"Cracking previous digital cash systems fell into the class of "hire a hacker" difficulty problems"
David Chaum would disagree with you. Cryptographers have published a very large volume of research on how to create provably secure digital cash.
"Right, the person/group who published what may be the most significant usage case of applied cryptography in human history is probably not educated in cryptography"
So you think that someone who made no mention of any of the widely studied notions of security, any of the previous work in multiparty computation, any of the previous work in digital cash, or even the fact that allowing polynomial time attackers is a significant weakness in the design was well-educated in cryptography?
"Have you even read the whitepaper?"
The Bitcoin whitepaper reads like the work of someone who took an undergrad course in cryptography and spent a summer thinking about decentralized payments. Where is the notion of security even clearly defined -- what does it even mean to prevent double spending? What does it mean to control "more CPU power" than an attacker?
Compare the Bitcoin paper to this (just a random result from Google):
http://eprint.iacr.org/2007/148.pdf
Look at how security is defined in Section 3.2; that is typical of cryptography research in this day and age. Those are definitions that can be reasoned about in a rigorous way, without hand waving, without poorly defined concepts like "CPU power," and without having to analyze any specific attack strategy (i.e. the attacker is a "black box"). Where in the Bitcoin whitepaper do you see that sort of analysis, or even a hint that such an analysis is possible, desirable, or commonly used?
The security analysis presented in the Bitcoin whitepaper assumes a specific attack strategy: the attacker will try to compute a longer block chain than the rest of the network. Ignoring for the moment that this can be done in polynomial time, what reason is there to believe that no other attack strategy is possible? Maybe the attacker can adaptively send malicious messages to particular nodes in the network to break the security of the system. In multiparty computation, security analysis is usually based on the "simulation paradigm" -- a model of security based on showing that no polynomial time attack has a higher probability of success in the real protocol as it would in a simulation of the real protocol where a trusted party does all the work (which rules out all polynomial time attacks). That is a widely accepted notion of security, and the cryptographers who first published it recently won a Turing award; while other models may be valid, one would have expected an experienced cryptographer to have at least mentioned that they were using a different model (and perhaps justify their approach).
The whitepaper also includes statements that suggest a profound lack of knowledge of the decades of previous work in the field of digital cash. There are claims that no system with a central authority can possibly allow irreversible transactions, but that is untrue -- cryptographers discovered systems that support offline payments, in a model where the bank takes some action when double spending is detected (and in some systems, the bank can produce a proof that double spending occurred and that it was performed by a specific party, allowing for a trustworthy blacklist). Privacy is mentioned, and again the author makes claims about systems with central authorities that are simply not true -- Chaum's original work on digital cash was motivated by the need for private electronic payments, and that was the first problem to be solved in this field.
"When I say "Bitcoin is secure", that's by my standard"
What is your standard, and why should anyone else accept it?