= 27d567bc2d48d9f40e9ccc20ea370b15)
SquirrelInHell comments on How It Feels to Improve My Rationality - Less Wrong Discussion
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 (50)
This I expect to be pretty universal, so if you think about how you do it you'll have a good idea. I'm still going to answer though. Briefly, it seems to be a combination of:
monitoring effectiveness, increase in ability to solve actual problems and make predictions,
intuition, sense of elegance, feeling that the theory "clicks",
checking against other people, both by listening to them and penalizing any solution that breaks existing rules/trends.
The problem is, if I go solely by internal perceptions/feelings I can't reliably distinguish the cases where I'm a beacon of light and reason and where I'm an arrogant self-deluding idiot. What I need is real-life testing.
So yes, I agree with the "effectiveness" point, but at least in my case I have doubts about elegance and "clicks". To figure out whether something "clicks" is easy for me, so that's an early threshold an idea/theory/explanation has to pass. And "checking against other people" is not terribly useful because if I'm right then they are doing it wrong so the check will only confirm that we see things differently.
All of this is true. Though in many cases when people "are doing it wrong" you find not that they have opinions opposed to you, you find that they don't have any consistent opinion at all. Which makes it OK to stick with your version until you find something better.
I'd mention that in many cases the best thing to do might be to lay off the topic for some time, work on other problems, improve your overall thinking, check facts known from respectable science, wait for your feelings of attachment to die, and revisit the original topic with a fresh perspective much later.
This can be repeated many times, and I guess it's actually the core of my description of caring about "pastures". This is a kind of a meta-technique that seems to be central to not becoming "stuck" in stupidity.
Well, they might not be expressing any consistent opinion, but if they are doing the same thing over and over, then there is a clear implied position (similar to revealed preferences).
Might be -- unless you need to make a decision in the near future. If the topic is something you can ponder for a long time without needing to come to any conclusions, well, the question that comes to my mind is "Are you sure it's important?" :-/ (yes, I know that's not applicable to science)
That's also frequently happening with people adopting wrong beliefs.
Yes. I'm not claiming to be infallible, but I also suppose that having done a lot of abstract math helps me to know good thinking when I see it. Especially in cases when I can go deep enough and follow the whole thing from "first principles".
Being convinced that a single theory derived from first principles explains everything about a complex domain seems to me like having a hedgehog perspective on the domain.
That means you are unlikely to be very good at predicting over the domain by the findings of Tedlock.
You are jumping to assumptions about what I do, and how I think.
Well, thanks for the warning anyway. It's good to keep it in mind.
That's part of trying to understand what somebody else thinks. It's good to make assumptions to prevent a statement to be to vague to be wrong. If you think I made incorrect assumptions feel free to say to correct mistaken assumptions.
I'm not SquirrelInHell, but I'll point out what looks to me like one substantial misunderstanding.
SIH said that s/he finds that mathematical training gives a good sense of good versus bad thinking in cases of the "rigorous reasoning from first principles" kind. You responded as if SIH were claiming to be explaining everything about a complex domain using such reasoning, but s/he made no such claim.
Perhaps this analogy will help. Suppose I write something about improving my abilities in graphic design, and am asked how I distinguish genuine improvements from (say) mere increases in arrogance. I list a number of criteria for distinguishing one from the other, and one of them is something like "When the design has a strong short-term commercial focus, like an advertisement or a page on a merchant's website, we can measure actual sales or conversions and see whether I've successfully increased them". And then you object that it's wrong to reduce everything to counting money. So it is, but that doesn't mean that when something is about money and it can be counted you shouldn't do so.
The situation here is just the same. Not everything is about careful logical reasoning from first principles, but when things are a good sense of when they're correct is helpful. And yes, mathematicians are good at this. (I don't know how much of that is selection and how much is training.)
That's not the only claim. If you look at the post there the claim that there's polarization. That being rational makes him see less shades of gray.
two sensible sounding ideas become one great idea and one stupid ideaFor that to happen he has to call those ideas that are in line with his first principle derived theory great and ideas that are not in line with it stupid.Let us take an example. An aspriring rationalist finds that status is important for social interactions. He then rethinks all of his thinking about social interactions based on the first principle of status. That person will see the signs that SquirrelInHell described in the OP as the signs for increased rationality about the domain.
Or take one of those libertarians who try to boil down all of politics to being about violence. That produces those signs that SquirrelInHell describes but has nothing to do with real rationality.
It's the one I thought you were responding to.
My interpretation was that all those signs are potentially separate; in a given place, some will apply and some won't. The situation you describe applies, at most, to those cases that (a) SquirrelInHell thinks are resolvable from first principles and (b) SquirrelInHell now feels more polarized about.
So let's suppose we're only talking about those cases -- but note, first, that there's no reason to think that they're very common. (If SquirrelInHell finds that most cases are like that, then I agree that may be a bad sign.)
In that case, I agree that it is possible to go wrong by leaping into some oversimple crackpot theory. But so what? SIH listed intuition/elegance/"clicking" as just one of several signs to distinguish real from fake improvements. Any one of them may lead you astray sometimes. (All of them collectively may lead you astray sometimes. Sometimes the world just screws you over.) The question is not "can I think of counterexamples?" -- of course you can -- but "will this heuristic, overall, make you more or less accurate?".
I don't know whether SquirrelInHell has watched to see whether that sense of elegance does actually correlate with correctness (either globally or in some particular cases -- heuristics can work better in some situations than others). For that matter, I don't know whether you have (but SIH's sense of elegance might differ from yours).
Suppose, as per your first example, someone runs across the notion of social status and completely reframes his thinking about social interactions in terms of status. They may, as you say, feel that "everything makes sense now", even though in fact their thinking about social interactions may have become less effective. So let's look at the other signs SquirrelInHell lists. Does our hypothetical would-be-rationalist become more effective in interacting with others after this status epiphany? (If so, I would take that as evidence that "it's all status" is a better theory than whatever s/he was working with before. Wouldn't you?) Does discussion with other people throw up obvious problems with it -- especially obvious problems that the previous theory didn't have? (If so, then again I would take that as evidence in favour; wouldn't you?)
Note that for "it's all status" to be an improvement in rationality it's not necessary for "it's all status" to be correct. Only that it be more correct than whatever our hypothetical would-be-rationalist thought before. (Kepler noticed that planets seem to move in approximately elliptical orbits with certain nice properties. This was wrong -- because of the gravitational effects of other bodies besides the sun and the planet, and because Newtonian physics is wrong -- but it was a lot better than what had come before.)
Thank you for arguing calmly and patiently. I don't trust myself to do this, seeing how I have already failed once to keep my composure in my line of discussion with ChristianKl.
I'm responding to a mental model of his position based on what he wrote. No single statement is responsible for the full model.
I don't think the concern is simple about crackpot theories. It's about trying to explain everything with one theory. You can do that successfully in physics but in many contexts it's you can't do everything with one theory.
Yes. I think the heuristic of following the Superforcasting principles is better. That means developing more shades of gray and thinking foxy instead of thinking like a hedgehog.
The status-hedgehog might be better at a few interactions at the cost of not being able to have genuine connections with others anymore. He would be more effective if he would be foxy and would say: Status is important, but there are also other important factors.
I don't think that looking for positive real world effects or looking at whether discussion with other people throw up obvious problems are filter that successful protect from hedgehog thinking.
There nothing wrong with using first-principle thinking. If you however use it to come up with a view and then call all ideas that align with that view great and all that don't align stupid you are making a mistake. You are using a bad heuristic.
Now you have made a general point that can be easily argued both ways.
Tell me the strongest counter-arguments you can think of against what you just said.
(I predict you to agonize over this, produce strawmans, and have a strong impulse to dodge my request. Am I wrong?)
Edit: This was a bad way to handle this on my part, and I regret it. The flip side to ChrisitanKl's statement is probably obvious to anyone reading this (confirmed with a neutral third party), and I wanted to somehow make ChrisitanKl see it too. I don't know a good way to do this, but what I wrote here was certainly not it.
Why do you think that would be helpful?
It seem to me like you don't want to engage with discussion. As a result it doesn't me to try to find counter-arguments against what I'm saying.
Notice how I made a successful prediction that you will try to dodge my request.
It would be helpful to you, if you want to improve your rationality, as opposed to feeling good.
Edit: I retract this, since it is not a helpful way to advance the discussion.
That happen to be false. You predicted something related but different. But predicting that people won't go along with unreasonable requests doesn't require much skill.
It's also intersting that you call it dodgin when I ask you to provide reasons for why you think what you recommend is good.
I don't see how going along with people who are evasive generally increases my rationality. In general the sequences also recommend against playing devils advocate and don't see it as raising rationality.