Lonewolf
It was an observation given to me over 20 years ago that the MOD still spent 90% of the upfront money derisking airframe and electronic issues, at a time when software had already assumed most of the complexity and accounted for 90% of the slippages in such programmes (this was in the particular example of IUKADGE.
I then encountered exactly the same problem on NATO ACCS, where they thought they could reuse 20 year old flat files for 90% of the software, and ended up replacing all of it with nearly a 10 year slippage.
I find little difficulty in finding the same issue occuring again - and am amused, but hardly surprised, when even a 4 star suggests that he'd like, at this stage, changing the entire softwRe over to a new open-architecture, in the expectation it can be done in 5 years - and will save money.....