The way that humans have traditionally solved these problems is with hierarchical organizations that distribute responsibility. I think the reason that computer systems aren't reliable isn't because they are complicated, its because they are dynamic - the thing that you want and the right way to make it are constantly changing, so everything must be constantly rebuilt, and so there is no opportunity to work the bugs out. Relatively static bits of code that have remained so over long periods of time have reached high levels of reliability and security.

As in all things, if you want reliability and security you must sacrifice being on the bleeding edge - you must slow the pace of change. This isn't new or unique to computers. The Catholic Church has known this for centuries. Presently, in computing, the value of change exceeds the value of reliability. If I can run three times as fast, I'm still better off even if I fall twice as much.

RE: trust and consequences