Which directly implies that financial systems that require massive investments of new software infrastructure to make them useful to the public are especially risky.

I should think that depends on if the massive investment is for paying hordes of typical competent programmers, or for paying fewer guys more like you.

My sense of the whole industry right now is that the demand for competent software security far, far outstrips the supply; Bank of America could want to build a gigantic new transaction processing infrastructure in a few years, but logistically would probably not be able to retain the talent required to secure it.

The major security advantage our legacy infrastructure has is that it's old, and its failings are well-understood and (from a risk management / loss mitigation perspective) mostly mitigated.

Bitcoin (or any other online currency or transaction infrastructure) has none of that.