Your Proving Too Much disproves too much: If we only allow reasoning steps that always work, we never get real-world knowledge beyond “I think, therefore I am.”. Some of these reasons for belief make their belief more likely to be true, and qualitatively that’s the best we can get.
(This is also incorrect, because considering a thinking you in a counterfactual makes sense. Many UDTish examples demonstrate that this principle doesn’t hold.)
Great critique! I’ve updated the post to show the purpose of this frame. We already have Bayesian updating for achieving beliefs to be more likely true. This framing of Proving Too Much is about differentiating useful reasons and those that claim”This reason leads to truth 100% of the time”.
There are lots of beliefs that people hold with 100% confidence for only 1 reason, and this is simply a way of jarring them out of that confidence. Encountering heuristics that account for less than 100% confidence will be covered in Finding Cruxes.
Your Proving Too Much disproves too much: If we only allow reasoning steps that always work, we never get real-world knowledge beyond “I think, therefore I am.”. Some of these reasons for belief make their belief more likely to be true, and qualitatively that’s the best we can get.
(This is also incorrect, because considering a thinking you in a counterfactual makes sense. Many UDTish examples demonstrate that this principle doesn’t hold.)
Great critique! I’ve updated the post to show the purpose of this frame. We already have Bayesian updating for achieving beliefs to be more likely true. This framing of Proving Too Much is about differentiating useful reasons and those that claim”This reason leads to truth 100% of the time”.
There are lots of beliefs that people hold with 100% confidence for only 1 reason, and this is simply a way of jarring them out of that confidence. Encountering heuristics that account for less than 100% confidence will be covered in Finding Cruxes.