The Problem of Old Evidence (Bayesian Epistemology)

[@AhsimNreiziev]

You claim that, in Bayesian Epistemology, one is _"only allowed to change [one's] beliefs when our degree of belief in E changes"_. I don't recall that being stated in any of your previous videos on Bayesian Epistemology, but even if it was, I don't see why Bayesians would hold that. So, I ask you, like any Scientist might, for your sources. Back up your claim and provide a link to an article on Bayesian Epistemology, preferably written by a Bayesian Epistemologist, that asserts the claim you made that Bayesian Epistemology holds that only a change in the actual Evidence E can change our belief in the Hypotheses H.

[@CarneadesOfCyrene]

Ahsim Nreiziev The reason that we cannot change our degree of belief is that we would be dutch book irrational if we did. I can do something better than an article, I can give you an argument (as most philosophers would). Based on the rules of Bayesian Epistemology, if you discovered that H>E, you would be in a diachronic dutch book if you changed your belief in H&E. Take the example in the video. At time t1 you have a .96 degree of belief in E, a .5 degree of belief in ~H given that E, and a .02 degree of belief in H and ~E. But at time t2 (remember that diachronic means that there are two times involved) you have a 0 degree of belief in H and ~E (because you have discovered that H implies E), but still a .48 degree of belief in ~H and E (since there is nothing to change your mind about this) and therefore have a .51 (approx. 50/98 is exact) degree of belief in H given that E. If you hold all of these degrees of belief, you are committed to a contradiction according to Bayesian Epistemology, just as surely as if you believed p and ~p. Here's why: Let's say I sell you a wager at time t1. I sell you a wager that pays $1.0 for $0.96 if E is the case. Your degree of belief in E is .96 so you accept. I then sell you another wager that pays $10 for $5 if ~H is the case given that E. If E is not the case, this wager is called off. At time t1 you have a .5 degree of belief in ~H given that E (48/96=.5) so you accept. Finally at t2 I sell you a wager that pays $10 for $5.10 if H is the case given that E. If E is not the case the wager is called off. Since you degree of belief in H given that E is now .51 (in fact slightly higher) you accept. Now, no matter what happens, you lose money. ~E: You lose $0.96 on the first wager, the other two are called off. So you have a net $0.04 loss. E, H: You win $0.04 on the first wager, you lose $5 on the second wager, and you win $4.90 on the third wager. You have a net loss of $0.06. E, ~H; You win $0.04 on the first wager, you win $5 on the second wager, but you lose $5.10 on the third wager. You have a net loss of $0.06. If you have watched my video on the dutch book arguments (kzbin.info/www/bejne/g5DEeqRjnL-efc0) you will be aware that if you are in a diachronic dutch book, according to Bayesian Epistemology, you are irrational. Therefore one cannot remain a rational Bayesian and change one's degree of belief in E based on the implication H>E. If you give up on this, you give up the whole of Bayesian Epistemology. Check out the SEP article on the subject for more information, if you really want an article.

[@MeltedCheesefondueGruyere]

I don't see the problem here. Let M="vehicle exhaust causes the world to get hotter" (I'm assuming we take the fact that there is a lot of vehicle exhaust as a given). Then we get some evidence for M. This causes both E and H to increase (unless they are already maximised). Here M is taking the evidential role that the logical H->E is taking, and, sure enough, finding evidence for M increases H. But how does the high E come in? A low E makes M very unlikely, so an increase in M requires more evidence. So, basically, it seems that having a large E allows M to update more easily, which means that H can increase more easily on evidence of M.

[@KabeloMoiloa]

BTW, this is a question unrelated to epistemology, but if you are familiar with partially ordered sets or category theory, I would love it if you were to answer this question: philosophy.stackexchange.com/questions/22053/is-there-a-poset-based-category-theoretical-definition-of-god

[@CarneadesOfCyrene]

Kabelo Moiloa I have only an elementary understanding of the concepts, but I can give it a shot here. First I assume that the author of the post means omnipotence (being able to do things) not omniscience. It seems to me that there are two things that we could be talking about when we are talking about when we say omnipotence, something that is more powerful than anything else (1), or something that is able to do anything (2). In the former case it is necessary that there exists some thing that is omnipotent (1), because it is just the most powerful thing. However in the latter case there is not necessarily something that is omnipotent (2) as it might be the case that the thing that is omnipotent (1) in that world is not able to do everything. The question that is posted seems to offer a definition of the first kind of omnipotence, but the theist is looking for the second kind of omnipotence, but I'm not sure that is what is being said. What do you think of such a response?

[@KabeloMoiloa]

Carneades.org Actually, I wrote that question. Now that I think of it, yeah theists want some kind of an "all powerful," entity not a "most powerful" entity. Ugh, oh well at least I tried :(

[@josiaphus]

Straight witchcraft