Bucky
Karma: 577
NewTop
Argh,you’re right,I didn’t check that one. P(OM) cancels on the P(G) equation so that one isn’t over constrained.
However for the equation for P(OM) 4 variables is over constrained, 3 is enough.
The 4 given probabilities are actually perfectly consistent within the equations you are using. It is provable that whatever 4 probabilities you use the equations will be consistent.
Therefore the question becomes “where did my maths go wrong?”
P(G|OM) = 0.055, not 0.55
I’m pretty confident that the only way probabilities can actually be inconsistent is if it is over constrained (e.g. in this case you define 5 relevant probabilities instead of 4). The whole point of having axioms is to prevent inconsistencies provided you stay inside them.
P.S. Good job on noticing your confusion!
Firstly I think the feature is a great idea and is working pretty well for a prototype.
Ideally the feature needs to do at least 3 things:
Get questions noticed
Get questions answered well
Present good Q/A combinations to the community
The current practice of promoting interesting questions to frontpage does the first but could be detrimental to the other 2 - many users will see the question but not get round to reopening and looking at the answers. This encourages people to answer quickly to get their response read and discourages more detailed answers.
If you’re planning on developing the feature further then addressing this issue would really help get the best out of it. One option would be the ability to promote the best answer(s) to frontpage after a week or so.
Thanks, got it! I hadn’t read the post as advocating a lack of decision but re-reading it now I can see how you read it.
I’m very confused about this. Here’s what I think you’re saying:
Choices are bad, particularly with regards to regret. It is better to make a decision based on instinct and forget about it than to consider your options carefully and potentially regret the decision.
Is that about right or is there something that I’m missing?
[Question] Experiences of Self-deception
Thanks for writing this, it’s not something I’d looked at before but I read some of the Promethease sample reports because you got me interested.
There does seem to be some weird normalisation going on when calculating magnitude. For instance this gene gives a score of 0 for (C;C) and bad magnitude of 2.7 and 3.1 for (C;G) and (G;G) respectively. So if you have (C;C) and you filter by magnitude 2 you miss out on the fact that you have an advantageous genotype.
This isn’t a problem if (C;C) is extremely common but actually it’s no more common than (C;G) (except for people of African descent), so the act of filtering prevents you from realising that you missed a 50:50ish chance of getting a disadvantageous genotype.
So probably to work this out properly you can’t filter by magnitude and you’d have to open up every genoptye details to check for what you’ve avoided getting hit by. You could only really work out how well you’ve done compared to other people where the data includes frequency so you could see just how lucky/unlucky you got for a particular gene.
Not all of the genotypes have this issue—for instance this gene seems to be more sensibly normalised. If they were all done like this then I’d be much happier with the system.
Good summary.
I’d like to add that sometimes definitions do matter, particularly in a public settings such as a blog. Even if t3tsubo and Benquo agree with both (a) and (b), it is possible that others reading the OP think that it is asserting (not-a).
If a significant number of people reading the blog are likely to think that it is asserting (not-a), rather than asserting (b), then it may be worth clarifying the OP to ensure that the correct message is received. I don’t know whether this would be a common misunderstanding, I can only conclude that at least one person read the post as asserting (not-a).
This is pretty much the same for me. I think the solution to bat and ball of “10cents, oh no, that doesn’t work. Split the difference evenly for 5 cents? yup that’s better” is all done on system 1.
Kahneman’s examples of system 1 thinking include (I think) a Chess Grandmaster seeing a good chess move, so he includes the possibility of training your system 1 to be able to do more things. In the case of the OP, system 1 has been trained to really understand exponential growth and ratios. I think that for me both “quickly check that your answer is right” and “try something vaguely sensible and see what happens” are both ingrained as general principles that I don’t have to exert effort to apply them to simple problems.
A problem which I would volunteer for a CRT is the snail climbing out of a well. Here there’s an obvious but wrong answer but I think if you realise that it’s wrong then the correct answer isn’t too hard to figure out.
From a frequentist perspective you’re right.
From a Bayesian perspective, my prior would be that the GS condition would make people more likely to help. The likelihood calculation reinforces this belief. In terms of bits, my prior would have been (conservatively) 2-3 bits in favour of an effect and the experiment adds another 3-4(?) so I end up on about 6 bits. 64:1 is pretty good.
I agree with what you say but feel like it’s the wrong kind of response for a post marked as speculative epistemic status. For such posts I think it’s fair to assume that the author knows they are over-simplifying and is just wanting to see where the idea goes.
A related question is when you should let people know where they stand with you. Prima facie, it seems good to make sure people know when they are safe. But that means it also being clearer when a person is not safe, which has downsides.
An interesting question.
Even if you are not specific about where people stand with you, they have the evidence of your past actions. So whenever you decide whether to stick with a friend or give up will provide evidence to your other acquaintances as to what they can expect from you.
If one is perceived as being too quick to ditch friends, it probably decreases availability of replacement friends. On the other hand, someone who is extremely loyal is likely to have greater availability of friends (up to a limit!) but also less need for new friends.
This surplus may give leverage for things one cares about—one might say “I’ll stand by my friends but I do expect they turn up when they say they’ll turn up”. Someone who is less loyal may not be able to be so picky.
Just spotted this comment is being put at the bottom by the magical sorting algorithm despite its high karma—maybe an artefact of having been marked as spam?
Done :)
Thanks for the comments, I’ve added section headings so hopefully it reads a bit easier now.
To be honest I really didn’t expect this to be as interesting to people as it was—glad to be proven wrong!
Ever wonder why people reply more if you ask them for a meeting at 2pm on Tuesday, than if you offer to talk at whatever happens to be the most convenient time in the next month? The first requires a two-second check of the calendar; the latter implicitly asks them to solve a vexing optimisation problem.
My experience is that this also makes people more likely to show up at the agreed time (using this method was suggested to me when I was working with students who were notoriously bad at showing up).
Possibly phrasing it this way creates an artificial significance to the time suggested, I don’t know, but it does seem to work. I generally offer as few options as is reasonable given other constraints.
Ok, I’m an idiot, this model doesn’t predict exponentially increasing required inputs in time—the model predicts exponentially increasing required inputs against man-hours worked.
The relationship between required inputs and time is hyperbolic.
This is an interesting theory. I think it makes some different predictions to the low hanging fruit model.
For instance, this theory would appear to suggest that larger teams would be helpful. If intel are not internally repeating the same research then them increasing their number of researchers should increase the discovery rate. If instead a new company employs the same number of new researchers then this will have minimal effect on the world discovery rate if they repeat what Intel is doing.
A simplistic low hanging fruit explanation does not distinguish between where the extra researchers are located.
Sticking with the throwing darts at a wall analogy, the linked post suggests that the problem is that no-one knows how close any of the darts are to the line. That problem would need to be solved before we could make progress.
Questions in humans often involve some kind of status interaction. A question is not only (or even always) a request for information but often also represents an offer to trade status for information.
(I realise that this is focusing on a very narrow sub-field of the question asked but you did ask for unusual framings!)
In a canonical case of requesting unknown information, the asker is lowering his status relative to the askee in trade for the information. The act of asking implies that, at least in this area, the askee has superior knowledge to the asker, thus increasing the askee’s prestige. In doing so it provides a trade to the askee for an answer.
This status transfer is often the reason that questions go unasked. The common refrain “there’s no such thing as a stupid question” is there to try to overcome this reluctance to ask by lowering the status tax on question asking. I often tell new starts at my work to ask as many questions as possible in the first 6 months because you’ll feel less silly during this period (i.e. the status tax is lower as expectations of your knowledge are lower). Asking questions after that will be necessary but it will be emotionally more costly as time goes on. One weighs up (not necessarily consciously) whether the information required justifies the expenditure of status.
One obvious result is a preference to ask questions in a smaller group so that the asker’s status is lowered in fewer people’s opinions. Of course this is reversed if the question is really an excuse to show off one’s own intelligence!
Similarly, one is tempted to search for information without making it obvious that you are asking a question thereby maintaining plausible deniability.
Questions generally being low status can be used to your advantage. In general you won’t tell your boss if he’s wrong but you might ask a question obliquely which might make him consider for himself that he might be wrong. If this works then you can correct his mistake without threatening his status (Caution: use with care!).
A manager might also use questions to correct an employee in order to be less status threatening to the employee.
I feel like questions in humans can’t be fully understood without some form of status interaction.
Meta Note
Having the question feature on LessWrong is interesting as it emphasizes that question asking is not a low status activity on the site. The potential issue is that askees might not feel like spending the time to answer a question provides sufficient reward. If this is overcome then I think the Q&A feature would count as a positive sum status interaction (at least on a status adaptation level).
I think responders need to be careful to avoid putting the status tax back on asking questions (e.g. by implying the question is stupid or the answer is obvious). I realise the need to distinguish good and bad questions (for clarity, I would include this question in the former) but I would prefer this to be done via moderation policy.
The issue with the status tax is that it optimises for whether an individual needs the information rather than whether it is a generally good question. For example, if a question is something that alot of people are interested in but no individual really desperately needs to know then the status tax makes it less likely to be asked. A good moderation policy should be able to encourage such questions.
(I’ve often had the experience that when someone finally overcomes their reluctance to pay the status tax and asks a question suddenly everyone says that they were wanting to ask that too—I wonder how often such questions just don’t get asked).