= 27d567bc2d48d9f40e9ccc20ea370b15)
Eugine_Nier comments on Harry Potter and the Methods of Rationality discussion thread, part 5 - 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 (648)
I suspect you can answer this question yourself: think about all the crimes you don't commit. Heck, think about all the crimes you didn't commit today. Why didn't you commit them?
If your answer is something other than "fear of being caught and punished," consider the possibility that other people might be like you in this respect, and threatening to punish you might not be the most cost-effective way to keep them from committing crimes, also.
But if you want more concrete answers, well, off the top of my head and in no particular order:
Increase P
Compare attributes of people (P1) who commit a crime given a certain perceived (p,EB,SP) triplet to those of people (P2) who don't commit that crime given the same triplet, and investigate whether any of those attribute-differences are causal... that is, whether adding a P2 attribute to P1 or removing an attribute from P1 reduces P1's likelihood of committing the crime. If any are, investigate ways to add/remove the key attributes to/from P1.
Decrease perceived EB -- for example, if a Weber's-law-like relationship applies, then increasing standard of living might have this effect.
Condition mutually exclusive behaviors/attitudes.
Arrange your society so that there are more benefits to be gotten by participating in it than by attacking it, and make that arrangement as obvious to the casual observer as possible.
I agree with your points in general; however, note that unlike increasing SP your suggestions can't simply be implemented by fiat.
Also given these things weren't done, I believe TDT requires us to use the values of p and EB at the time the crime was committed when calculating SP because those are the values would be dark lords are using to determine whether to start an overthrow.
Re: by fiat... yes, that's true. In behavior-modification as in many other things, the thing I can do most easily is not the thing that gets me the best results. This is, of course, not an argument in favor of doing the easiest thing.
Re: TDT... I don't see where TDT makes different requirements from common sense, here.
Re: using p/EB at the time of the crime... of course. If I want to affect your decision-making process now, the only thing that matters is the policy I have now and how credibly I articulate/ that policy. But that's just as true of my policy around how I investigate crimes (which affects p) as it is of my policy around how I select punishments (which affects SP).
Relatedly: yes, most of my suggestions require lead time; if you're in a "ticking time bomb" scenario your options are more limited. That said, I distrust such claims: it's far more common for people to pretend to exigent circumstances than it is for such circumstances to actually occur.
My point is simply that you shouldn't reduce the punishment after the fact, by say rescuing Bellatrix, simply because you have since changed the value of p and/or EB.
On the account you've given so far, I don't see why not.
If I've followed you correctly, your position is that severe punishment of prisoners is justified because it deters crime in the future.
But if I implement a 100% effective crime-deterrent -- say, I release a nanovirus into the atmosphere that rewires everyone's brains to obey the law at all times -- then from that moment forward severe punishment no longer deters crime. That is, I will get the same crime rate in the future whether I punish my current prisoners or not.
So why should I continue punishing them in that case? It seems like wasted effort.
Granted, none of the suggestions I've proposed are 100% effective. But it seems like the same argument scales down.
You're claiming that in order to deter crime today, I should establish an SP inversely correlated with p (among other things). If I raise p today, then, it follows that I should lower SP today to keep deterrence constant. What benefit is there to continuing to punish existing prisoners under the old SP?
Otherwise your new value of SP isn't credible. After all, you're likely to lower it again in the future and then apply the change retroactively.
If I assume that changes to SP are retroactive but that changes to p and EB aren't... for example, if I assume that if today I increase my ability to catch criminals (say, by implementing superior DNA scanning), this only affects criminals who commit crimes today or later, not criminals who committed a crime last year... then I agree with you.
If that's not true, then I don't agree. The same logic that says "Dave will probably lower SP in the future, so I should apply a discount factor to his claimed SP" also says "Dave will probably raise p in the future, so I should apply an inflation factor to his claimed p." And since what's driving the reduction in SP in this toy example is precisely the increase in P, the factors should offset one another, which keeps my level of deterrence constant.
Now, I grant you, this assumes a rather high degree of rationality from my hypothetical criminal. In the real world, I strongly doubt any actual criminals would reason quantitatively this way. But in the real world, I strongly doubt any actual criminals reason quantitatively from EB, SP, and p in the first place.
Well, retroactive changes to p tend to be much smaller since most evidence degrades with time.
Also in this case since the crime is attempting violent overthrow of the government retroactive changes in p are almost non-existent, after all a successful overthrow by its nature virtually eliminates your chances of getting punished for it.
That's a fair point. So, yes: if p is effectively constant and SP is not, you're right that that's a good reason to keep applying the old SP to old prisoners. I stand corrected.
So are you saying the SP-setting strategy you're proposing doesn't apply to crimes that don't destabilize the criminal justice system itself?
I'm saying what I said and hopefully what's true, redo the calculations yourself if you like. Here I'm saying that if a crime has the potential to destabilize the criminal justice system itself, that should be taken into account when calculating p.