I think that a rational agent would donate the $500 eventually either way because the utility value of a $500 contribution would be greater than that of a $0 contribution, if the matching $500 was not forthcoming. Thus, the precommitment to withhold the donation if it is not matched seems to be a bluff (for even if the agent reported that he had not donated the money, he could do so privately without fear of exposure) Therefore, it seems to me that the matching arrangement is a device designed to convince irrational agents, because the matcher’s contribution does not affect the amount of the original donor’s contribution.
He may actually refrain from donating, by the reasoning that such offers would work iff someone deems them reasonable and that person is more likely to deem it reasonable if he does, by TDT/UDT. I could see myself doing such a thing.
But whether he does or doesn’t donate does not affect how such offers are responded to in the future, since he is free to lie without fear of exposure. Given such, it seems that he should always maximize utility by donating.
Are you familiar with UDT? There’s a lot about it written on this site. It’s complex and non-intuitive, but fascinating and a real conceptual advance. You can start by reading about http://wiki.lesswrong.com/wiki/Counterfactual_mugging . In general, decision theory is weird, much weirder than you’d expect.
I think that a rational agent would donate the $500 eventually either way because the utility value of a $500 contribution would be greater than that of a $0 contribution, if the matching $500 was not forthcoming. Thus, the precommitment to withhold the donation if it is not matched seems to be a bluff (for even if the agent reported that he had not donated the money, he could do so privately without fear of exposure) Therefore, it seems to me that the matching arrangement is a device designed to convince irrational agents, because the matcher’s contribution does not affect the amount of the original donor’s contribution.
Am I missing something?
He may actually refrain from donating, by the reasoning that such offers would work iff someone deems them reasonable and that person is more likely to deem it reasonable if he does, by TDT/UDT. I could see myself doing such a thing.
But whether he does or doesn’t donate does not affect how such offers are responded to in the future, since he is free to lie without fear of exposure. Given such, it seems that he should always maximize utility by donating.
Future offers do not matter. His precommitment not to donate if others do not acausally effects how this offer is responded to.
I’m not sure I understand what you mean. Would you mind explaining?
Are you familiar with UDT? There’s a lot about it written on this site. It’s complex and non-intuitive, but fascinating and a real conceptual advance. You can start by reading about http://wiki.lesswrong.com/wiki/Counterfactual_mugging . In general, decision theory is weird, much weirder than you’d expect.
I’ve read some of the posts on Newcomblike problems, but am not very familiar with UDT. I’ll take a look—thanks for the link.