Basically, it seems like the government sourcing process has shrunk the pool of potential bidders too small to provide a properly competitive marketplace for software for USGov customers.
And don't think for a second that this is unintentional.
I think it's actually achieved that goal. It comes at the expense of competence, however, as those with the best lawyers (not best developers) win contracts.
Okay, what if I do?
If you are anything but cynical regarding the government contract bidding process, you're asking to be made a fool of.
But put that in the government, and you're potentially disenfranchising 5% of the citizenry which is not only politically foolish but quite likely illegal, given constitutional requirements about equal treatment and so forth. If you have to provide universal service of some kind, then your marginal costs go way up. Suppose 99% correctness were the acceptable standard, such that Social Security, Medicare, VA etc. could just ditch that 1% of claimants that caused the most administrative problems; the administrative savings would probably be far more than 1%, I'm guessing more like 10-15% because once the administrative burden of dealing with a given citizen rose above 1 or 2 standard deviations you could just dump them from the system and cut your losses.
http://www.theatlantic.com/technology/archive/2012/11/when-t...
It seems today's politicians are (at least in some cases) familiar with what it takes to build reliable software. Perhaps the problem is that they are only willing to see that it's done when it benefits them directly, but not when their constituents need it.