nyan_sandwich comments on Open thread, September 2-8, 2013 - Less Wrong
You are viewing a comment permalink. View the original post to see all comments and the full post content.
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (376)
I found this interesting post over at lambda the ultimate about constructing a provably total (terminating) self-compiler. It looked quite similar to some of the stuff MIRI has been doing with the Tiling Agents thing. Maybe someone with more math background can check it out and see if there are any ideas to be shared?
The post: Total Self-Compiler via Superstitious Logics
This is the same basic idea as Benja's Parametric Polymorphism, with N in the post corresponding to kappa from parametric polymorphism.
The "superstition" is:
And from the section in Tiling Agents about parametric polymorphism (recommended if you want to learn about parametric polymorphism):
Anyway, it's interesting that someone else has a very similar idea for this kind of problem. But as mentioned in Tiling Agents, the "superstitious belief" seems like a bad idea for an epistemically rational agent.
It is neat that this problem is coming up elsewhere. It reminds me that MIRI's work could be relevant to people working in other sub-fields of math, which is a good sign and a good opportunity.