Please try to follow the "Norm One Principle." In particular, please keep in mind:

But the pattern I'm proposing doesn't restrict the "generation" process, the creative aspect which produces new hypotheses. It doesn't restrict the topics that can be discussed. It only restricts the criticism of those hypotheses, such that they are maximally useful to the source of the hypothesis.

You are encouraged to act more like a hypothesis-generator, and when you are a hypothesis-critic, you should aim to make sure your critique produces a useful "gradient" as if your output was equivalent to running you in backprop. (In machine learning, the discriminator gives its gradient to the generator, so the latter knows how to make the discriminator like it better.)

Instead of telling someone they are wrong, tell them what they would need to do to make themselves not wrong anymore.

Posts

Wikitag Contributions

Comments