Quote:
Originally Posted by browni3141
Are you just trying to be obnoxious now? If 'p' is the prior probability that Sklansky is God, the probability that Sklansky is God given the new evidence is 2p/(p+1), per Baye's Theorem. The chances ARE "basically doubled" for small 'p'. Do you think it's likely that Sklansky is God?
If you want to get into it, it has a lot more to do with the general approach to thinking about belief as being merely probabilistic in nature. Among other things, it's an error that one should assign a 50% probability to getting a single coin flip right under these types of circumstances and use that to update my beliefs. I would literally conclude nothing after watching one correct coin flip. I wouldn't allow for such nonsense in my sense of experimental design. (Note: This is highly asymmetric and I'm okay with that. That is, if he gets it wrong, I will conclude that he can't do it on the spot. But there's no reason to do this symmetrically since the weight of evidence isn't symmetric.)
The example that was given by the other poster is a clear example of how this actually isn't how we update our beliefs. You can assume whatever you want, but if those assumptions don't match how I (or most anyone) actually updates our beliefs, then his claim that we "must" do anything with our beliefs is clearly going to be wrong.
(DS also a history of him making comments of this type here and in SMP, and it's a well-trodden path of objections to his insistence of using this model for how people update their beliefs and what people "must" conclude as a result of his approach.)
Quote:
Kelvis seems right to me. Explain exactly which part of his argument is fallacious?
The all-or-nothing fallacy is basically that we have a binary classification scheme, and that all things must go into the same category. Specifically, because there are small probabilities of events happening by chance, that therefore all events *ARE* the result of random chance.
Here's the statement again:
Quote:
Originally Posted by Kelvis
Like yeah, that one time highly improbable event happened, but there are a lot of events that happen that are very low frequency. It's just numbers, if there are 10^20 things can happen with very low probability and some of them happen, that is expected. I myself witnessed a 10^-8 event in a board game throwing dice, but when there are so many events of low probability can happen that doesn't mean jack.
Yes, it's true that rare things happen as a result of random chance. It's not true that because a rare thing can happen by random chance that a rare thing that happened *IS* random chance. And pointing to other rare events happening as a result of chance does not imply that some particular event is the result of chance. There's little justification to support such a claim and it's almost certainly wrong. It stands as a possible explanation (I would even grant a "probable" explanation in some cases), but that only results from actually assigning various probabilities to things and doing such an analysis, and making other observations about the situation. And much of this is quite impossible to do with real life coincidences or situations.
It is logically fallacious to simply argue that because there are lots of events that we can therefore conclude nothing regarding low probability events.
(The addition of the anecdote "I've seen a rare random event and therefore <conclusion>" is just an added bonus of bad argumentation.)