Chuck, also see (apart from whatever else you are pursuing to answer the above question) the CTM = Computational theory of mind, and cognitivism. This is a theory proposed by Jerry Fodor and others (not only are mental operations akin to computations, but also their outcome, beliefs, are truth-preserving for this, and not empiricist, reasons) which follows naturally out of some of the work done on computability in the earlier half of the 20th. What Gödel's (and Turing's) findings imply relies heavily on the validity of this view.
My apologies to you on the lack of response on your question regarding NAT/DSL/etc. Fortunately, later messages suggest you have the problem under control. I was going to respond to point out that kernel recompiles are not at all a matter to worry about and you should definitely follow that route. I assume you also solved your problem of routing from your old box through your new one? Note that NATed clients are [mostly] unaware that they are being NATed, so all they require is the default router configuration. You could supply that information either through DHCP (as most HW NAT boxes do) or, as you did, hard code it in your old box's routing table.
--ravi