The article lampshades, but imo doesn't really address, the fact that it doesn't really address the title? Like, it brings up the "you haven't said what probab

The article lampshades, but imo doesn't really address, the fact that it doesn't really address the title? Like, it brings up the "you haven't said what probability can't do" response, and then kinda jukes by combining that with the "probability is just math and the math is fine", and responding primarily to the latter part. It brings up the "it would be unreasonable to use probability theory to plan a picnic" point. Sure, granted. But the title doesn't say "It isn't reasonable/feasible to compute explicit probabilities in all cases." , it says what probability _can't_ do. And the article does not demonstrate the existence of something that probability _can't_ do. The main content of the page might be, fine I guess (I'm having trouble finding it), but it doesn't address what the title says it addresses. What are the things that probability can't do? What I'm asking for, which is my initial interpretation of the title, is not merely a single example, but rather an explanation of a _class_ of things which "probability can't do". It doesn't have to be a complete characterization, but it should give a sufficient condition for something to be something that "probability can't do". If you have a single example which you can demonstrate that "probability can't do", you should be able to take that example, and then ask "what things could I change in this example without changing the validity of the argument?", which should then give you a broader class of things which "probability can't do". The article by nostalgabraist which this article links to, I like much more. Especially section 8. I saw another post on lesswrong semi-recently which talked about Jeffry updates, and Jeffrey Bolker's broader generalized probability theory ideas. That was pretty good I thought. The point about updating on an uncertain observation (the "Jeffry updates") part I especially thought was convincing. ___ So far, the real shortcomings in the base form of ("strong") bayesian probability that I see/understand are : 1) observations we make are not suddenly gaining perfect certainty of a proposition, but rather some sense experiences. We may not be entirely sure of what we are seeing. 2) we have bounded computation and memory, and as such: 2.i) we cannot enumerate all the possibly hypotheses that could be relevant 2.ii) we cannot process all of the logical implications even just between the different hypotheses that we do consider