edit: nevermind, did read the post before explanation was added. Guys, seriously, this stuff is confusing enough and its really hard to see what is the point of exercise even, when theres no actual hints on that. Furthermore, it is not mainstream knowledge of any kind. The updateless decision theory is specifically lesswrong thing. It’s not easy for those who write practical deciding agents to get the point of exercise with the newcomb’s , as in practical approach, there’s simply two ways to write down the payoffs in newcomb, which looks too much like typical problem formalizing problem descriptions written in English (and newcomb’s just doesn’t even look like a problem description that different people would understand exactly in same way)
It doesn’t prove U()=100, it proves [A()=1 implies U()=100]. Simultaneously having the proof that [A()=1 implies U()=5] allows you to disprove A()=1, which is perfectly fine, since actually A()=2.
edit: nevermind, did read the post before explanation was added. Guys, seriously, this stuff is confusing enough and its really hard to see what is the point of exercise even, when theres no actual hints on that. Furthermore, it is not mainstream knowledge of any kind. The updateless decision theory is specifically lesswrong thing. It’s not easy for those who write practical deciding agents to get the point of exercise with the newcomb’s , as in practical approach, there’s simply two ways to write down the payoffs in newcomb, which looks too much like typical problem formalizing problem descriptions written in English (and newcomb’s just doesn’t even look like a problem description that different people would understand exactly in same way)
It doesn’t prove U()=100, it proves [A()=1 implies U()=100]. Simultaneously having the proof that [A()=1 implies U()=5] allows you to disprove A()=1, which is perfectly fine, since actually A()=2.