"most attempts will fail, but if we are allowed 3^^^3 attempts at a significant technological gain, we will expect to advance." I think you need some sort of prior distribution for number of attempts to make it analogous to SIA doomsday.

And you need a similar prior distribution for Katja_Grace's SIA argument, which makes it underdetermined as well. (My "no anthropic reasoning" paradigm is really starting to pan out!)

James Miller is the author of this top-level post.