re: Evolution in action
Not being able to discuss specifics about one's work is pretty common in engineering. On the commercial side companies want to protect their trade secrets (which are, in my experience, usually trivial and more a matter of corporate ego than substance). You might be able to tell that when I write about my eBay gig, I carefully craft my words to avoid giving away anything that eBay might consider proprietary. I even gave some thought to how to describe their server farm, since too much info could open them up to attack.
... system development. It's hard to describe just how badly the current way of doing things is broken.
I can believe it, based on examples of government software that I've seen over the years. NASA seems ahead of the rest, but even then that seems to define a standard of mediocrity. The problem isn't the skill level of government programmrs, I think it's just the ordinary pernicious influence of bureaucracy on work quality. It's got to be hard, in particular, to integrate code from different teams when bureaucratic politics encourages empire-building and separatism, and discourages cooperation.
Not that that behavior is the exclusive province of governmental organizations. Most commercial software organizations tie compensation to "metrics" like bug counts, which leads directly to playing politics with bugs--find somebody, anybody, you can reassign the bug to (preferably in another department to protect your boss's bonus); ping-pong the bugs endlessly in the hopes that they just disappear at the next release; horsetrade with your QA counterpart to reduce the severity level in exchange for future back-scratching; and on and on. You get more points for making bugs disappear than for fixing them.
So what else is new, really? The domain today is software, but human nature's been producing the same result for thousands of years. This is about as good as it gets with homo sapiens sapiens 1.0.