This is a blog to discuss philosophy, chess, politics,
C. S. Lewis, or whatever it is that I'm in the mood to discuss.
Wednesday, June 27, 2007
Punishing the innocent by mistake
Is it better for a hundred guilty criminals to go free than for an innocent person to be wrongfully convicted and deprived of their liberty? Why do you think this? Does this affect how you think when it comes to the death penalty?
6 comments:
Anonymous
said...
Is it better for a hundred guilty criminals to go free than for an innocent person to be wrongfully convicted and deprived of their liberty?
No.
Why do you think this?
For utilitarian reasons (not that I'm a utilitarian). Depriving someone of her liberty is a very serious offense. There should be a lot of room for judicial appeal, etc., but it is in society's best interest to prevent the harm perpetrated by the hundreds of wrongly released criminals.
I do believe, though, that standards should be high for guilty verdicts (e.g. innocent until proven guilty, burden on the prosecution, etc.), so maybe this does put me in the "yes" camp afterall. I don't know.
Does this affect how you think when it comes to the death penalty?
I don't believe in the death penalty because I believe retribution is a base and immoral motivation. There is nothing redemptive about retribution/vengeance.
The real relevance of these questions to the issue of the death penalty is this:
Punishing someone wrongly is always a very serious issue and we should go to great lengths to avoid it, but when the punishment in question is the death penalty we should be as near certain as possible that the person is guilty. This is because if we later discover that the person we executed was not guilty, there is no longer anything that can be done in terms of reparation. We can never give a person back the years they have spent behind bars, but financial compensation and an apology (and other things?) can be given in these cases, but the best we could do where the death penalty has been used is to give these things to the family.
This is, I think, the most powerful argument against the death penalty: we can never be so certain of a person's guilt that we would be justified in killing them and risking not being able to make any sort reparation should we turn out to be wrong.
Interestingly, many of the people who use this argument seem unaware that similar considerations are relevant to issues of euthanasia (physician assisted suicide) and abortion. In many such cases we may be unsure (at least naturalistically speaking if not generally) whether the person to be killed is really a "person" in the ethically relevant sense ... but the doubt itself should give us reason for pause, just as does the any doubt about a person's guilt in the death penalty case. In both cases a very serious risk is being taken.
Could the same argument be relevant to, say, the war in Iraq. Something the American president and the British Prime Minister should have thought about before invading?
The issue here is a really a basic element of theories of choice. One on the relevant metrics in decision making is "expected values", which has to do with assessing both the value and the probability of some outcome.
If we can put numbers to each of these things for all the possible possible (and mutually exclusive) outcomes, then we will then multiply probabilities by values and sum all the results. This sum will give you the expected value of the action in question.
Suppose we have available two actions A and B, each having three possible outcomes X, Y and Z. The value of outcomes X, Y and Z is fixed across all possible outcomes. For the sake of argument, let's suppose these values are
X: 0 Y: 200 Z: 1000 (Negative values will work too.)
On A and B the probabilities of these outcomes we'll assume to be
On A: X: 0.3 Y: 0.1 Z: 0.6
On B: X: 0.0 Y: 0.5 Z: 0.5
Applying our decision making theory:
On A: For X: (0.3 x 0)+ For Y: (0.1 x 200)+ For Z: (0.6 x 1000) Total: 620
On B: For X: (0.0 x 0)+ For Y: (0.5 x 200)+ For Z: (0.5 x 1000) Total: 600
On these assumptions the right action, according the this theory of choice, is action A as it has an expected utility of 620 against an expected utility of 600 for action B.
Now make up your own cases, using your own probabilities and values (with more than two choices or more than/less than three possible outcomes if you like). You could use:
A = Invade Iraq B = Don't Invade Iraq X = Lots of suffering and we're responsible Y = Lots of suffering and we're not responsible Z = Some suffering, but not lots.
A = Keep child B = Have abortion X = Unborn has full rights Y = Unborn has some rights Z = Unborn has no rights
A = Don't use death penalty B = Use Death penalty X = Verdict is wrong Y = Verdict is unsound Z = Verdict is right
The values/probabilities I've used above are not those I'd use for these scenarios, but you could imagine some people thinking along each of these lines.
For completeness, I should also point out that Pascal's wager uses a version of this theory of decision making. In that case there are (to put it mildly) some serious disagreements about how to enumerate the possible actions and outcomes, and one they are enumerated, how to assess the probabilities and values of those outcomes.
6 comments:
Is it better for a hundred guilty criminals to go free than for an innocent person to be wrongfully convicted and deprived of their liberty?
No.
Why do you think this?
For utilitarian reasons (not that I'm a utilitarian). Depriving someone of her liberty is a very serious offense. There should be a lot of room for judicial appeal, etc., but it is in society's best interest to prevent the harm perpetrated by the hundreds of wrongly released criminals.
I do believe, though, that standards should be high for guilty verdicts (e.g. innocent until proven guilty, burden on the prosecution, etc.), so maybe this does put me in the "yes" camp afterall. I don't know.
Does this affect how you think when it comes to the death penalty?
I don't believe in the death penalty because I believe retribution is a base and immoral motivation. There is nothing redemptive about retribution/vengeance.
Is it better for a hundred guilty criminals to go free than for an innocent person to be wrongfully convicted and deprived of their liberty?
I would not put a quantitative value ("a hundred") upon it but in general, yes.
Why do you think this?
Justice is not served by injustice.
Does this affect how you think when it comes to the death penalty?
Yes, though it is not my primary argument against it.
The real relevance of these questions to the issue of the death penalty is this:
Punishing someone wrongly is always a very serious issue and we should go to great lengths to avoid it, but when the punishment in question is the death penalty we should be as near certain as possible that the person is guilty. This is because if we later discover that the person we executed was not guilty, there is no longer anything that can be done in terms of reparation. We can never give a person back the years they have spent behind bars, but financial compensation and an apology (and other things?) can be given in these cases, but the best we could do where the death penalty has been used is to give these things to the family.
This is, I think, the most powerful argument against the death penalty: we can never be so certain of a person's guilt that we would be justified in killing them and risking not being able to make any sort reparation should we turn out to be wrong.
Interestingly, many of the people who use this argument seem unaware that similar considerations are relevant to issues of euthanasia (physician assisted suicide) and abortion. In many such cases we may be unsure (at least naturalistically speaking if not generally) whether the person to be killed is really a "person" in the ethically relevant sense ... but the doubt itself should give us reason for pause, just as does the any doubt about a person's guilt in the death penalty case. In both cases a very serious risk is being taken.
Steve
Steve,
Very interesting remark about abortion/euthanasia. I'd never quite made that connection before.
Could the same argument be relevant to, say, the war in Iraq. Something the American president and the British Prime Minister should have thought about before invading?
The issue here is a really a basic element of theories of choice. One on the relevant metrics in decision making is "expected values", which has to do with assessing both the value and the probability of some outcome.
If we can put numbers to each of these things for all the possible possible (and mutually exclusive) outcomes, then we will then multiply probabilities by values and sum all the results. This sum will give you the expected value of the action in question.
Suppose we have available two actions A and B, each having three possible outcomes X, Y and Z. The value of outcomes X, Y and Z is fixed across all possible outcomes. For the sake of argument, let's suppose these values are
X: 0
Y: 200
Z: 1000
(Negative values will work too.)
On A and B the probabilities of these outcomes we'll assume to be
On A:
X: 0.3
Y: 0.1
Z: 0.6
On B:
X: 0.0
Y: 0.5
Z: 0.5
Applying our decision making theory:
On A:
For X: (0.3 x 0)+
For Y: (0.1 x 200)+
For Z: (0.6 x 1000)
Total: 620
On B:
For X: (0.0 x 0)+
For Y: (0.5 x 200)+
For Z: (0.5 x 1000)
Total: 600
On these assumptions the right action, according the this theory of choice, is action A as it has an expected utility of 620 against an expected utility of 600 for action B.
Now make up your own cases, using your own probabilities and values (with more than two choices or more than/less than three possible outcomes if you like). You could use:
A = Invade Iraq
B = Don't Invade Iraq
X = Lots of suffering and we're responsible
Y = Lots of suffering and we're not responsible
Z = Some suffering, but not lots.
A = Keep child
B = Have abortion
X = Unborn has full rights
Y = Unborn has some rights
Z = Unborn has no rights
A = Don't use death penalty
B = Use Death penalty
X = Verdict is wrong
Y = Verdict is unsound
Z = Verdict is right
The values/probabilities I've used above are not those I'd use for these scenarios, but you could imagine some people thinking along each of these lines.
For completeness, I should also point out that Pascal's wager uses a version of this theory of decision making. In that case there are (to put it mildly) some serious disagreements about how to enumerate the possible actions and outcomes, and one they are enumerated, how to assess the probabilities and values of those outcomes.
Happy case constructing.
Steve
Post a Comment