Same with systems. A small team of hackers can work wonders on certain problems with no methodology and some TDD, but imagine 1000 of those small teams, many of them now with below average programmers, all working toward a single vision, that now need to integrate their individual projects, some of which can only be tested once our twice ever due to cost.
Huge budget => massive feature creep => massive manpower => huge overhead, extremely difficult communication/sync => failure to meet deadlines.
[1] : http://en.wikipedia.org/wiki/Apollo_program#Program_cost
Either terraforming mars or we might make it to Alpha Centauri by the next century.
> In 2009, NASA held a symposium on project costs which presented an estimate of the Apollo program costs in 2005 dollars as roughly $170 billion.
So multiply by nine then.