Date: Fri, 11 Sep 1998 12:17:55 -0700 To: "416720" From: "302146" Subject: Re: [MCG] Re: Knowledge is not an objective scalar Ed, Thank you for the clarification of the usage. I now understand the intent. I have had debates recently with some over the (non?)determinism of the universe, cosmology, and so I transitioned to applying this general definition to the discussion at hand. I guess my point was that for a _fixed_ set of past-present-AND-future events, the "output" of a NDFSM is completely determined. This was my intent when I said (to someone) I never intended to imply that if we were to roll-back the clock-and-universe, the MC-ware would produce a different result. "Parallel-Coupled" FSM works for me, or "Dynamically-Coupled" or "Dynamic-Transition". These would have produced less "discord". Overall, (ignoring potential implementation efficiencies) I fear that the discussion of NDFSMs tends to obscure the real issues. That is to say, a choice between a P-state NDFSM and a 2^P-state FSM does not speak to how MC-ware can do what it can(/will) do, above and beyond alternative contemporary certification systems. That is, I believe MC is a world above the limited certification regimes being established to date. This is rooted in its dynamic ability to retain and apply histories, as well as many other features. We should not give the impression that MC "works" because of some (magic or mysterious) qualities of NDFSMs. I am sure that there are varied ways to implement the precise functional features of MC. We should focus upon how MC obtains, retains, and treats data/objects in support of its unique functionalities. I can recall many students who believed that LISP would make AI possible. It may make the definition/description of some things easier, and also ease the construction of higher-level functionality, but there is nothing that can be written in LISP (or Prolog or...) that cannot be done in any other general purpose programming language. And yet the proselytizers of each language would tend to have you believe otherwise. This obscures the truth, and diverts discussion from the salient features of the subject matter (be it AI, MC, etc.) I do not consider myself an MC-expert, but my overall impression is that it represents an automated "intelligence" far surpassing the functionality envisioned for the limited certification systems deployed to date. Perhaps an exposition of its functionality in anthropomorphic terms would be illustrative, as a primer for the novice (I include myself). That is, I recall having been introduced at a young age to what a "computer" was, by analogy to a human servant who could store and retrieve data, and perform "pencil and paper" calculations, given (erzatz) high-level directives. If one attempted to describe MC-ware in terms of human agents, we could quickly get beyond arguing over implementation details, and focus instead on the what-when-and-where of information transfer, and how the directives given to these agents promotes security in ways not addressed by current systems. It may also help us to locate the "boundaries" of MC (hard as that may be ... it seems potentially boundless!) ___tony___ Tony Bartoletti LL SPI-NET GURU LL LL Computer Security Technology Center LL LL LL Lawrence Livermore National Lab LL LL LL PO Box 808, L - 303 LL LL LLLLLLLL Livermore, CA 94551-9900 LL LLLLLLLL email: azb@llnl.gov phone: 510-422-3881 LLLLLLLL