Lambda the Ultimate

I just finished reading the book.

At first, I opened the book, and it said, "First and foremost, I wish to thank Basile Starynkevitch ..." -- Immediately I thought you just pulled a snow job on me :) I sort of figured you had an advance review of it, since Amazon.com said the book ships June 15th and you posted this June 10th.

But this is a really nice book, aside from the strange side discussion about robot sexuality.

It is exactly what you say it is: a great explanation of reflective software systems. I found it far more approachable than my other two books on the subject: Advances in Object-Oriented Metalevel Architectures and Reflection eds. Zimmerman et al and The Art of the Meta-Object Protocol by Kiczales et al. Unlike these other two books, the pacing is consistent, is clearly written by a single author, and uses surprisingly clear and crisp English.

Thanks for the recommendation. At some point Lambda the Ultimate should probably have a discussion of one of the papers Pitrat mentions... the one by Luc Steels about robots that build languages.

I spend many hours reading J.Pitrat drafts, and J.Pitrat gave me a sample of his book (in april 2009).

In France you can find (or order) this book since may 2009.

What seems strange (and sad) to me is that many of the ideas in this book are quite old (e.g. D.Lenat papers on Eurisko, or J.Laurière's papers on Alice, and of course J.Pitrat's papers - most of these papers have been published in the 1970s or 80s) but most are forgotten.

However, there might be a (social) explanation. Today's Computer Science papers are usually more about theory than about implementations, and any reflexive system should be implemented before you publish on it, and (compilatory) reflection is necessarily complex, because you need to build a fairly complete code generation (& introspection) system. Producing theoretical results take less coding (and coding is painful) than coding reflective systems.