Quote:
Originally Posted by Nichlemn
The point of the terrorism example is that there are lots of policies that we can't prove have statistically significant effects, but that doesn't mean we should assume they're completely useless. In the most extreme cases, like unique foreign policy decisions, there may be no data whatsoever to go on. Could you prove say, that going after bin Laden caused a statistically significant reduction in terrorism? Of course not. Should we assume then, that absolutely no good came from it other than feelings of vengeance or justice (which you would probably call "barbaric"?) Of course not. When we lack empirical evidence, we should lean towards theory, and the theory that getting rid of a terrorist mastermind would help reduce terrorism is perfectly reasonable.
FWIW, conversations about assassinating terrorist leaders often include struggles predicting blowback and issues discerning whether tensions between terrorist groups and the States would worsen as a result. I think your assumption that assassinating Osama and/or targeted drone killings of other terrorists unequivocally decreases terrorist activity is a bit of a jump.
Perhaps more directly related to our topic, I wonder how negative unintended consequences of assassinations might decrease if these individuals were (hypothetically) apprehended, tried, convicted, and put in prison for life and NOT executed. I know the reasons why apprehension isn't usually opted for, but I'm just putting it out there to think about and to keep us on topic.
Quote:
Originally Posted by Nichlemn
The death penalty is a milder version of this. We do have some empirical evidence. However, the death penalty is applied so infrequently (relative to the number of violent crimes) with so many confounding variables that the evidence should not cause us to adjust our priors very much. The evidence certainly points to there not being a very large deterrent effect, otherwise it would have likely shown up despite all the confounding variables. But if the death penalty caused a small decrease in homicides, we very likely wouldn't be able to reliably detect it with statistical significance.
There have only been a few dozen executions per year in the US, so even if each one deterred several murders, there are about 15,000 homicides in the US every year - the number of deterred murders would be less than 1% of the total. I would personally be happy with a deterrent rate considerably lower than that, so we could be talking about a <0.1% decrease in homicides. The evidence just doesn't support a 0.0% decrease in homicides as being significantly more likely than 0.1% decrease. At that level of precision, the margin of error is so high that the empirical evidence is basically useless. Theory is a much better guide in such cases.
I accept the limitations of our relatively small sample here in the States, but many interested specialists would disagree that the sample is insufficient. Moreover, a healthy portion of studies showing a deterrent effect don't control for the deterrence of life sentences, which really decreases the credibility of pro-deterrence studies and lop-sides the pro/con data debate. We've got data showing that murder rates in States with the death penalty consistently being higher than States without it, year after year, for decades. Criminologists overwhelmingly agree (like 9 out of 10) that the death penalty doesn't deter murder. These are impressions formulated from social scientific study and reviews of empirical data. I'm not convinced that the phenomenon can't be sufficiently analyzed through use of data and, therefore, we gotta turn to theory alone.
Quote:
Originally Posted by Nichlemn
You talk of this as if it's trusting my "instinct" over the "evidence" which refutes my "instinct", but that's not true, it's about being Bayesian. My priors were that there is very likely a small deterrent effect. The fact we haven't found it yet, when it would be hard to find if it did exist, makes me only slightly less likely to believe it. Your priors were probably that there was no deterrent effect, so the lack of its discovery should increase your credence slightly. Both these responses are rational. The only way to come to agreement, bar non-existent smoking gun empirical evidence, is to convince the other person that their priors were wrong to begin with. That means discussing theory.
Any thoughts on what kind of sample you would need to respect research findings, or what controls you would need to see that aren't being covered currently by social researchers? Like, what would it take for you to recognize results which can't disprove the null (if I said that right...)?
A better description of my con position is that empirical data and the majority opinion of relevant specialists sway me to believe a deterrence effect isn't present above and beyond that of life imprisonment.
Quote:
Originally Posted by Nichlemn
All I'm saying is that people exonerated ≠ number of innocent people executed. I do agree that it's troubling that a significant number of innocent people may have been executed and it's why I support a higher standard of proof.
I find it troubling that you can't even admit that a significant number of innocent people HAVE SURELY been executed.
Quote:
Originally Posted by Nichlemn
You could have a larger jury, but that's probably not necessary. After all, civil trials can be conducted with a lower burden of proof without requiring anything special. It would probably be sufficient to simply instruct juries to apply a higher standard of doubt. The problem with insisting on DNA evidence is that you could have overwhelming evidence in every other area but you lack DNA evidence for some reason, such that the accused is a billion-to-one likely to be the perpetrator but can't get the death penalty, whereas it might be a lot less than a billion to one that someone else who was convicted on DNA evidence was guilty. Still, if you're sufficiently suspicious that juries can be trusted with such leeway, such a standard would be better than nothing.
So a higher higher burden of proof than "beyond a reasonable doubt?" Wouldn't a higher burden be "beyond unreasonable doubt?" I don't even know what that means. I guess I need to know more about what exactly you're getting at.