(Talk given at an event on Sunday 16th of August. habryka is responsible for the talk, Jacob Lagerros and Justis Mills edited the transcript.
If you're a curated author and interested in giving a 5-min talk, which will then be transcribed and edited, sign up here.)
(This was all very off-the-cuff about stuff that I am very confused about, so take all of this with a grain of salt. There are at least a few things in here I disagree with myself, that I wouldn't have put into written form directly.)
habryka: I’m going to talk about three frames I have involving relationships and sociology. I’ll present the frames, some short justifications for why I believe them, and discuss how they connect to each other.
1. I think most relationships go better if you lean into conflict.
2. Most conflicts are hierarchically embedded within different rules, and maintaining the integrity of those rules is quite important.
3. Professionalism is really interesting. I like thinking about it, and I've gotten a bunch of value from thinking about it, because I didn't realize how much of my life has been shaped by professionalism.
Leaning into conflict
One of the things that has been pretty useful for me in life, is a general heuristic of realizing that conflict in relationships is usually net positive. (It depends a bit on the exact type of conflict, but works as a very rough heuristic.) I find it pretty valuable too, if I'm in a relationship, whether it's a working relationship, a romantic relationship, or a friendship, to pay a good amount of attention to where conflicts could happen in that relationship. And generally, I choose to steer towards those conflicts, to talk about them and seize them as substantial opportunities.
I think there are two reasons for this.
First, if startups should fail fast, so should relationships. The number of people you could have relationships with is much greater than the number of people that you will have relationships with. So there is a selection problem here, and in order to get as much data as you want, I think going through relationships quickly and figuring out whether they will break or not is quite valuable.
Second, I've found that having past successful conflicts in a relationship is a very strong predictor for that relationship going well more generally, and for my ability to commit to the relationship and get things done within it. In fact, I find it a better predictor of my capacity to coordinate with that person than the length of the relationship, the degree to which we even enjoy spending time with each other, or other obvious indicators. I have some models of why successful conflict is such a good predictor, which brings me to my second frame.
Honoring the rules of engagement
I have this frame of thinking about rules of conflict and hierarchically embedded conflict in a lot of different situations. The basic situation that happens relatively frequently is that I have been interfacing publicly with some organization or person on the internet, and they take some small action that hurts me in some way. They might write an angry comment, for instance, or they might try to insult me.
A thing that I find useful to think about with these conflicts is: do we have any rules of engagement that we could rely on? That me and whoever I'm in conflict with, that we could together coordinate on, in order to prevent that conflict from spreading over into other parts of our life? With online commenting, for example, there's a generally widely accepted rule that we will not let our conflict leak into other platforms. If we're in a pseudonymous context, you won’t seek me out in a non-pseudonymous context and try to continue the conflict there.
Similarly, let's say LessWrong is in conflict with some other organization. There is a general expectation that if we have some kind of debate on the internet and there's a bunch of conflict happening there, we won't take that conflict to a broader venue like Twitter and then spark an angry mob to attack the other organization. The reason for that is pretty straightforward: the collateral damage of high levels of conflict tends to be quite large. I've often, in situations of conflict, got a lot of value out of asking myself, "What are the lowest collateral damage rules that I could embed and act on here in order to make this conflict go well?"
This leads me to frame three.
I argue that this frame is the culmination of everything I’ve said thus far.
I’d like to talk about professionalism as a culture. Professionalism requires navigating conflict in lots of situations successfully, and it often encourages having important conflicts early on. Furthermore, it’s a really important component of people's lives.
I've found it useful to think about professionalism, or “presenting consistent APIs to other people”, as a safeguard against serious conflict. These APIs enable very narrow interfaces, which also enables very narrow potential for collateral damage. If I try to get a plumber, it's pretty easy. I don't have to think exactly about whether the plumber will like me or not, or other broader concerns about them as a person. Instead, I can pick a plumber based only on their fulfilling that specific role.
And also, another key component that I find really interesting to think about is the degree to which professional culture is optimized for replacing any individual within a given system. So that, for example, if you do professional writing, I would say that one key component is the ability to switch out writers in the middle of the writing process, which is false for 99% of other situations involving writing.
More broadly, I find professionalism is a modern religion whose details and culture it can be quite valuable to analyze.
jacobjacob: Regarding the first point: Nassim Taleb, who sure likes conflict, has a metaphor of forest fires. The basic idea is that if you prevent the small fires that come up naturally, it leads to an accumulation of excess flammable material, such that the next big fire is exceptionally bad. Does this resonate with your model?
habryka: Sorry, my first reaction was just that Taleb is my key example of failing at point two, and escalating way too much. There's this thing where you disagree with Taleb on some topic and suddenly he calls you deeply disingenuous and sends an internet mob after you. That was my first reaction -- but that is mostly about Taleb as a person.
I do think that the idea of not having conflict explode violently later on is pretty important here, but that actually most of the time we manage to avoid such explosions, and that’s interesting. Humans are pretty good at predicting when big conflict will happen and actually, most of the time, we manage to prevent it. I think it’s true that small fires help guard against big ones, but I also think that big fires are rare in any case because people are good at predicting and averting them.
mingyuan: I have previously heard you talk about professionalism in a more negative sense, and I'm wondering if you've changed your perspective over the past couple months?
habryka: My current relationship to professionalism is very similar to my relationship with Christianity, which is that it’s really damn important to understand how Christianity has influenced the West, because it's everywhere. There are obviously central concerns about Christianity. The dominant one being that it is horribly wrong. But my current relationship to professionalism is: "Oh god, it is doing massive amounts of damage in many different contexts." It's also quite useful in other contexts, in the same way that Christianity is.
mingyuan: So you're not a practicing professional?
habryka: I am in some dimensions. I definitely recognize that in certain situations it's really important to present a consistent API that's consistent with professional norms. I also try to avoid situations where that's truly necessary, because the constraints that come from professionalism are quite substantial.
Ozzie: For relationships, I was curious if you’ve thought about “anti-dates” or purposely difficult situations that you could put yourself in, to generate helpful conflict. Like taking care of a few kids that are difficult for a day or organizing a friend's wedding. Things that maybe should be common practice for people dating each other.
habryka: I mean, I don't know. The first thing that came to mind, I think it was someone in our community who wrote the “questions to ask if you wanted to end your relationship”.
Maybe it's a good idea for everyone to ask themselves those questions early on. It’s well understood that a lot of typical relationship advice is actually about vulnerability, about finding places where there might be a potential source of conflict, and stuff like that, so I think all this is relatively entangled here.
It’d be hard to think of specific tasks that would expose potential conflicts across all relationships, though, since the situations that produce conflict are very different per relationship. If I want to work with someone at an organization, the dimensions in which I need to understand how future conflict will go differs a lot from if I want to have a Dungeons & Dragons group with someone. So it's very hard for me to come up with one thing that you would want to do in both, but it doesn't feel very hard to come up with stuff I'd want to do in each individual situation to stress-test.
jacobjacob: I'm pretty confused about the claim that professionalism does conflict fast. My impression is that many professional environments are incredibly risk averse, and they’re characterized by things like “putting your emotions to the side and just showing up to do the work”. This is an important norm for some situations, but it doesn't enable conflict.
The outlier environment that most have embraced conflict seems to be Bridgewater, a large hedge fund centered around something like “having the most crucial conversations as early as possible”. (If people know what a Hamming Circle, or even a Doom Circle, is, my impression is that at Bridgewater they basically do that all the time.)
So could you say a little more about professionalism’s relationship to conflict?
habryka: I think the key thing here is to distinguish professional norms within and between organizations. Professionalism has very different rules for these two cases. One of the core components of professionalism between organizations is contractualism. All parties agree on a contract, and negotiations on that contract frontload a bunch of conflict. All parties figure out what they will do in the relevant different scenarios very early on. Compared to other norms, contractualism is really costly. If I hang out with a friend and we already have an existing trust relationship we usually don't need a contract, because we trust that we will later figure out how to negotiate if something weird happens or one of us falls through on an implied obligation.
In professional relationships, if you were to rely only on norms and ad hoc negotiations, it would often end very badly. So what we do is put a lot of the negotiation and conflict very early on, where we sign a contract and the contract is very well specified. All parties negotiate the rules of the contract, bring in some lawyers, and they figure out all the ways in which the contract could go wrong. Then, once negotiations are done, the contract codifies a working relationship. Without that frontloaded conflict, a professional relationship would go very badly. Contractualism is really powerful here.
Anonymous: Where do professional hierarchies start and end? Does professionalism have some sort of consistent variation with power or scale? And is there a predictable way it changes as you go up a power hierarchy?
habryka: Let me restate the question to make sure I get it: “Let's say we have conflicts that are hierarchically embedded. What does that hierarchy actually look like?”
The top of the hierarchy would be something like outright war. You’re in conflict with each other and you're both just out to kill each other. And then, even above that, you're out to blackmail each other using threats to directly harm the other person's values. In the superintelligence example, this could look like torturing trillions and trillions of simulated beings that you really like. On the very lower level, it's very minor conflict, like we're playing Scrabble and/or we are playing a video game and killing each other's units. Even though there's some kind of damage we're causing each other, it's an extremely limited type of damage.
I feel like there’s a component of professionalism here that boils down to a bunch of very well-negotiated rules of conflict. These rules work pretty well when you're talking about tens of thousands to millions of dollars. I think it starts working substantially less well when you start talking about billions and trillions of dollars. Wars are not fought, for instance, within the professional culture, nor are other things with stakes in the trillions of dollars. There are still rules of conflict in these cases, but they are fought with different sets of norms.