A Glossary of Accidental Social Defections

We all remember the core conceit of the Prisoner's Dilemma: defecting is advantageous for the individual, but cooperating is advantageous for the collective. In iterated games, strategies that incentivize the maximum amount of cooperation through a mixture of punishments and rewards are the ones that win. Since you make everyone's lives better by cooperating, I wouldn't hesitate to say you need to be someone who cooperates to be a good person.

When we describe someone as "well-adjusted", I think what we're actually saying is they default to cooperating in social situations. Cooperation is the lubrication that conversation and all social interaction glide upon. Nevertheless, the allure of the defection is there. Remember, defecting against a cooperator puts you ahead. According to the Gervais Principle (which I do not wholeheartedly endorse, nevertheless I find it a useful framework and vocabulary) the cooperators are "Losers" and the defectors are "Sociopaths", and the Sociopath becomes the CEO. You don't want to be the Loser, right?

The problem with accidentally defecting is that you manage to gain all of the advantages without the disadvantage of a guilty conscience. An accidental defection is the most addictive strategy of all. Since most people you interact with will likely be well-adjusted Losers that bias to cooperation, you can kinda steamroll them without ever being made aware that you've done something wrong in the first place. This is really going to bite you when you run into a Sociopath and they observe you defecting, because you will have marked yourself as a threat. It will also bite you every time you defect against a cooperator because they're going to want to stop cooperating.

Worry not, little ones. I am here with a handy reference for some of the ways you may accidentally be running roughshod over your conversational partners so you can wake up to your defections and become a virtuous Loser like the rest of us.

Prelude: The Yapper's Dilemma

Let's just quickly reiterate the game theory of a social interaction. Two people begin a conversation, let's call them Angela and Bianca. And let's say Angela wants to accomplish objective A and Bianca wants to accomplish objective B, and both objectives are achievable within the bounds of this conversation. For example, Angela and Bianca are discussing a job offer than Angela received, and Bianca's objective is to get Angela to accept it because she believes that is the best course of action, while Angela's objective is to get advice and insight on the job offer. They also carry with them an uncountable number of hidden secondary objectives that are modeled more or less by Maslow's hierarchy of needs: they both want prestige, power, safety, resources. All of these things are possible to influence within the bounds of a simple conversation, but especially prestige. You can gain or lose face simply from how you conduct yourself during the exchange.

There are a variety of cooperative strategies Bianca can deploy to achieve her objective. She can list her reasoning in a measured manner, ask follow up questions, make Angela feel heard, and ultimately respect her decision. But there are also a host of defective strategies she may use to accomplish her goal faster or more decisively, maybe without even noticing that's she using them.

Let's explore these social defections and think about how we may be falling into the same patterns.

Defections

1. Overconfidence

The standardized tests you take in high school are sometimes kind enough to let you know how you stack up against other test-takers, to give you an idea how to interpret your scores. I always enjoyed picking up my colorful little sheet of statistics that told me I was 99th or 98th percentile in every category. I hear a lot of people my age complaining that being a "gifted" child ruined their life, though sometimes it can sound like a humblebrag instead of anything actionable. While I have heard some reasonable-sounding rationale for why academic excellence early in life leads to failure and burnout later on, the symptom I most frequently encounter is one I've never heard discussed elsewhere, probably because it is uncomfortable to think about. As a recognized smartypants you can usually get away with a lot of intellectual shortcuts and your peers will defer to you. You can exaggerate the depths of your own knowledge and almost nobody will call you out on it because they will simply assume you're more informed on the topic. Sometimes you are, but not always.

This behavior already crosses the line into a defection (pushing around the epistemically humble despite having worse rigor) but where this will really hurt you is past high school. Something funny happens between high school and college: one third of the students don't move on to higher education, and another full third end up in community colleges, for-profit scams, technical colleges, and "party schools". If you're gifted and don't burn out in high school then you're likely going to find yourself suddenly surrounded by an already highly selected group of peers and your place in the intellectual pecking order is going to take a dive. Post college, you're going to be rubbing elbows with people who have been in your industry for decades and you're going to be even more of a bottom-feeder. Being 99th percentile really stops feeling like being 99th percentile once you start spending all your time around other 99th percentilers.

I think early adulthood is really rough for a lot of kids because it subverts our narrative expectations for life. It feels natural that as we grow older and gain skills and experience, our standard of living and social status should only ever increase. In reality, there is a sharp dive in standard of living once we reach adulthood and stop taking freely of our parents' wealth. In college and beginning a career, our relative status also decreases sharply (even if the intellectual distribution were to stay completely static, just going from high school senior -> college freshman and college senior -> junior hire are total hierarchical resets).

It is essential one gets out of the habit of confidently stating their beliefs as facts. Beyond your intellectual betters calling you out and embarrassing you in front of an audience, an even worse outcome is your betters might believe you. They'll just assume in good faith that you have done your homework and allow themself to be misled. If you end up being wrong and you—for instance—steer a business in the wrong direction, this is all going to eventually land on your shoulders and you're going to burn all of your credibility in one incredible conflagration.

The most trustworthy people constantly express uncertainty. Just like other cooperative strategies, this opens you up to be taken advantage by defection as your conversational partner can seize on that uncertainty to dismiss your concern/claim with their own overconfident assertion. Hopefully this will drive you toward more virtuous pursuits like actual research where you compile sources and weigh options and do all the things that would make an investigation into a hard problem productive in the first place.

2. Isolated Demands for Rigor

Scott Alexander explains it better than I. Many of the conversations that touch upon the philosophical and the empirical will stop before they reach their most aggressive conclusion, but if you are desperate for a win it is possible to go nuclear. In the case of the philosophical/ethical, you can interrogate someone's system of belief for an example of inconsistency (which you will almost certainly find, for who among us is perfectly egosyntonic?) and then claim victory. This strikes deep at their sense of identity and blows up your relationship for a cheap and often irrelevant win. In the case of the empirical, you can just smirk and say "source?" and sit back while they have to grovel and proffer you what evidence they can scrounge together, often without you ever having to provide any competing sources of your own. It's aggravating to see this strategy deployed with such great effect in online conversations, but it's a far worse trick to pull out in face-to-face conversation because it's demeaning to have to pull out your phone and scroll through a bunch of search engine results just to be allowed to continue participating in the conversation.

There are good reasons to probe at the frayed edges of why people believe things, but there are much more cooperative ways of going about it. I really enjoy the Socratic method, where you help your interlocutor build up and flesh out their argument through only asking questions, and cooperatively discover the holes in it. You can only really effectively deploy this technique if you stop viewing your conversation partner as an opponent and stop viewing your arguments as soldiers.

3. Teasing

I grew up in a family where we teased each other a lot. I got teased most of all, being the youngest child, something that continues to this day. Through much of my teens and young adulthood I have defended teasing as mostly harmless banter. I think in extremely high-trust groups it can serve as an equalizing measure that regularly humbles everyone involved (you will see this effect play out a lot in groups of young men, where the jokes at someone's expense can get pretty wild). But it has some pretty terrible failure modes. If any shadow of a social hierarchy (and the vast majority of social settings already come with a pre-existing social hierarchy: school, church, politics, the workplace, a party) ever manages to take root, teasing turns into a sort of class warfare where the high-status punch down and the low-status feel justified in getting increasingly personal and genuinely nasty with their targets when punching up. Teasing is never 100% ironic, no matter what the teaser wants you to think. It's a low-stakes way to deliver gentle criticism and the target ends up looking insecure if they let people know their feelings are hurt. Thus, it is a defection.

The solution is easy: just say nice things about people instead of mean things. If you identify behaviors or beliefs that you seriously think need addressing, there's almost always a more productive and peaceable way of addressing it than roasting and potentially humiliating them (like writing a 3000 word essay that they will never read).

Nevertheless, I still feel conflicted about this one. I do think teasing works to keep people self-aware, and it can feel overly solemn and hostile to address group members' suboptimal behavior through mini struggle sessions instead. The social hierarchy can be an unexpected tool to help everyone save face: a manager can deliver serious feedback much more effectively than a peer. It would definitely be inappropriate for the manager to do so by teasing their report, especially in front of an audience. It seems at the very least you should be overly conscious about where you sit in the hierarchy and make sure you never punch up or down with your teasing, if you really must continue to tease.

4. Psychoanalysis

Learning about developmental psychology, evolutionary psychology, sociology, or any other -ology that purports to contain explanatory power for human behavior can fry your brain. Once you build a framework to justify and understand the actions of others, the natural inclination is to begin sorting and tagging behaviors. This is fine to do in your own head, in fact I find it can really lessen the blow of the more hurtful social experiences you will endure by wrapping them in a comforting narrative (e.g. "she was just gossiping about me because she's competitively signaling her own worth to the group to ensure she's not at the bottom of the pecking order and liable to get expelled when there's a resource shortage, a fate that would spell death in the EEA").

But psychoanalyzing someone out loud, in front of others, is a major defection. Not only is it an ad hominem, but it is impossible to come to a rigorous consensus one way or the other if what you're saying accurately explains their behavior or beliefs because you cannot know their thoughts. The standard way people do deep dives about themselves—professional therapy—usually takes years of self-reflection and cooperative frame-building. It's not a productive topic to dive into in the middle of a disagreement. Seriously, steer clear of "(You/they) (did/believe) X because of Y" statements in public settings. It immediately puts your target on the back foot because these sort of claims can be extremely persuasive to an audience due to how easy it is to produce anecdata justifying your new narrative.

My least favorite form of psychoanalysis is a claim about someone's emotional state. Something as simple as "you are mad". This is a kafka trap, there is no way to defend yourself against the accusation that you are mad without looking like you're mad, at least online where there aren't any body language hints to assist you. However valuable or relevant you think the psychoanalysis is to the conversation, understand that you will only serve to derail it and place yourself in a position of power over your target, so don't do it.

5. Misunderstanding

One of Amazon's principles of leadership sometimes makes me laugh:

This feels like the equivalent of gamers telling each other to "get good". Just be right. Have you ever just tried being right? You should try it.

But I will say something very similar: misunderstanding someone's point is actually a form of defection. You may respond that it is the speaker's responsibility to make sure they are understood correctly, but this doesn't tell the whole story. Conversation, as a cooperative endeavor, entails heavy-lifting on both the part of the speaker and the listener to make sure information is communicated accurately and efficiently. When you're dealing with a bad faith listener you will constantly have to stop to clarify points and address objections and it gets exhausting. It is not enough to simply not be bad faith, as a listener you need to be good faith and do the intellectual labor to understand what the speaker really meant rather than merely what they literally said.

There are two straightforward tools in your arsenal for this: steelmanning and an Ideological Turing Test (ITT). Steelmanning exists as the inverse of strawmanning: rather than responding to the weakest form of a speaker's argument that you can imagine, respond to the strongest form you can imagine. This may entail altering it to a form that the speaker doesn't entirely endorse, so even better is the Ideological Turing Test. In the original Turing Test, a computer attempts to be indistinguishable from a human in conversation. In an ITT, you attempt to be indistinguishable from somebody the speaker agrees with. Rather than taking the form of an actual test or deception, this usually boils down to repeating the speaker's claims back in your own words, ensuring you understand them in a way the speaker endorses and finds charitable before proceeding to respond to the argument.

If you identify a critical and obvious flaw in somebody's argument, do them a favor and stop a moment to consider that maybe they're not a dummy and it is you that is missing something. It only requires one person for a misunderstanding to take place, and that person is the listener, not the speaker.

So just, stop misunderstanding people bro. Get good.

6. Applause Lights

Eliezer explains it better than I. There are substance-free statements that almost nobody will disagree with, that function like an "applause light": hey everyone, look at how reasonable I am! Please clap. Quoth Eliezer:

In a speech, applause lights are a waste of breath and brainpower. In a conversation or a debate, they are much more nefarious. Saying them aloud implies that they needed to be said. Not only do they sound like an argument in favor of your policy proposal, but they sound like an argument against another policy proposal. They are neither of those things, they are just an applause light. A "+0" tacked on to the end of an equation. The motte in the motte-and-bailey. Respect everyone's time and intellect and stay away from applause lights.

7. Fault-finding

When you first start learning about logical fallacies, the most important one to learn so that all your knowledge of fallacies doesn't serve to actively harm you is the fallacy fallacy:

Like applause lights, fallacies are not a deathblow to an argument but merely a "+0", neither supporting nor detracting from the conclusion. Pointing them out merely serves to remove pieces from the table and simplify the argument, not checkmate the king. But take caution! pointing out logical fallacies can seem really aggressive, unless you have strong social norms around naming-and-shaming fallacies to save time. This is a form of fault-finding that I think one should be very careful with, but it at least serves a utility even if it can embarrass the fallacious arguer.

There is a family of fallacies that I think are even more pernicious than the fallacy fallacy: the nirvana fallacy. In essence, the nirvana fallacy involves arguing against actionable prescriptions by pointing out drawbacks without establishing a case for why the drawbacks actually exceed the benefits (or even acknowledging the benefits at all). For example:

The reason this crosses the line from a mere fallacy to an actual social defection is twofold. First: it's annoying. One theme you might notice that runs throughout all of these defections is they often force your victim to go out of their way to explain something to you that really should be obvious: that you are hurting their feelings, that you are misunderstanding their argument, that you are pointing out flaws that don't serve to kill their conclusion. Rather than losing you status for defecting, the defection tends to harm the victim because it disrupts the flow of their argument and makes them look weaker, if only illogically so. Second: just like applause lights, fault-finding casts you in the role of the wise moderate and the victim as the dangerous radical. It's a dark art that allows you to farm credibility without engaging substantively with the proposal. Should you be successful in your defense of the status quo you will be ensuring that nothing gets worse, but nothing gets better.

There is a superior way to go about this: point the cons toward the pros and show how they cancel each other out until you feel you have demonstrated that there are more cons on the board than pros. Even just acknowledging the existence of upsides to a proposal you disagree with is already a huge step forward to breaking out of the arguments-as-soldiers mindset and turning a contentious debate into a collaborative exploration.

8. Aggressive Coupling and Defensive Decoupling

As I said in 5. Misunderstanding, a vast sea of shared context underlies each interaction. If it were not so, you would have to stop to explain every term, cultural reference, relevant current event, etc. and nothing would ever get done. We join context onto conversation in a process called Coupling that embeds meanings in people's statements that they don't need to say explicitly. The issue arises from the fact that we don't all share the same sea of context, and we don't all swim so deeply in it. Missed connections and incorrect assumptions can arise from mismatched couplings, leaving conversational partners defensive and irritable.

Here's what this often looks like in practice: Angela wants to establish some positive ground truth, perhaps to build toward a greater point or perhaps simply because she finds the ground truth valuable by itself. Then Bianca draws upon her personal background or knowledge of broader societal realities and claims that Angela is saying the truth, yes, but pointed in a problematic direction.

An example would be if Angela cited American crime rates broken down by race. America has a lot of history and contemporary context around the subject of race. Angela's audience is going to—understandably—join that context with the positive claim and wonder why she cited those statistics. Because, you know, sometimes people cite those statistics to justify racism(!) and sometimes people cite those statistics as evidence to prove that systemic racism is occurring, and sometimes people cite those statistics for innumerable other reasons, like determining resource distribution or health outcomes or something else you can't even think of.

Both Angela and Bianca have an opportunity to defect here. Bianca can defect through Aggressive Coupling: accusing Angela of impure motives, claiming she has chosen to highlight these facts for hateful reasons. Angela can defect through Defensive Decoupling: denying that societal and historical context should play a role in considering the ground truth. Should Bianca commit Aggressive Coupling, she will have lobbed an extremely serious accusation at someone that she better be 1000% sure of because she stands to do extreme damage to Angela's reputation. Should Angela commit Defensive Decoupling, she is simply denying reality. Wail about dispassionate facts and logic as you might, you cannot stop your audience from joining context onto content, and it's irresponsible to expect that to ever happen. It's Angela's responsibility to ensure that the appropriate context is accessed for the point she is trying to make, and that she is fastidiously avoiding dogwhistles or any other form of disingenuity.

Postlude: The Yapper's Paradise

A cooperative conversator is more pleasant to be around. They are more liable to be described as "charismatic", "frank", "intimate", and other kind words.

But we mustn't forget that there are two players in any Prisoner's Dilemma, and cooperating opens yourself up for defection. Almost all of these defections only work if your conversational partner is already cooperating, so you may seize the advantage and push them around. By stopping your defections and raising the collective bar, you are sacrificing your personal social safety. Alas there is no solution to this, for all intimacy is ultimately vulnerability. If someone really is misbehaving socially, while you may suffer temporary humiliation at their hands, staying cooperative allows you to access a new set of interventions by virtue of your innocence: third parties may jump in to defend you, managers or guardians may discipline their wards, and you will always eventually escape without the blemish of having stooped to their level. I guess I should end by quoting the ultimate authority on this and every other subject. Matthew 5:43-47: