The curse of the gifted programmer is apparently the need to deal with occasional direct emails from Eric Raymond.

The $400,000 dollars per year isn't all going to product development. $350,000 of it is to compensate the consultants for navigating the nightmarish government bidding process laced with procedural road blocks to newcomers, cronyism and incompetence. Been there and done that, would not do business with again.

I used to put together proposals for a huge general contractor as a marketing assistant. The completed government proposals usually consisted of between 15 and 30 three inch ring binders of information. It would take me 20 minutes just to load it into FedEx via hand truck from my car. (This would be for something like a military barracks, I'm not talking about skyscrapers or stadiums. Not to mention that it wasn't uncommon to have several rounds of RFP's.) Contrast that with a civilian proposal that would be two or three binders. Software is not construction, but I'm willing to bet the process of bidding is similar.

In other words, it's a fat payday if you can stomach the process, but few can. This is also doubly sinister in that it's a process that is going to be particularly distasteful to just the sort of person that would make a great developer; one who holds efficiency and ingenuity on high, etc... Which is exactly how you end up with terrible DMV sites and CIA, FBI and DHSA databases that are unsearchable, and won't talk to each other.

Addenda: I'm not trying to put people off. I just want to shed some light on exactly how much pain is here: A LOT. Maybe the game-changer isn't someone getting the contracts and doing great, but someone rejiggering the bidding process?

I completely do not understand this sort of thinking. It's like saying, "All you people driving motor vehicles: trains are better". It's a meaningless comparison. NoSQL systems are really great for certain things. RDBMS systems are really great for certain things. What I would really like to see is if all this effort writing these kinds of articles went into well-thought out pieces that discusses very specific cases NoSQL backends provide an advantage and why. Use cases that are not a blog, digg, or twitter clone would be most helpful. Some of us have to work outside the bay area for companies like banks, insurance companies, hospitals, etc... Can I use Cassandra as a data warehouse for electronic medical records? Hell if I know, without actually having to learn it and implement it to see if it works.