In detail, an inference strategy has to deal with the following three kinds of indeterminisms: 1. the selection of the initial path, 2. the selection of the next branch B to be expanded, 3. the selection of the next linking step to be performed on B.

Clearly I have a duty to say why design strikes me as a nonsilly hypothesis, and even as a simple and attractive one.

Never give the least touch with your pencil till you have well examined your design.

There's no use in complaining. It doesn't change anything.