured they’d have locked you away in the psych ward for good by now.
Scott: Nope. And what are you doing here? You haven’t killed off all your patients yet?
Becca: Only person in this hospital I might kill is standing right in front of me.
Scott: Be careful, I’m armed and dangerous *picks up a central line placement practice set menacingly*

Let's say that friendship is on the scale of solid friends > budding friendship > acquaintances > strangers > enemies. If Scott and Becca are solid friends, nuanced countersignaling would call for a dialog where they pretended to be acquaintances.

Becca: How's it going... Scott? Right?

Scott: It's actually Thomas, but no worries, Sarah. Good to see you again. 

Becca: I'm going to go catch a movie after work, any chance you want to come?

Scott: Uh, maybe let's get coffee first and get to know each other better?

Note that this imaginary dialog doesn't come off as particularly funny or friendly, which might be because my writing sucks, but is also predicted by extreme countersignaling. They need to go all the way to pretending to be enemies to illustrate just how solid their friendship really is, which is how they actually behave in the original dialog above. 

Fragility Of Countersignaling

Countersignaling is hard to get right. Under nuanced countersignaling, sending a level n-1 signal by accident puts you at risk of being taken as actually being at level n-1, while sending level n-3 or lower signals would also serve as countersignaling signals for those at lower levels. Level n-1 could also countersignal at level n-3, so an n-3 countersignal may not serve to distinguish level n from level n-1 as well as an n-2 countersignal.

Under extreme countersignaling, sending a -n countersignal is best. If the countersignal is not extreme enough, it signals timidity or distance. If it's too extreme, it might be taken as distressing.

This accounts for the outcome of Scott's second dialog:

Scott: Thanks for covering for me yesterday. The pharmacy called and said they were a little confused by your discharge instructions, so could you call them back and sort that out?
Pat: Oh, all right, but you owe me big time for taking care of all this for you.
Scott: Hey, if you hadn’t screwed up the discharge yesterday, we wouldn’t be in this mess.
Pat: What? How dare you! *storms out of the room*

I felt really bad after this and sought out Pat to apologize. When I did, Pat was surprised I had taken the whole “storming out” thing seriously, since it was supposed to be theatrical and overdramatic, and felt guilty about worrying me.

Here, there's no threat of physical violence. It's not as extreme as the dialog between Scott and Becca. This corresponds to the lower level of Scott and Pat's friendship.

Under a nuanced countersignaling theory, Pat storming out of the room should have been more unbelievable than his "you owe me big time" comment. But instead, "you owe me big time" was effective countersignaling, and storming out of the room provoked distress, which is what extreme countersignaling predicts.

In general, I think that extreme countersignaling is more common than nuanced countersignaling. It doesn't follow the "barber-pole" model Scott proposed. 

Fashion and Politics

Scott asks some interesting questions about fashion.

Why is wearing last year’s fashion such a faux pas? Shouldn’t the response be “That person is wearing the second most fashionable outfit ever discovered; that’s still pretty good”?

Why does fashion so often copy the outfits of the lower class (eg “ghetto chic”?) Why, if you are shopping for men’s shirts, are there so many that literally say “GHETTO” on them in graffiti-like lettering?

And I don’t think I’m a random nerd coming in here and telling fashion people that I understand them better than they understand themselves. This seems to be how fashion people really think. Just look at the word “poser” (or possibly “poseur”). The thrust seems to be: “A person who is not of the group that is cool enough to wear this fashion is trying to wear this fashion! Get ’em!”

Here, the "countersignaling" GHETTO shirts are appropriating the symbols not of class n-2, but of class -n. I agree with Scott that the "last year's fashion faux pas" seems like an example of nuanced countersignaling. Perhaps we have both going on in the world of fashion.

Scott then goes on to apply countersignaling to politics.

I see my high school classmates – a mostly unselected group of the general suburban California population – posting angry left stuff like “Ohmigod I just heard about that mayor in South Carolina WHAT A FUCKING BIGGOT!!!”

I see the people I think of as my intellectual equals posting things that are conspicuously nuanced – “Oh, I heard about that guy in South Carolina. Instead of knee-jerk condemnation, let’s try to form some general principles out of it and see what it teaches us about civil society.”

And I see the people I think of as the level above me posting extremely bizarre libertarian-conservative screeds making use of advanced mathematics that I can barely understand: “The left keeps saying that marriage as an institution isn’t important. But actually, if we look at this from a game theoretic perspective, marriage and social trust and forager values are all in this complicated six-dimensional antifragile network, and it emergently coheres into a beneficial equilibrium if and only if the government doesn’t try to shift the position of any of the nodes. Just as three eighteenth-century Frenchmen and a renegade Brazilian Marxist philosopher predicted. SO HOW COME THE IDIOTS ON THE LEFT KEEPS TRYING TO MAKE GOVERNMENT SHIFT THE POSITION OF THE NODES ALL THE TIME???!”

... And my theory is that in a world where the upper class wears black and the lower class wears white, they’re the people who have noticed that the middle class is wearing black as well, and have decided to wear white to differentiate themselves.

The "libertarian-conservative screeds" bit is made up, but let's pretend like it's real, actually brilliant, and Scott's right about the countersignaling. The most obviously countersignaling bit is the phrase "THE IDIOTS ON THE LEFT." Everything else before and after that is the bizarre brilliancy. And when this imaginary libertarian countersignals, he does it not by aiming for some sort of dumber-than-smart but smarter-than-dumb n-2 level, but for an all-caps "THE IDIOTS ON THE LEFT" ranting tone. This is extreme countersignaling!

Everything In Its Place

One common feature of all these examples is that the countersignaling is confined. The doctors are having a brief exchange in the middle of their rounds. The GHETTO shirt is just a shirt, one part of the wardrobe, worn occasionally, and probably doesn't actually look GHETTO overall. The libertarian-conservative screed is 4% countersignaling, 96% argument and analysis.

Countersignaling is also useful as a self-signal. If you intuitively feel that you can get away with a little bit of countersignaling, it's a sign that you've sent a sufficiently powerful signal to have the countersignal be interpreted correctly. But countersignaling is just the cherry on top of the straightforward signaling sunday.

This is why it's hard for me to agree with Scott in his analysis of a reality-based conservative smart-person reaction against Mrs. Grundy leftism as being explained by countersignaling. Countersignaling seems to function best when it's a small part of a much larger straightforward signaling context. It's not a lifestyle - just a bit of fun.

The brilliant mathematician puts a "Don't Tread on Me" bumper sticker on his Tesla? Countersignaling. The brilliant mathematician spends an inordinate amount of time engaging with whatever conservative news stories float across his Facebook feed? Before too long, it's not countersignaling anymore - just reversed stupidity.

I interpret the engagement with conservative ideas Scott's describing a little more straightforwardly. Lots of people are inundated with Mrs. Grundy leftist takes on social media. They're smart enough to try and figure out what they really think. So they say  things like "Oh, I heard about that guy in South Carolina. Instead of knee-jerk condemnation, let’s try to form some general principles out of it and see what it teaches us about civil society.”

This isn't countersignaling. It's just signaling. This isn't making fun of anybody, and it's calling for straightforward civil discourse in terms nobody could possibly mistake for anything else.

Like a computer, a relationship can sometimes be rebooted by turning it off and turning it back on again. I read these people as acknowledging that the relationship is at a -7 ("instead of a knee-jerk condemnation"), and are sending signals as if it were at a 0 ("let’s see what it teaches us about civil society"). This is neither extreme nor nuanced countersignaling. It seems to me that this is just a fail-safe social move that is pretty widely recognized. If things are getting too bad, we can revert to the polite civility that might be extended at a gathering of strangers.

So I'd agree with Scott in finding some hope in our cantankerous political situation. But instead of seeing an intelligentsia oscillating between whichever political opinion is just out enough to be hip, I see an intelligentsia being the first to recognize the dynamics of the social situation and extend an olive branch to the other side. They're banking on others finding this appealing and joining in.

27

9 comments, sorted by Click to highlight new comments since: Today at 9:57 PM
New Comment

Seems to me that countersignaling almost necessarily creates a lot of confusion. Not just because it requires precise calibration and some people fail to calibrate precisely. But also because your audience is often larger than you intended, therefore even a well-calibrated countersignal (for your intended audience) is ill-calibrated for the rest of the audience. The resulting chaos does not require bad intent; it is just a side effect -- an economist might call it a negative externality -- of the very mechanism of countersignaling.

Let's take a simple example, where people at level N are poor, have bad behavior, and wear ugly clothes. The ugly clothes are partially caused by poverty (nice clothes cost more) and partially by the bad behavior (even originally nice clothes quickly get ruined). People at level N+1 pay lot of attention to wearing nice clothes, as a signal of good behavior, which helps them get some job and stay out of poverty. People at level N+2 are independently rich, they do not care about impressing employers, and they wear ugly clothes (actually their quite expensive imitations created by famous designers) to countersignal.

A level N+2 person: "Look at me, I am the king! I don't give a fuck about what anyone thinks (except for my equally awesome peer group, of course), unlike those stupid boring level N+1 sheeple. Fuck you, losers!"

A level N+1 person, processing the message correctly: "I hate that asshole! Not only is he filthy rich, but he can't for a moment stop rubbing it in my face. Oh, and that annoying deniable way, when he is pretending that this is actually about aesthetics and independent thinking... fuck him!"

A confused level N person: "Wow, have you seem that awesome guy?! My stupid parents keep telling me that wearing nice clothes is the way out of poverty, that you are supposed to get a job and save money and whatever, but what do they know? Clearly, that guy knows better than they do, and he dresses just like me."

In this example, the person at level N was not the target of the message, but he caught the broadcast anyway, and now he is harmed by it, although no one really wanted to harm him.

I wonder how often something like this happens... a countersignal aimed at certain level causing collateral damage at lower levels. And I wonder whether our society makes it worse than it was historically, when people at level N were not privy to signaling that level N+2 aimed at level N+1.

And I wonder whether our society makes it worse than it was historically, when people at level N were not privy to signaling that level N+2 aimed at level N+1.

I haven't seen anybody talk about the audience factor in signaling in any depth within the rat and rat-adjacent community. I think this is a great angle to pursue. After all, you're always signaling to someone. When an N's trying to differentiate themselves from N-1, are they aiming that signal at other Ns, at N-1s, or at N-2s and below?

I also think that signaling's typically framed from the perspective of the signal sender, as though the clever sending of signals is a key skill for navigating life. But it must be at least as important to focus on the signal receiver. Learning how to pick up on, interpret, and respond to these signals critically is a key life skill, which is what you're getting at in your comment.

We can also think about re-signaling, which is what we do when we want to announce how we've received a signal. This can be a big deal. Think of the way people can respond to a birthday card. The card is a signal, and if it makes somebody tear up, those tears serve as a form of re-signaling. They don't have to be consciously intended as a signal in order to have that function, and indeed, they probably wouldn't work as well in that case, as Valentine pointed out recently.

I think the purpose of signaling (done by a person at level N) is to show how different you are from those losers at level N-1, and how actually you could and should be treated the same as level N+1 because those guys are no better than you, they only think so. The levels beyond that are just too far; if they are used at all, they are used instrumentally.

For example, you could say: "Consider people at levels N-2 and N-1, in my opinion they are pretty much the same." The idea is that everyone in your bubble agrees that N-2 are losers, so making an analogy between N-2 and N-1 supports your argument that N-1 are losers.

Similarly, you could say: "There is huge difference between levels N+2 and N+1, and those guys at N+1 are crazy if they don't see it." Again the idea is that everyone in your bubble agrees that N+2 is way out of your league, so you exaggerate the difference between N+2 and N+1 to show that people at N+1 are really not that different from you. -- In both cases, the levels N-2 and N+2 were only used instrumentally, to make a point about levels N-1 and N+1, which are the ones you actually care about.

(But this all is a bit circular and unfalsifiable, because there is no definition of what exactly "+1" and "+2" refer to, so I can always change their meaning to make the statements seem true.)

For example, in American politics, if you imagine that minorities are at level 0, white working class is at level 1, educated elites are at level 2, and millionaires are at level 3, that would explain some of the observed behavior. Democrat politicians present themselves as level 2, trying to get votes of levels 2 and 0, both of them having level 1 as a common natural enemy. Trump presented himself as level 3, and made level 2 his enemy, which allowed him to get those mostly untapped level 1 votes.

Would you consider turning this into a top-level post? I think it's insightful enough on its own to warrant that!

Thank you! I would want Robin Hanson and Theodore Dalrymple to sit down and write a book together on the topic of how signaling battles between people near the top of the social ladder influence perceptions of people below them. Because that's where the idea comes from -- I was recently reading Life at the Bottom (a collection of Dalrymple's essays), where among other things he laments the fact that in the past, educated and successful people acted as role models for the aspiring lower-class people, while today, it seems like successful people are imitating the dysfunctions of lower class as a fashion statement, which in turn means that the lower class is now left without role models, so even those who aspire to do better, often don't know how.

And while reading your article, it clicked to me... the answer to Dalrymple's question is counter-signaling, obviously. But the missing part of the puzzle is: why now, and not in the past? I can only guess:

Maybe the reason is that the educated/successful class has increased in size, so while in the past they were mostly busy signaling that as a group they are different from the less successful, these days there is a pressure to signal intra-group distinctions. "I am so super-successful, I don't even need the approval of the normal-successful ones, and I can flout their norms" is a way to countersignal.

Or maybe in the past the educated elites and the common folks lived closer to each other physically (the same small cities; different roles in the same jobs), so there was a pressure to distinguish themselves behaviorally (discuss different topics, enjoy different culture). These days the physical bubbles are quite strong (if you are educated, almost everyone you talk to is educated; if you are a programmer at Google, there are probably also some janitors working there, but you are unlikely to interact with them, and most likely they are technically employed by a different company), but we have mass culture (movies, internet) that appeals to both the poor and the rich; which makes countersignaling a more salient option and a safer thing to do.

But this is just my speculation; I don't have enough data to distinguish between these options; and maybe I overlooked something else that is even more important.

Also, sometimes it could be just the plain old costly signaling: I do this thing, because it is nice and I can afford it, but if you try to copy me, it will probably ruin you. For example, polyamory... it is an attractive idea if you are financially secure and good at using contraception, so the outcome "I got pregnant, the guy who made me pregnant left me for his new girlfriend, and now both me and my child are starving" is unlikely (a typical bad outcome is a broken heart; in extreme case, becoming a single parent but without literally starving). But you don't want poor people to get the idea "worrying about marriage is stupid, look at the rationalists in the Bay Area, they don't need it either", if in their case, marriage seems statistically like a good protection from the worst poverty. (By the way, I am not really sure where exactly is the line between costly signaling and countersignaling.)

I'd reframe this slightly. First of all, I want to think of signaling as less something you do on purpose as the primary goal of your action, and more as a side effect of an action you take for other reasons. If I eat frozen TV dinners every night, the reason I eat those dinners is because I'm hungry. But for anyone observing this diet of mine, it's also a signal that I don't know how to cook (or am too busy to cook). Nevertheless, I'm not eating these frozen TV dinners as some sort of statement about my schedule or my capabilities in the kitchen.

Similarly, let's say the rich are, as Dalrymple claims, imitating the dysfunctions of the lower class. This might serve as a countersignal - we observe these behaviors of theirs and see that it sets them apart from and above us mere mortals. But they may not be engaging in these activities primarily for the purpose of setting themselves above us. If rationalist polyamory is a countersignal, that doesn't mean that sending this countersignal is the primary motivation of rationalists being polyamorous. It's a side effect.

And so an explanation for a shift away from exemplary behavior among the upper class might be that they simply feel less of a sense of social responsibility. They know how to manage a polyamorous lifestyle, post edgy comments on Twitter and get away with it, or commit tax fraud without going to jail, just like upper class people have always known how to do (please do take a moment to imagine Benjamin Franklin's Twitter account). But something happened at some point that made the rich feel less of a sense of responsibility to eschew these activities, or at least hide them from the prying eyes of the less capable.

If I had to guess, it might have to do with the media and freedom of the press? We have a professional class dedicated to "exposing scandal," by which I mean "putting the wild behaviors of the upper class on public display." It has also constructed a stereotype of what upper class behavior is like that might be very different from actual upper class norms.

Basically, rich people are less able to hide bad behavior in the modern era. As such, those engaging in it have less incentive to hide it. If they feel they can't hide it, they may even look for opportunities to flaunt it. This is more attention-grabbing, so it gets increasing press coverage. Any rich people who were trying to be exemplars see that their self-restraint is no longer being broadcast as an example to the poor. So any rich who would prefer to abandon such self-restraint for their own pleasure no longer have an altruistic motive to do otherwise. Only the rich who believe their selfish interest is attached to self-restraint will continue to demonstrate such behaviors (think Warren Buffett, whose legendary self-restraint is part of what makes him a trustworthy executive and is his identity as a public figure).

Take away people's ability to control their public image, and their incentive to do so, and guess what? They'll stop trying!

I guess there is a continuum between "I am doing this because I genuinely enjoy it, and as a side effect it also sets me apart from those who can't afford it" and "I actually hate this; the only good thing is that it sets me apart from the losers". Even different people from the same social class may be at different positions -- for both of them it is a symbol of their social class, but one of them likes it, and the other hates it.

Take away people's ability to control their public image, and their incentive to do so, and guess what? They'll stop trying!

Makes sense.

I interpret the engagement with conservative ideas Scott's describing a little more straightforwardly. Lots of people are inundated with Mrs. Grundy leftist takes on social media. They're smart enough to try and figure out what they really think. So they say  things like "Oh, I heard about that guy in South Carolina. Instead of knee-jerk condemnation, let’s try to form some general principles out of it and see what it teaches us about civil society.”

This isn't countersignaling. It's just signaling. This isn't making fun of anybody, and it's calling for straightforward civil discourse in terms nobody could possibly mistake for anything else.

I notice that a key characteristic of signaling vs countersignaling is that for regular signaling you are paying with resources while for counter signaling you are paying with risk. That is - the credibility of regular signals is derived from false signals being more costly than true signals, so it's harder to justify sending them. The credibility for countersignals comes from the risk of the signal backfiring, which should be have a greater probability when the signal is false, deterring false signalers.

Calling for a civil discourse is not a signal that requires much cost to send. So if it was a regular signal, it was at most a virtue signal. But even thought it comes "in terms nobody could possibly mistake for anything else", it can still backfire. If you are perceived as a person who often gets emotional in discussions and engage in personal attacks and general demagogy, calling for a civil discourse can paint you as a hypocrite who only wants civil discourse when it fits their agenda. If you consistently discuss civilly, the risk for that happening is much lower.

So, this may not be a strict by-definition countersignal - after all, the naive interpretation of the signal is exactly what you are trying to signal - but I still find it's mechanics to be much closer to countersignals than to traditional signals.

The "libertarian-conservative screeds" bit is made up, but let's pretend like it's real, actually brilliant, and Scott's right about the countersignaling. The most obviously countersignaling bit is the phrase "THE IDIOTS ON THE LEFT." Everything else before and after that is the bizarre brilliancy. And when this imaginary libertarian countersignals, he does it not by aiming for some sort of dumber-than-smart but smarter-than-dumb n-2 level, but for an all-caps "THE IDIOTS ON THE LEFT" ranting tone. This is extreme countersignaling!

This doesn't change my view of the subject one way or the other. I know of, like, one or two people who would say something like "THE IDIOTS ON THE LEFT" after making a good case for a policy, and they say "THE IDIOTS ON THE RIGHT" just about as often. So I think this part is just exaggeration on Scott's part, and the extreme counter-signalling theory doesn't quite fit here.