I'll confess, I'm not a big fan of academia. I mean education is a great thing as long as it's taken in moderation and not too seriously. A classic example is the way academia approached artificial intelligence. They promised the moon and delivered...well...a lot of papers. The AI scam burned people so badly that by now it's pretty hard to get money for research that even uses the term AI.

And yet the study of complex artificial systems with an eye toward "intelligent" behavior is still a legitimate concern. Several years ago a chunk of money was available for this type of research and somehow I ended up reviewing proposals from universities.

I'd like to think my e-mail reprinted below summarizes what I found fairly well. Names are omitted to protect the guilty...

Bob,

It's hard to say anything nice but I'll try summarize the highlights of my opinions without resorting to profanity.

1.1 They give lip service to how control theory has improved things but ironically miss the central point of its failure. Control theory allowed us to design and build robust system that are more complex than we can understand. This was by modelling assumptions (chiefly linearization) and encapsulation (minimizing the number of inputs, outputs and parameters). This allowed us to build hierarchical systems. The interesting thing is that it only bought us so much. Encapsulation and hieararchical design can only buy us so much complexity before things start going to hell and acting in unpredictable (and undesirable) ways. Software engineering is at the forefront of this problem (Which is why almost no modern software works completely as intended!)

1.2 "We are now attempting to build control systems for processes that are vastly more complex or that are to be operated much closer to their performance limits in order to achieve much greater efficiency of resource use." The examples that they talk about are ones that make the previous statement absurd. e.g.The problem with autonomous control of an automobile is not "efficiency" but working at all!

4.1 The discussion on the bottom of page 10 is pseudo-mathematical tripe intended to impress the natives. The last paragraph of this section seems to be stating that they've discovered hierarchical encapsulation.

4.2 The first paragraph on the top of page 12 (Finsler manifolds, differential geometry, etc.) appears to be intended to render the natives senseless.

The last paragraph of 4.2 talks about a "game theoretic approach" to controller design. Isn't this just multidimensional optimization by a spiffy and obscure name?

4.3 This section discusses "verification methodologies and tools" and is either the most incredible piece of garbage I've ever encountered or these guys are overdue a Nobel prize. The statement "if under given environment assumptions, every subsystem is guaranteed to perform correctly, then the compound system performs correctly" is obviously true but is the most absurd mission statement I have ever seen. Rephrased in English it might as well say "if the system design is perfect, and our understanding of the universe is perfect, then the system will work perfectly." Wow!

The top of page 13 talks about piece-wise linear models and being analyzable. That's nice but that's one of the many standard non-linear controls modeling and analysis tricks. The problem is they don't always work. The whole problem with even simple single input/single output control systems is that the damn systems and their environments are NEVER what you think they are. This is why increasingly strong non-linearities usually require increasingly paranoid and cautious designs to keep the system from doing something naughty.

On page 13, paragraph 3 they mention they have discovered phase domain analysis is good. Check any non-linear controls book written in the fifties to discover how leading edge this thought really is.

Page 13 paragraph 5 is the most amazing techno BS I have ever read. Verification diagrams, easy to construct proofs, intuitive proofs of complex hierarchical systems. If this is true these guys are gods who put Newton, Gauss, Euler and Von Neumann to shame. There is nothing coherent I can say about this section as its claims are so outre as to leave my mind boggled by their audacity.