February 1, 2008 by jimstogdill
FCS, SOSCOE, and the Big Bang
It bums me out to read statements like this one in this article about FCS/SOSCOE:
The software program “started prematurely. They didn’t have a solid knowledge base,” said Bill Graveline, a GAO official involved in the government’s ongoing review. “They didn’t really understand the requirements.”
That isn’t to say that I think FCS/SOSCOE is on track and being developed the best way, it’s just that I think these kinds of statements perpetuate the idea that software of this magnitude should be written as a Big Bang after every requirement is fully understood. There are 3000 developers working nine years at a cost of $6B and the expected value curve is supposed to look like this:
Contrast this with something like Linux where value has tracked much more closely with effort:
Why the difference? Well, unlike an open source project like Linux that slowly moves its way up the food chain from departmental web servers to mission critical applications as it matures, the government acquisition system tends to assume that at 8 years 364 days SOSCOE is an entry in an earned value report. Then, suddenly as the calendar turns over 9 years it is hatched as a fully functioning completed system ready for operational deployment. In the meantime, as it hasn’t been “delivered” it isn’t available to be used anywhere.
I would love to see large scale software developments like this thought of in much more incremental / evolutionary terms. I’d also like to see a greater degree of transparency and openness so that incremental value could be provided along the way, even to completely unrelated programs. After all, many of the component parts of SOSCOE are lego blocks that could be readily used in other environments.
In fact, my first graph is probably completely wrong. Without the hardening that comes from incremental use it is much more likely that the budgeted 9 years / $6B grows dramatically (like it did for Vista) before the value bit can be flipped.