And when it failed, the government would be criticized for hiring a start-up to do a job that was obviously so enormous that only a big established contractor could take it on. And rightfully so.
Fixed that for you.
Sure, a mythical online website built from scratch in Node.js with a Mongo backend can be out together by a bunch of college graduates, but without talking to a dozen other systems this site is not going to do much, and as soon as you take people out of the comfort zone of shiny toys we'll see how long the remain interested when you end up in SOAP and CORBA or are screen scraping mainframes. The people willing to tolerate this mess don't generally intersect with the startup crowd.
Interaction with legacy systems is all the future has left, plus its a problem - and the most elegant hack to scrape the aging DoJ mainframe wins the prize.
No, its clear public data architecture has to change.
But this is a space that could be disrupted (albeit not easily). And isn't Clever[1] trying to do something similar in the education space?
[1] http://www.businessinsider.com/clever-y-combinator-student-i...