By Aumann's agreement theorem (and some related/inferred ideas), two rationalists can't agree to disagree.

Isn't this claim obviously false? Doesn't agreement map directly to the halting problem, which is undecidable?