Peer review is just a slightly more formal kind of debate, but debate doesn't work and isn't about finding truth.

Absolutely!

To go further, we must discover new mechanistic principles of truth-seeking. We should never ask: "What would it be good to know?" That road leads to alchemy.

Or engineering. Perhaps even aerodynamics (wouldn't it be good to know how to fly?). The main problem with alchemy was that it was too difficult to make any genuine progress.

I'm not sure I'm with you on this one. Wanting to know stuff is a rather important motivator for finding out stuff. Often you'll even end up finding out stuff completely different to the stuff you wanted to know.

Rather we should ask: "For what types of questions can the answers be evaluated by mechanistic principles?"

Lots of really boring things that I don't particularly care about. Also, some that I do care about. It'd be good to know those ones.