Eliezer,
The seed AI concept has been rattling around in my head for a while. What's buggin me is the concept of environment. I asked you what environment you expected would be necessary for the AI a while back, and you said that the environment would be the AI's code itself (I think - I could have missed your point, please have patience with a mere mortal).
Absolutely, this must be the core of the environment. But there must be more than this, mustn't there? For an AI to optimise itself, there must be some definition of optimal, which implies a frame of reference. I think that frame must be external to the code, because the idea of optimising you code to make you better at optimising your code has an unfortunately circular and empty feel to it.
You talk about domdules to perform various functions in "Coding a Transhuman AI" - do you envision trying to deal with the whole real world from the get go? Or is there some initial minimal closed environment in which to start the AI, which the AI can begin in, and upon which you can expand when the AI gets comfortable?
I would be in favour of trying the approach of designing a minimal simulated environment (including the code for the AI itself), with option to expand (aiming to eventually include a connection 'out', into the net).
Does this sound something like what you are envisaging? Emlyn, Computer Knackerer