Research

The question I keep coming back to is how something decides what to do when it does not have enough information. That is my PhD, and it is also what the applied work keeps walking into from the other direction.

I try to test a claim in the situation it is supposed to hold in. Fourteen people reviewing their own lunches tell you more about a review app than any number of lab sessions, and a trust study means more when the robot is standing in the corridor while people answer.

Publishing is not the point, but it is a useful discipline. It forces the claim to survive a reader who has no reason to be generous about it, which is a harder test than any I would set myself.