The part you're missing is the assumptino that your pre-judgements are the bigger picture.
I know you think you are being eminently reasonable, but it's really just another way of saying you intend to keep to those prejudices and not be distracted by treating individuals as individuals. That would be edge case thinking.
You're saying that my assumption is "Majority of the fat people are fat because of intake of food and not because of actual health problems." In this context, yes, that is exactly what my Bayesian priors (so to say) assume but it's not relevant to the point I'm trying to make. I've read things that seem to support this many times over the years so obviously that is what I would assume. If I am wrong, would you kindly care to educate me.
Actually no, don't. Because that is not the point. The point is (like I said in other comments) my gut feeling suggests that we as a society are moving towards optimizing for edge cases and that's not a good thing. (Immediate examples that come to mind are parent fearing children going out and playing, security theatre, crime television, crazy safety labels etc.)
I know you think you are being eminently reasonable, but it's really just another way of saying you intend to keep to those prejudices and not be distracted by treating individuals as individuals. That would be edge case thinking.