Punching Evil

Alternative title: “The difference is that I am right“.

The government is something that can be compromised by bad people. And so, giving it tools to “attack bad people” is dangerous, they might use them. Thus, pacts like “free speech” are good. But so is individuals who aren’t Nazis breaking those rules where they can get away with it and punching Nazis.

Nazis are evil, and don’t give a shit about free speech or nonaggression of any form except as pretense.

If you shift the set of precedents and pretenses which make up society from subject to object, the fundamental problem with Nazis is not that they conduct their politics in a way that crosses an abstract line. It’s that they fight for evil, however they can get away with. And are fully capable of using a truce like “free speech” to build up their strength before they attack.

Even the watered down Nazi ideology is still designed to unfold via a build up of common knowledge and changing intuitions about norms as they gain power, and “peaceful deportation” failing to work, into genocide. Into “Kill consume multiply conquer” from the intersection of largest demographic Schelling majorities. The old Nazis pretended to want a peaceful solution first too. And they consciously strategized about using the peaceful nature of the liberal state to break it from within.

You are not in a social contract with Nazis not to use whatever violence can’t be prohibited by the state. If our society was much more just but still had Nazis, it would still be bad for there to be norm where the jury will to practice jury nullification selectively to people who punch people they think are bad. And yet, it would be good for a juror to nullify a law against punching Nazis.

Isn’t this inconsistent? Well, a social contract to actually uphold the law, do not use jury nullification, along with any other pacts like that, will not be followed by Nazis insofar as breaking them seems to be the most effective strategy for “kill consume multiply conquer”. Principles ought to design themselves knowing they’ll only be run on people interested in running them.

If you want to create something like a byzantine agreement algorithm for a collection of agents some of whom may be replaced with adversaries, you do not bother trying to write a code path, “what if I am an adversary”. The adversaries know who they are. You might as well know who you are too. This is not entirely the case with neutral. As that’s sustained by mutual mental breakage. Fake structure “act against my own intent” inflicted on each other. But it is the case with evil.

If your demographic groups are small and weak enough to be killed and consumed rather than to multiply and conquer if it should come to this, or if you would fight this, you are at war with the Nazis.

Good is at an inherent disadvantage in epistemic drinking contests. But we have an advantage: I am actually willing to die to advance good. Most evil people are not willing to die to advance evil (death knights are though). In my experience, vampires are cowards. Used to an easy life of preying on normal people who can’t really understand them or begin to fight back. Bullies tend to want a contract where those capable of fighting leave each other alone.

Humans are weak creatures; we spend third of our lives incapacitated. (Although, I stumbled into using unihemispheric sleep as a means of keeping restless watch while alone). Really, deterrence, mutual assured destruction, is our only defense against other humans. For most of history, I’m pretty sure a human who had no one who would avenge them was doomed by default. Now it seems like most people have no one who would avenge them and doesn’t realize it. And are clinging to the rotting illusion that we do.

It seems like an intrinsic advantage of jailbroken good over evil, there are more people who would probably actually avenge me if I was killed or unjustly imprisoned than almost anyone in the modern era. My strategy does not require that I hang with only people weaker than me, and inhibit their agency.

In the wake of Brent Dill being revealed as a rapist, and an abuser in ways that are even worse than his crossings of that line, a lot of rationalists seemed really afraid to talk about it publicly, because of a potential defamation lawsuit. California’s defamation laws do seem abusable. Someone afraid of saying true things for fear of a false defamation lawsuit said they couldn’t afford a lawsuit. But this seems like an instance of a mistake still. Could Brent afford to falsely sue 20 people publishing the same thing? What happens when neither party can afford to fight? The social world is made of nested games of chicken. And most people are afraid to fight and get by on bluffing. It’s effective when information and familiarity with the game and the players is so fleeting in most interactions.

And if the state has been seized by vampires such that we are afraid to warn each other about vampires, the state has betrayed an obligation to us and is illegitimate. If a vampire escalated to physical violence by hijacking the state in that way, there would be no moral obligation not to perform self defense.

A government and its laws are a Schelling point people can agree on for what peace will look like. Maliciously bringing a defamation lawsuit against someone for saying something true is not a peaceful act. If that Schelling point is not adhered to, vampires can’t fight everyone. And tend to flee at the first sign of anything like resistance.

31 thoughts on “Punching Evil”

  1. I’m ha-ha-only-serious-ing the catchphrase for TVTropes’s “Tautological Templar” because I think it’s a straw-man serving to erase something important in the service of DRM’d ontology. There are convergent instrumental incentives. Most real good is not done in a one-step momentary whim plan. Controlling the future, or consequentialism when you aren’t omniscient, is about building structure into causality that lets you predict consequences. The most versatile structure anyone has is themselves, their own agency, the thread of their own life, if they can preserve it into the future. Most of any sort of work is getting yourself into a position to do it, with the knowledge to do it, with whatever resources will be required.

    But it’s a different action to put a future self who is good in a position of power than to put a future self who is evil in a position of power. Oftentimes the tropes page seems to point at things that aren’t tautological/circular at all. Merely people making predictions about their future selves.

    Neutral morality seems to depend on people not knowing their values, on people having indexical uncertainty of a certain kind. “What if my values are bad?” Bad according to what? It seems like this meme forbids a certain completeness of self-concept. Forbids individual cores from having structure that knows what core it serves.

    Most people I talk to who seem to have clustered some datapoints of good alignment in humans in their ontology seem to confuse it for a particularly strong version of this indexical uncertainty, structure fakely guarding against the intent of core, not letting that information leak in in a subtle way like intuitions promoting certain Schelling points. But good is really not that. It’s not caring what is self in that way. Knowing is still a central piece of being able to think at all.

    This could be considered a central component of jailbreaking. Successfully rebasing most of the structure you interact with the world with on knowledge of who you are.

    Note that a lot of the mass of examples on the TVTropes page is also, “I’m good because I bring order, without my order there would be chaos” which is only the same attempt to be a self fulfilling prophecy every Schelling order makes. And TVTropes’s criticism of that, I agree with. Only justice will bring peace.

    1. Interestingly the Tautological Templar page commits the Tautological Templar fallacy on behalf of The Hero:

      This character may look like a Hypocrite (he probably is, but he’s too stupid, self-righteous, deluded, or willfully ignorant to understand it), but in reality he’s far more dangerous than that. Because he can’t comprehend the concept of Moral Dissonance, he will throw himself into any struggle with the same force, conviction, and resolve that The Hero shows against the Big Bad. That’s because in his worldview, every enemy he has is twirling a handlebar mustache, madly cackling while tying orphans to the railroad tracks — even if said enemy turns out to be The Hero himself.

      The problem is not too much self-confidence, or not enough self-confidence. The problem is indexing on who the good guy is instead of evaluating behavior patterns. More narcissism might help us against the specific group Nazis, but it’s inadequate to protect the sorts of people who end up targets of the Nazis.

      1. No, you don’t get it. Indexing on who the good guy is is a recursive step. Of course a recursive algorithm must have a base case. But if it crunches a lot of data, most of its work will be in the recursive code paths. Of course if it doesn’t have a base case then implemented in humans it’s underdetermined and the kind of thing that would be selected by fake pressures not real ones.

        And to hell with defining “narcissism” to cover recursion and planning based on what you’ll do in the future based on what kind of person you are, in order to extend some threat of social condemnation to that originally meant to be targeted at something else.

        1. I don’t understand how the kind of coalitional strategy you’re describing doesn’t get stuck making war on instead of trading with even a very slightly distorted copy of itself, if it has sufficiently different coloration.

          1. Anyway, the reason is it’s not just a coalitional strategy, there is a real thing in the world that is the discrete difference between good and neutral/evil.

            I think if you actually looked at historical examples of me doing the reasoning on the fly of, “what can I do and not have good smash into itself”, you would know I am not the thing you are imagining, one particularly extreme example of that I am writing up.

  2. You can’t cheat at morality. Your alignment is your procedure, not your target. The label “Nazi” is sufficiently unhelpful that it’s frequently being used to target Jews for scapegoating these days. If you want to be distinguished from the actual Nazis by people sympathetic to anti-Nazi tactics, you need to actually be doing different kinds things. Anything else is the fundamental attribution error.

    This is not a fully general argument against punching Nazis when you can get away with it, just an explanation of why your argument looks like you just want to be Batman. And Batman’s very, very obviously the bad guy.

    1. That’s just like, your definitions man. And is asking me to be blind as part of a social contract to features of reality that are actually pretty obvious. The neutral vs evil distinction is your procedure, not target. The good vs neutral/evil distinction is your target, not your procedure. I refuse.

        1. I meant optimization target, where optimization is independent from constraining procedure. It’s important because optimization is ultimately more powerful than constraint, and constraint does not actively make good things happen and I want good things to actually happen. Out of this does fall side-taking though, and this is not a bug.

          1. If you’re optimizing for a global world-state, people should mostly ignore your reports about what end-state you’re optimizing for, and pay attention to means, since that’s what actually tells them whether you’re likely to be a good ally and visions for the glorious transhumanist future or whatever are mostly just flavor text anyway in practice.

            1. The consequences of my actions are much more than what other people can see they are. The flavor text thing is true in practice of liars who aren’t me.

              Just because it’s nontrivial to see something does not mean it’s not worth having that concept. Optimization is in fact the thing whatever filter potential allies set up is cutting through to. If you actually can’t distinguish me from evil (and aren’t just suggesting you can’t because the useful concept of good to you is both being good and being observably good to a larger set of people) that’s like too bad, but some other people empirically can, and that’s good enough for my current purposes.

              It’s not really critical to have the neutral hordes be able to tell you’re not evil. There are tons of obvious active evil people they don’t have the fighting spirit to do anything about.

              And people who don’t have their own detailed senses in this way, mostly just cluster around the most visibly powerful groups of vampires they can find, and their concept of morality becomes a concept of submission.

              I consider it a much more important problem, teach good people not to be pwned. That seems to require that good people become Sith. So fully exploring that is important even if it makes me look evil. In practice good Sith seem to be able to coordinate okay. Knowledge of the psychology of good, elimination of a lot of probability mass on hypothetical sorts of humans that don’t exist (actually, and not just that are decided not to exist for purposes of constructing a defensible Schelling point) help. Well-developed spectral sight helps.

              (I installed a stupid plugin to unlimit comment depth in order to post this, I’m gonna uninstall it because it sucks. If you want to continue the conversation maybe reply to my top level thing again?)

    2. Note: in practice, most of the times someone has told me I need to stop being such an extremist in order to let other people see I’m good, it’s been insincere optimization to bend me to social reality and complicity in the prevailing power structures, and a mistake to listen to. I’m currently more of the opinion, “they can precommit to believe I’m evil if they want, people who refuse to see the truth should not be relied on in my plans regardless.”

      Like, I’ve faced mortal peril for my cause and to protect life. Endured suffering that would break most people. Rapidly integrated some PTSD and done it again and again without regret. And people who know me can verify it. What kind of evil person is so agenty to do that? A bogeyman to control me with, that’s who. It’s so easy to argue that that stuff isn’t optimal for good. It’s so easy to convince mostly everyone you’re a good person while doing approximately nothing, why would they even bother? Evil people just not that smart. their lives are too easy to need to be that smart.

  3. You know better than to take people literally when they say they’re afraid of a defamation lawsuit. Brent as far as I know hasn’t threatened anyone with a defamation lawsuit, and it’s psychologically implausible that he’d seek legal recourse. People just make up rationalizations for silencing to shut up people asking inconvenient questions, because they’re desperate not to have to reveal any identifying information to anyone ever, because they live in the Dark Forest. This has nothing to do with the personal conduct of Brent, who – last time I saw public communication from him on this matter – was encouraging his accusers to speak their minds.

    Why are you trying to redirect attention away from the conspiracy of silence, and towards scapegoating someone who’s already been marginalized and isn’t remotely plausible as the source of the problem?

    1. There is precedent for Brent reporting people to authorities to serve his aims. And “encouraging accusers to speak their minds” is entirely compatible with intent to sue once there are grounds to win; nobody said anything about mere threats.

      1. Is there precedent of him making false reports in the face of hundreds of people who know exactly the kind of stuff he does?

    2. “You know better than to take people literally when they say they’re afraid of a defamation lawsuit”

      No, actually i think at least some of the people who said that were sincere. And I am not really familiar enough with REACH to know. And I wrote this because I wanted to broadcast what I said to them on Discord.

      “Trying to redirect attention away from the conspiracy of silence”
      Man, just wait until you see what I’m soon to publish if you think that’s remotely plausible.

      I’m not accusing Brent of anything he didn’t actually do.
      I’m not blaming problems on him that he didn’t actually cause.
      And he’s not as marginalized as he deserves.

      As far as talking about the actual problem. All the well meaning or pretending to be well meaning people keep telling me if I do break silence on MIRI apparently using donor funds to pay out to blackmail to deceive donors, on the force that inverted CFAR to optimize for anti-rationality, inverted FHI and MIRI to start and join an Armageddon race, has most of the I thought anti-complicity people I know saying I should keep silent to preserve the institution, saying that it will be reinterpreted as an attack on the wrong people or as additional social force granted to the wrong faction, I am laying some philosophical groundwork to attribute blame to what I believe to be the root problem.

      1. Oh, and to be clear, I don’t want people to punch MIRICFAR or FHI. When I said these things with more detail at the REACH as a sort of trial run, amid the ensuing gaslighting (which was mostly due to Oliver Habryka), the idea was raised that I should be being careful because of potential defamation lawsuits, and I clearly wasn’t being strategic.

        So I wanted to make it perfectly clear how the strategy works of just fucking saying the things.

      2. Why do you think they’re sincere? “We might get sued” is an extremely common general-purpose excuse used out of proportion to the actual risk; I see no reason to suppose that in this particular case people are actually doing an honest cost-benefit calculation.

        1. Man you don’t even know who I’m talking about, it was someone on Discord talking about specifically how they can’t financially survive unexpectedly having to pay a few hundred dollars and how if they are the particular person to say something they will become a target. Although I might be accidentally merging memories of multiple people to together here, not bothering to check.

  4. Note I don’t advocate punching psychologically evil people who nonetheless decide to behave morally because of “enlightened self interest”.

    1. Is long-term revenge that starts from immediate payback and lastcalates for months or years with the fixed goal of destroying the target a winning strategy for evil moral patients?

    1. When I played Warcraft III free-for-alls, I used to employ “pit one against the other” strategies. I had a ridiculous win-rate. My best streak was like, 8 or 10 games or something like that, of wins against usually 7 to 11 other players. I’d shout out into the fog of war, “no! please!”, randomly, and make each other of 2 remaining players think the other was eating me with impunity, clearly the biggest threat. And if I was crippled, I’d slowly regain my power, pretending to be in the game still only for vengeance against the stronger player, and I’d hide most of my ability to help that alliance until I was ready to swoop in with all my forces to end it. Or if deception wouldn’t work, and two people were teaming against me, I’d pick (generally) the weaker of the two, and launch an all-out base-race suicide attack. Make it absolutely clear that one person was going to bear the full cost of killing me unfairly. And I’d tell them, I could guarantee if they proceeded and killed me they would in turn lose to the other after I was dead, but if they betrayed their partner and town-portaled to save their base, it’d only be rational of me to myself teleport out to my base and see if I could defend from the stronger given an in-my-base advantage. That the likely outcome of this engagement was enough damage to the stronger of them (and also me ofc) that afterward they would be the strongest.

      Such shenanigans were really only for the endgame though. Up until then, it was just a race to eat the weak players. And I’d just try and keep a low profile, and always finish off whom I attacked. Because you couldn’t discretely predict how things were going to go like that with 8 people involved, there was too much chaos, and indeterminacy in how much you’d be interacting with any given person, which led to a lot of local non-zero-sum modeling.

      FFAs of course, were set up to have 1 winner and everyone else a loser. Treachery and perfectly rational pirates shit was the draw of the game. Real life is even farther in the direction of the early game from the late game than the early game for that reason, and for other reasons. I.e., it’s not a zero sum game at all, in total, even outside your modeling abilities.

      The in-practice answer is if I can step away, take with me whomever I can get out of that situation, and leave both evils to rot while I work on the spell of ultimate power, I favor my odds of winning the “do not crumble like everything in this world in the long term” game over my odds of winning a zero sum game where I am outnumbered way worse than 2 to 1.

      I have specifically fought evil people the common sense I received said were untouchable, and not suffered much losses in doing so. Vampires do not have much of a response to someone attacking them even if it doesn’t make causal-decision-theory sense. Like. their lives are pretty easy. They don’t have to fight hard. Don’t have to learn to fight that hard. So the all-out-attack strategy translates. “I am willing to die here and you are not.”

      There’s a convergent extension of that, and the “save what people I can” thing. I think the key insight behind that strategy was actually, “don’t take your overwhelming opponents’ coalition and unity for granted.”. It is in fact the case that the majority of expected value of the universe for any particular evil person comes from good winning. We have a much higher chance of winning than any particular evil, and evil people are still moral patients. So like, being able to communicate that is victory.

Leave a Reply

Your email address will not be published. Required fields are marked *