Defence outsourcing

After hearing about the problems the MOD has with controlling defence spending I was musing on the problems that Government has with high-profile procurement projects that go wrong.
In terms of IT, the Government's record on this is no better or worse than private industry. About 30% of IT software projects fail completely (i.e. get canned). This is true for private and public IT projects. People think that the private sector is more efficient - that's not strictly true.
What happens in the private sector is that the inefficiency is spread across the entire industry. Let's take Search as an example. In order to have Google we also had to have Yahoo, Altavista, Lycos and any number of other false starts that I haven't heard of. As consumers we don't have to pay for the failures (at least not directly). In constrast, if you undertake a big public project that has a large element of R&D then all the chance of failure falls on that one project. Taking the example above, the chance of a publicly-funded internet search solution being successful was at best 25% and probably a lot lower than that.
The solution? Dont't outsource the hard stuff. Outsourcing is fine for things that are commodities or where all the R&D has already been done - no-one, apart from a printer manufacturer would develop their own printer, whether in-house or outsourced, for example. For the more complicated stuff (e.g. fighter planes) we have to ask whether we should bother getting cutting edge stuff developed. I'm sure the Generals, Air Vice Marshalls and Admirals will say that we need cutting edge equipment but there's not much point if it doesn't work
Imagine how well resourced the armed forces would be if everything was bought off the shelf...