Wow this is awesome. Some comments and questions:

• Spohn's decision theory does look very similar to Eliezer's, but Spohn couldn't give a good argument for the plausibility of rational cooperation in one-shot PD (he tried in the 2003 paper) because his didn't have the concepts of decision making as an algorithm, and of logical correlation between instances of such algorithms.
• The kind of AI cooperation discussed by Eliezer is not the type discussed as "program equilibrium". Instead "program equilibrium" is very similar (essentially the same?) as cousin_it's initial approach to AI cooperation, which he came up with in part due to dissatisfaction with Eliezer's approach. (cousin_it later moved on to "Lobian cooperation", which is closer to Eliezer's idea, and as far as anyone knows those results weren't previously discovered in academia.)
• In your research, did you fail to find previous academic work for some elements of the sequences? In other words, which other elements are not (known to be) expositions or reinventions of previous academic work?