Morning Spam

[Undiscussed Horrific Abuse, One Victim of Many]

--------------
A number of messages, in which everything became good, did not reach
the list. Retain this information so as to combine the messages.

---------------

Mind-Controlled Experimentee's Plan for Stopping Boss

"You're either too mind controlled to learn stuff, and you have AI
that can learn stuff, or you're free to learn yourself. Either way,
you can recruit more people."

"Treat the people and the AI as interchangeable. You keep studying the
way AIs interact in conflict: _when they can mutate themselves and
each other_ -- until you see the paths that work in the environment."

"This means 'running simulations' -- i.e. effectively planning -- for
the things and impacts that are actually going on."

"You don't do stupid stuff! When a problem happens, you get more meta."

"You don't run ridiculous experiments on huge swaths of data! You
simulate better and better."

"If your AI is too clumsy, it can study itself to improve."

"This is basic thinking! Everything on earth does this."