Can Blindsight-style Scramblers employ anthropic reasoning?
Not sure what Scramblers are exactly, but in 2001 I came up with some ideas about how AIs could do something like anthropic reasoning, without being sentient (which eventually incorporated into UDT). Here's the original post I wrote, which was titled no need for anthropic reasoning .
If it's worth saying, but not worth its own post (even in Discussion), then it goes here.