Imagine you’re debugging a program that crashes. Your friend Bob is smarter than you and has already figured it out, so for any input data he can tell you in advance whether it will crash or not. You formulate a hypothesis: “the crash happens within this function”. It does not yet allow you to predict crashes accurately, because there’s a variety of things that could go wrong within this particular function. Now you tell Bob, “I think it crashes somewhere in this function”, and Bob nods, “Yeah”. This means Bob belongs to the subset of possible predictors that satisfy your hypothesis :-)
But what predictions Bob should make to satisfy Islam seems to be a more difficult question.
Edit: I don’t think you’re disingenuous. Asking the right questions can help all participants achieve clarity where there was none. It’s an art that I’m slowly learning here on LW—most of the results in my posts come out of discussions.
Imagine you’re debugging a program that crashes. Your friend Bob is smarter than you and has already figured it out, so for any input data he can tell you in advance whether it will crash or not. You formulate a hypothesis: “the crash happens within this function”. It does not yet allow you to predict crashes accurately, because there’s a variety of things that could go wrong within this particular function. Now you tell Bob, “I think it crashes somewhere in this function”, and Bob nods, “Yeah”. This means Bob belongs to the subset of possible predictors that satisfy your hypothesis :-)
But what predictions Bob should make to satisfy Islam seems to be a more difficult question.
Edit: I don’t think you’re disingenuous. Asking the right questions can help all participants achieve clarity where there was none. It’s an art that I’m slowly learning here on LW—most of the results in my posts come out of discussions.