Less Wrong is a community blog devoted to refining the art of human rationality. Please visit our About page for more information.

Wind comments on Guessing the Teacher's Password - Less Wrong

62 Post author: Eliezer_Yudkowsky 22 August 2007 03:40AM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Comments (92)

Sort By: Old

You are viewing a single comment's thread. Show more comments above.

Comment author: Wind 05 June 2016 09:41:48PM 1 point [-]

I am unsure if we are disagreeing or not. I think that it is bad if the system encourage people to go for the wrong incentives. My point is that, I believe that people know when they are hacking the system. I think that the students themselves know that their hypothesis is about the teacher and not the plate.

This is a life thing. One programming example might be people running code they've copy-pasted off of StackOverflow to see if it solves their problem--they don't understand what it will do, but they have a vague hope that it will be the magic incantation that will do what they want it to do.

If my goal is to just make the program work, then copy-past from StackOverflow might be a good idea. As long as I know what I am doing, and don't fool myself in to thinking that I understood what I just copy pasted, I don't see the problem.

I have done a little amateur programming and I admit that I have used this method. Of course I would prefer to understand everything, but at one point of an other, I just wanted some lines to do X for me, so that I could get to the part of the code that I was actually interested in.

Probably a better example is dysfunctional organizational dynamics, where guessing what the boss wants you to say serves you better than trying to estimate what best accomplishes organizational goals.

Yes, that is a good life example. However, in this example I think that it is even more clear that the employee has accurate beliefs about the world. The error is with the system, not with the employee.

This is not a hypothesis about the metal plate. This is not even a proper belief. It is an attempt to guess the teacher's password.

Guessing the teacher's password is obviously a hypothesis--but it's a hypothesis about the teacher, not the plate.

I agree with you, Vaniver, as you say: "it's a hypothesis about the teacher" But I disagree with Yudkowsky. A belief about the teacher is a proper belief.

Yudkowsky claims that Guessing the teacher's password is a behavior that occurs because the student does not understand their own knowledge or lack there of.

I claims that Guessing the teacher's password is an example of perverse instantiation. The students have correct beliefs and are doing the rational thing, given their incentive structure. They don't think that they understand heat conduction, and they don't care, because understanding heat conduction is not their goal. Their goal is to get acceptable grades with minimum amount of effort.

Using proxi-intensives works badly on intelligent agents, even if they are made out of flesh.

Comment author: hairyfigment 06 June 2016 06:44:06AM 0 points [-]

While you could be right, you're claiming the students have a clear and accurate model of their own beliefs. What does that mean? Could they explain the nature of technical explanation?

There's a certain popular series of books which portrays intelligence as a matter of parroting facts, without trying to connect any two of them - not even the disappearance of every child in the world and an event in world politics shortly thereafter. Now, you could try to explain this by saying the authors (plural) are deliberately selling their customers garbage to maximize return-on-investment. And you could try to claim that their fans are buying the books just to signal tribal membership. And that explanation may have some power - but I draw the line at saying that they all understand clearly what the books lack in terms of credibility.

Comment author: Wind 27 July 2016 09:44:56PM *  0 points [-]

This answer have taken some time because I wanted to read the link you gave, before writing back. I still have not read it most of it, but I think I have read enough to get your reference.

I can't comment on the books, because I have no idea what series you are talking about. But your tone do suggest that you expect me to know what series this is? I am guessing that these books are very popular in USA, but more or less completely unheard of in Europe? Probably something with at Christian theme?

I know for certain that there exists students in the world, that uses teachers password, or similar techniques, fully knowing of what they do. I am slowly accepting that there probably also exists people who think they know stuff when all they have is a statement they do not actually understand. I have currently no good estimate as to which of these are most common.

As to weather the first type of students could explain the nature of technical explanation? Why do you mean exactly? I am not absolutely sure about what Yudkowsky mean by this concept, but that only mean that I am uncertain about his mind, not about my own.

To me, an explanation does not feel like an explanation, unless I understand all the bits, and I can not remember ever thinking differently. If I would be told for the first time that light is a wave, then I would try to fit my current best understanding of light with my current best understanding of wave, to try to figure out what "light is a wave" could possibly mean, and then I would ask for more information, because that is clearly not an explanation. This must have happened at some time, even though I can't remember the exact event. I do know that something in my childhood triggered me to want to know more about water waves.

For me it is really hard to imagine that anyone could confuse a teachers password with knowledge, which makes me biased towards other explanations. So maybe I am wrong. But also, do not underestimate peoples willingness to knowingly use tricks to pas a class, or get better grades. Here are two examples that I rememberer classmates openly talking about:

  • Use extra sloppy handwriting to council spelling mistakes.
  • On an open question, just write everything you can think of, that seems at least semi relevant and hope that you included what ever the teacher was getting at, some where in there.