I agree with you. "Complex reasoning" without the guardrails of empirical tests and lived implementation is very error prone. The number of variables is gigan

I agree with you. "Complex reasoning" without the guardrails of empirical tests and lived implementation is very error prone. The number of variables is gigantic. In the armchair you can reduce them and tweak the remaining ones, and they will do whatever you like within the boundaries of your imagination. That method seems just not very predictive of what will really happen.