What if reasoning models structured their thoughts as nested bullet points; branching and backtracking through different ways of solving a problem?

Comments