I'm the kind of person that never fully believes anything, especially not myself... I'm sure there are many cases were continual deployment is a benefit, but also cases, where it is bad thing.
I use the Chrome nightlies, and every couple of weeks it'll break and I can't use it at all, other times it'll break and some features don't work (like, the address bar stopped working once, and the menu bar stopped working another time). I know that Chrome has very good automated tests, and if you look at the commit log, many commits include unit tests, and most of the bug fixes come with unit tests. But stuff still falls through the cracks.
With the Chrome Canary builds, fixes for badly breaking changes have normally come out in a few hours (a very good thing).
I think with mission critical - especially life dependent systems - continual compiling, testing (unit testing, + meaty testing). Often, even if there are no known bugs, testing needs to continue for a long period of time before deployment - and even then there will be bugs after the fact.
But I guess, the whole premise doesn't apply to most mission critical (not the finance term - because finance is just a big arse raping competition) systems, because these are usually embedded in firmware, and it is EXPENSIVE to deploy.
I will therefore agree conditionally, that continual deployment is a good thing for all (99%) of web based systems - because no matter how much separation of concerns you have - a web system is content based, and even the business logic is an integral part of that content. And content needs to be updated. Fast.
Big finance simply fails at software (I could get specific here, but won't), and I think its part of the whole financial way of thinking. Finance companies tend to have slow releases, and yet, their modern systems are riddled with bugs.
The bugs are generally never fixed and teams of operational personnel exist purely to fix the problems that the systems cause.
Software developers COST A LOT. If you can save money today by firing 5 programmers (or get them working on something else), you only need to hire ONE person to fix the problems in the system. (Add one person every second month, and it soon become 20 people - but that's the FUTURE, that's not NOW). We need to make money NOW.
Typical short term finance company thinking.
I think finance is the area that could most benefit from your methodology.