You don’t compare harms and benefits of the same process. You compare properties of the alternative processes. Comparing harms with benefits of a single possible decision is a heuristic for comparing the decision with the default of doing nothing, but often there is no such default option.
(This remark is not relevant to the conversion, just a technical note on your usage of harms vs. benefits.)
Comparing harms with benefits of a single possible decision is a heuristic for comparing the decision with the default of doing nothing, but often there is no such default option.
There is another good reason to analyze one process into harms and benefits, even if we don’t know how to avoid taking them as a package. The reason is that we may someday figure out a way to modify the process so that we get the benefits with less of the harm. Such a subtle refinement of the process may be beyond us now, but we are more likely to gain that capability someday if we perform the analysis now. That is one way that we learn of the worthwhile technical problems to work on.
This analysis wouldn’t make sense if we thought that the concomitance of the harm and benefit were a law of nature or logic. Otherwise, it is always possible that we will figure out a way to separate them.
In the meantime, a side benefit of this analysis is that its results are ready made to plug into expected-utility calculations. I think that that is a good reason for comparing harms and benefits of the same process.
But then you won’t compare harms with benefits, you compare a particular harm, or a particular benefit, with its alternative counterpart achieved by changing the process. The error is in comparing different factors with each other, like comparing the price of a good with cost of delivery, not in paying attention to detail of all the different factors. So you do want to compile lists of harms and benefits, as instrumentally or morally relevant characteristics of a plan that you want to optimize, but not for the purpose of comparing them to each other.
So you do want to compile lists of harms and benefits, as instrumentally or morally relevant characteristics of a plan that you want to optimize, but not for the purpose of comparing them to each other.
By “comparing harms with benefits”, I mean computing the values of the benefits, harms, and U = (benefit) – (harm) that follow from an action. Of course, these values alone don’t determine whether you ought to take the action. You need to compare U to the analogous quantity for the other available actions first. And of course the quantities of (benefit) and (harm) have to be measured with respect to some arbitrary “zero”, which doesn’t have to be a “default” as this word is intuitively understood. You could set U itself equal to zero. The computation of U is only an intermediate step to action.
Nonetheless, these are all handy quantities to store, because you can cheaply store them and compare them to the analogous quantities for other actions, including ones that you haven’t thought of yet.
Of course. But I still can’t convince my mother that paying $10 for delivery of a $3 peg is better than driving all the way to a remote storehouse to get it yourself, even though the cost of the delivery is significantly greater than cost of the goods and hence feels “not worth it”.
It’s useful to hold all sorts of quantities in your mind while considering a decision, but it’s important to know what they’re for, so that they aid in accuracy of the decision, and not confuse you instead. To avoid this confusion, first thing is to know a reliable (if inefficient) methodology, and only then develop advanced tricks.
But I still can’t convince my mother that paying $10 for delivery of a $3 peg is better than driving all the way to a remote storehouse to get it yourself, even though the cost of the delivery is significantly greater than cost of the goods and hence feels “not worth it”.
I agree that that is a common and buggy way to think. I probably tend to do that myself when I’m not careful, so I concede that it is a problem. But your example has someone comparing part of the cost of an action with another part of the cost of that same action. I’m talking about comparing the cost with the benefit of the same action. Are you saying that even doing that could put one at more risk of making the mistaken comparison that you describe?
At any rate, that wasn’t the kind of comparison that lukeprog was doing. Could you elaborate on the kinds of mistakes that you see flowing from what he said? One possible confusion is using the sign of (benefit) – (cost) of an action (measured in lives saved, say) to decide whether to do it. Do you see others?
That kind of comparison just completely ignores opportunity costs, so it will result in mistakes any time they are significant.
Making the comparison is not the last step before decision. The comparison itself ignores opportunity costs, but it doesn’t keep you from going on to perform an opportunity-cost check. The output of the comparison can then be combined with the output of the opportunity-cost check to determine a decision.
True.
Do you think it’s likely that the benefits of AIDS denial, homeopathy, exorcism, and faith healing come close to outweighing the harms?
Note that I also did not list possible benefits of anti-vaccination.
You don’t compare harms and benefits of the same process. You compare properties of the alternative processes. Comparing harms with benefits of a single possible decision is a heuristic for comparing the decision with the default of doing nothing, but often there is no such default option.
(This remark is not relevant to the conversion, just a technical note on your usage of harms vs. benefits.)
There is another good reason to analyze one process into harms and benefits, even if we don’t know how to avoid taking them as a package. The reason is that we may someday figure out a way to modify the process so that we get the benefits with less of the harm. Such a subtle refinement of the process may be beyond us now, but we are more likely to gain that capability someday if we perform the analysis now. That is one way that we learn of the worthwhile technical problems to work on.
This analysis wouldn’t make sense if we thought that the concomitance of the harm and benefit were a law of nature or logic. Otherwise, it is always possible that we will figure out a way to separate them.
In the meantime, a side benefit of this analysis is that its results are ready made to plug into expected-utility calculations. I think that that is a good reason for comparing harms and benefits of the same process.
But then you won’t compare harms with benefits, you compare a particular harm, or a particular benefit, with its alternative counterpart achieved by changing the process. The error is in comparing different factors with each other, like comparing the price of a good with cost of delivery, not in paying attention to detail of all the different factors. So you do want to compile lists of harms and benefits, as instrumentally or morally relevant characteristics of a plan that you want to optimize, but not for the purpose of comparing them to each other.
By “comparing harms with benefits”, I mean computing the values of the benefits, harms, and U = (benefit) – (harm) that follow from an action. Of course, these values alone don’t determine whether you ought to take the action. You need to compare U to the analogous quantity for the other available actions first. And of course the quantities of (benefit) and (harm) have to be measured with respect to some arbitrary “zero”, which doesn’t have to be a “default” as this word is intuitively understood. You could set U itself equal to zero. The computation of U is only an intermediate step to action.
Nonetheless, these are all handy quantities to store, because you can cheaply store them and compare them to the analogous quantities for other actions, including ones that you haven’t thought of yet.
Of course. But I still can’t convince my mother that paying $10 for delivery of a $3 peg is better than driving all the way to a remote storehouse to get it yourself, even though the cost of the delivery is significantly greater than cost of the goods and hence feels “not worth it”.
It’s useful to hold all sorts of quantities in your mind while considering a decision, but it’s important to know what they’re for, so that they aid in accuracy of the decision, and not confuse you instead. To avoid this confusion, first thing is to know a reliable (if inefficient) methodology, and only then develop advanced tricks.
I agree that that is a common and buggy way to think. I probably tend to do that myself when I’m not careful, so I concede that it is a problem. But your example has someone comparing part of the cost of an action with another part of the cost of that same action. I’m talking about comparing the cost with the benefit of the same action. Are you saying that even doing that could put one at more risk of making the mistaken comparison that you describe?
At any rate, that wasn’t the kind of comparison that lukeprog was doing. Could you elaborate on the kinds of mistakes that you see flowing from what he said? One possible confusion is using the sign of (benefit) – (cost) of an action (measured in lives saved, say) to decide whether to do it. Do you see others?
That kind of comparison just completely ignores opportunity costs, so it will result in mistakes any time they are significant.
Making the comparison is not the last step before decision. The comparison itself ignores opportunity costs, but it doesn’t keep you from going on to perform an opportunity-cost check. The output of the comparison can then be combined with the output of the opportunity-cost check to determine a decision.