Death-penalty analysis reveals extent of wrongful convictions

Are we ever 100% sure though, when the possibility exists that the prosecution is corrupt and the confession could be coming from someone that is mentally ill?

And stepping aside from morality, if it costs 10X[0] the cost of a life prison sentence to kill someone, isn't it really better to just put them away for life with no chance of parole?

[0] 10X is a made up #, but it's certainly a large multiple of non-capital punishment cases.

Well, never being 100% certain - about anything involving the world and not a pure tautology - is always correct. As they say on Less Wrong, "zero and one are not probabilities". When you say you're 90% certain, you mean if you answered 10 independent questions of that difficulty, you'd tend to get about one of them wrong. When you say you're 99% certain, you raise that to 100 independent questions. When you say you're 100% certain, that's not just a few more questions, or a lot more questions; if your assessment is well calibrated, you're claiming you could answer infinite such questions and never be wrong. Can you do that? Even stepping (infinitely far) back to 99.999% certain, that's a lot of questions (100,000) to get only one wrong. Consider the following anecdote (by way of illustration and to drive intuitions, not as any sort of proof of course): http://www.spaceandgames.com/?p=27

Having said that, we do not in any sense need to be 100% confident and it's kind of a weird distraction here. Yes, with life-in-prison we have the ability to "fix a mistake" in a sense, but 1) people don't get back the time they already spent in jail obviously, and 2) we often don't do that: remember that the death penalty is so much more expensive in large part because we allow appeals that we don't allow with life-in-prison; in any case where one of those appeals would have demonstrated innocense, we're foregoing fixing our mistake so the hypothetical ability means nothing.

Having said all of that, I oppose the death penalty. It's a needless shift of power toward the already powerful government, without really any gain. Studies repeatedly show it has little deterrent effect over life (or even just a significant chunk of time) in prison. Another study suggests it may do bad things to juries and make them more likely to wrongfully convict when the death penalty is an option. It's super expensive compared with just keeping people incarcerated (which is expensive compared to sending them to college, but that's another matter...). And it doesn't do good things to us - forgiveness is much better than vengeance once we've already dealt with deterrance. And from a Utilitarian perspective (again, once we've already factored in deterrance) harm is still harm and is - first order - bad, even when it's to "bad people".

There are two ideas.

One thought is that you can back out of a mistake on a very long prison case. You can't back out of the mistake once someone is dead. So let's say you're 99.9% sure someone molested a bunch of kids. You can sign them up for life in prison on those odds. But you shouldn't do it for life, because there's a chance that you're wrong. Please don't ask what the right % is, as I don't know that #. But I do accept that we can never be 100% of any crime. (Is the prosecutor corrupt? Was the suspect mentally ill and not really confessing?) But that doesn't mean we don't punish crime at all. It means we need to have a method of redressing mistakes, which we lose in capital punishment cases.

There's a separate economic argument that just says it's not economic to go through the hassle of it, but this is not a moral arguement.