= 27d567bc2d48d9f40e9ccc20ea370b15)
Jackercrack comments on A discussion of heroic responsibility - Less Wrong
You are viewing a comment permalink. View the original post to see all comments and the full post content.
= 783df68a0f980790206b9ea87794c5b6)
Loading…= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
Powered by Reddit
= f037147d6e6c911a85753b9abdedda8d)
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (215)
I think heroic responsibility is essentially a response to being in a situation where not enough people are both competent at and willing to make changes to improve things. The authority figures are mad or untrustworthy, so a person has to figure out their own way to make the right things happen and put effective methods in place. It is particularly true of HPMOR where Harry plays the role of Only Sane Man. So far as I can tell, we're in a similar situation in real life at the minute: we have insufficient highly sane people taking heroic responsibility. If we had enough sane people taking heroic responsibility, things would look rather different and likely a lot better run. It would be easy to be a happy gear, if you knew the machine was properly designed, the goal was a good one and the plan was likely to succeed.
There is clearly some ratio of heroically responsible:useful gears that works the best for each situation, some optimal equilibrium. Too many people trying to unilaterally change things in different directions and you get chaos and infighting. Too many useful gears and you have a wonderfully maintained, smooth running machine working at 40% maximum efficiency towards a goal that doesn't make much sense. I propose heroic responsibility be fit into a larger framework, that of filling the role that is required. I can't think up a snappy name for it, but you essentially mould your actions into the shape that best maximises your group's outcome given your abilities. If there is already someone doing the job of heroic responsibility and doing it well, you aim for the next empty or poorly-done role down where you do most good. If all the important positions are filled by competent people or by people more competent than you, then be a gear.
The problem lies in knowing whether you actually could do better in someone's situation with the resources available to them. The ethical injunctions seem to say that humans are rather bad at figuring out when they could do better than those in power. There is only an injunction against cheating to gain power though, not against legitimate means. Still, it's difficult to say whether one is actually more competent and a large amount of research into the potential role should be done first before action is taken. If you can see a decision for the role coming and make a strong prediction of what a good choice would look like, then seen what the actual decision the person made was and predict the effects and finally get it right, with a better batting average than the person in the role, if you can do that, then it's time to go for the position.
Disclaimer: There is a possibility that this theory is an extension of my belief that I should eventually be in charge.
I don't know about that. Let me offer you an example of a not-mad person who took heroic responsibility: Lenin.
Generally speaking, it's all very tightly tied to values. If you share the values, the person "takes heroic responsibility", if you don't share the values, the person is just a fanatic.
You say he's not-mad, but isn't he the spitting image of the revolutionary that power corrupts? Wasn't Communism the archetype of the affective death spiral?It would appear he was likely suffering from syphilis, a disease that can cause confusion, dementia and memory problems. Anyway, isn't that an ad hominem argument?
No. It is an argument which happens to use the perceived negative consequences of an individual's actions as a premise. Use of 'ad hominem!' to reject a claim only (legitimately) applies when there is a fallacy of relevance that happens to be a personal attack that doesn't support the conclusion. It does not apply whenever an argument happens to contain content that reflects badly on an individual.
Lenin in the 1920s is not relevant to this argument, I would say he "took heroic responsibility" around, say, 1915-1918, and It looks to me that it would be hard to make the argument that he was already corrupted by power at this point.
But if you don't like this example I'm sure I can find others. The underlying point is rather simple -- imagine "enough sane people taking heroic responsibility" with these people having a value system you find unacceptable...
I think we're using a different meaning of the word sane. See, I hold sanity to a rather high standard which excludes a huge breadth of people, probably myself as well until I've progressed somewhat.
I am imagining enough sane people taking heroic responsibility, the world looks rather different than this and it seems to be better run. We already have people in charge with value systems unacceptable to me, making them at least competent and getting them to use evidence-based strategies seems like a step forwards. People will have a normal range of value systems, if a particularly aberrant person comes with a particularly strange value system, then they'll still have to outsmart all the other people to actually get their unacceptable value system in place
Honestly lumifer, I'm beginning to thing you never want to change anything about any power structure in case it goes horribly wrong. How are things to progress if no changes are allowed?
Why is it a step forward? If these people have value systems unacceptable to you, presumably you want them stopped or at least slowed. You do NOT want them to become more efficient.
That, um, is entirely non-obvious to me. Not to mention that I have no idea what do you mean by "normal".
Oh, I do, I do. Usually, the first thing I want to do is reduce its power, though :-D
But here I'm basically pointing out that both rationality and willingness to do something at any cost (which is what heroic responsibility is) are orthogonal to values. There are two consequences.
First, heroic responsibility throws overboard the cost-benefit analysis. That's not really a good thing for people who run the world to do. "At any cost" is rarely justified.
Second, I very much do NOT want people with values incompatible with mine to become more efficient, more effective, and more active. Muslim suicide bombers, for example, take heroic responsibility and I don't want more of them. True-believer cultists often take heroic responsibility, and no, I don't think it's a good thing either. It really does depend on the values involved.
See, you're ignoring the qualifier 'sane' again. I do not consider suicide bombers sane. Suicide bombers are extreme outliers, and they kill negligible numbers of people. Last time I checked they kill less people per year on average than diseases I had never heard of. Quite frankly, they are a non-issue when you actually look at the numbers.
It is not obvious to me that heroic responsibility implies that a thing should be done without cost/benefit analysis or at any cost.
Of course it depends on the values systems involved, I just happen to be fine with most values systems. I'll rephrase normal values systems to be more clear: People will on average end up with an average range of value systems. The majority will probably be somewhat acceptable to me, so in aggregate I'm fine with it.
Is there a specific mechanism by which reducing government power would do good? What countries have been improved when that path has been taken? It seems like it would just shift power to even less accountable companies.
Well, would you like to define it, then? I am not sure I understand your use of this word. In particular, does it involve any specific set of values?
Things done on the basis of cost-benefit analysis are just rational things to do. The "heroic" part must stand for something, no?
Ahem. Most out of which set? Are there temporal or geographical limits?
That's a complicated discussion that should start with what is meant by "good" (we're back to value systems again), maybe we should take it up another time...
I had always assumed it was intended to stand for doing things that are rational even if they're really hard or scary and unanticipated.
If you do a careful cost-benefit calculation and conclude (depending on your values and beliefs) that ...
and if you are a normal person then you shrug your shoulders, say "damn, that's too bad", and get on with your life; but if you are infused with a sense of heroic responsibility then you devote your life to researching AI safety (and propagandizing to get other people thinking about it too), or become a missionary, or live in poverty while doing lucrative but miserable work in order to save lives in Africa.
If it turns out that you picked as good a cause as you think you did, and if you do your heroic job well and get lucky, then you can end up transforming the world for the better. If you picked a bad cause (saving Germany from the Jewish menace, let's say) and do your job well and get lucky, you can (deservedly) go down in history as an evil genocidal tyrant and one of the worst people who ever lived. And if you turn out not to have the skill and luck you need, you can waste your life failing to solve the problem you took aim at, and end up neither accomplishing anything of importance nor having a comfortable life.
So there are reasons why most people don't embrace "heroic responsibility". But the premise for the whole thing -- without which there's nothing to be heroically responsible about -- is, it seems to me, that you really think that this thing needs doing and you need to do it and that's what's best for the world.
("Heroic responsibility" isn't only about tasks so big that they consume your entire life. You can take heroic responsibility for smaller-scale things too, if they present themselves and seem important enough. But, again, I think what makes them opportunities for heroic responsibility is that combination of importantly worth doing and really intimidating.)
If you're a normal person, the fact that you shrug your shoulders when faced with such things is beneficial because shrugging your shoulders instead of being heroic when faced with the destruction of civilization serves as immunity against crazy ideas and because you're running on corrupted hardware, you probably aren't as good at figuring out how to avoid the destruction of civilization as you think.
Just saying "I'm not going to shrug my shoulders; I'm going to be heroic instead" is removing the checks and balances that are irrational themselves but protect you against bad rationality of other types, leaving you worse off overall.
Well, here is a counter-example. I can't imagine that was too intimidating :-/
I'll put this in a separate post because it is not to do with heroic responsibility and it has been bugging me. What evidence do you have that your favoured idea of reducing political power does what you want it to do? Are there states which have switched to this method and benefited? Are there countries that have done this and what happened to them? Why do you believe what you believe?
Well, before we wade into mindkilling territory, let me set the stage and we'll see if you find the framework reasonable.
Government power is multidimensional. It's very common to wish for more government power in one area but less in another area. Therefore government power in aggregate is a very crude metric. However if you try to imagine government power as an n-dimensional body in a high-dimensional space, you can think of the volume of that n-dimensional body as total government power and that gives you a handle on what that means.
Government power, generally speaking, has costs and benefits. Few people prefer either of the two endpoints -- complete totalitarianism or stateless anarchy. Most arguments are about which trade-offs are advantageous and about where the optimal point on the axis is located. To talk about optimality you need a yardstick. That yardstick is people's value system. Since people have different value systems, different people will prefer different optimal points. If you consider the whole population you can (theoretically) build a preference distribution and interpret one of its centrality measures (e.g. mean, median, or mode) as the "optimal" optimal point, but that needs additional assumptions and gets rather convoluted rather fast.
There are multiple complicating factors in play here. Let me briefly list two.
First, the population's preferences do not arise spontaneously in a pure and sincere manner. They are a function of local culture and the current memeplex, for example (see the Overton window), and are rather easily manipulated. Manipulating the political sentiments of the population is a time-honored and commonplace activity, you can assume by default that it is happening. There are multiple forces attempting the manipulation, of course, with different goals, so the balance is fluid and uncertain. Consider the ideas of "manufacturing consent" or the concept of "engines of consent" -- these ideas were put forward by such diverse people as, say, Chomsky and neoreactionaries.
Second, the government, as an organization, has its own incentives, desires, and goals. The primary among them is to survive, then to grow which generally means become more powerful. Governments rarely contract (willingly), most of the time they expand. This means that without a countervailing force governments will "naturally" grow too big and too powerful past that optimal point mentioned above. Historically that has been dealt with by military conquests, revolutions, and internal coups, but the world has been quite stable lately...
I'll stop before this becomes a wall of text, but does all of the above look reasonable to you?
Okay, my definition of sane is essentially: rational enough to take actions that generally work towards your goals and to create goals that are effective ways to satisfy your terminal values. It's a rather high bar. Suicide bombers do not achieve their goals, cultists have had their cognitive machinery hijacked to serve someone else's goals instead of their own. The reason I think this would be okay in aggregate is the psychological unity of mankind: we're mostly pretty similar and there are remarkably low numbers of evil mutants. Being pretty similar, most people's goals would be acceptable to me. I disagree with some things China does for example, but I find their overwhelming competence makes up for it in aggregate wellbeing of their populace.
gjm gives some good examples of heroic responsibility, but I understand the term slightly differently. Heroic responsibility is to have found a thing that you have decided is important, generally by reasoned cost/benefit and then take responsibility to get it done regardless of what life throws your way. It may be an easy task or a hard task, but it must be an important task. The basic idea is that you don't stop when you feel like you tried, if your first attempt doesn't work you do more research and come up with a new strategy. If your second plan doesn't work because of unfair forces you take those unfair forces into account and come up with another plan. If that still doesn't work you try harder again, then you keep going until you either achieve the goal, it becomes clear that you cannot achieve the goal or the amount of effort you would have to put into the problem becomes significantly greater than the size of the benefit you expect.
For example, the benefit for FAI is humanities continued existence, there is essentially no amount of effort one person could put in that could be too much. To use the example of Eliezer in this thread, the benefit of a person being happier and more effective for months each year is also large, much larger than the time it takes to research SAD and come up with some creative solutions.
The definition you give sounds like a pretty low bar to me. The fact that you're calling the bar high means that there are implied but unstated things around this definition -- can you be more explicit? "Generally work towards your goals" looks to me like what 90% of the population is doing...
Is it basically persistence/stubborness/bloodymindedness, then?
Really, last time I checked there is now a Caliphate in what is still nominal Iraq and Syria.
To me Snowden is one of the best examples of taking heroic responsibility. All the way to potentially breaking the laws and getting into the harms way to make the world a better place.