Stop calling them "Billion £ contracts" when its £100m a year running for 10 yrs might be a start
A £1Bn contract split into 10 sub contracts is *still* £100m each.
Which is still firmly in "usual suspects" territory.
In *reality* that's £10m per *year*, which is definitely in the (larger) end of SME land.
The bottom line is *always* proprietary lockin by the supplier.
This is *all* about the rules in the contract on documentation, source code, compiled versions etc.
Government systems are *likely* to be 1 off sales, as it is *highly* unlikely you can build one flexible enough to accommodate the UK NI systems (NIRS II anyone?) and then sell it to someone else (Although IIRC that's *exactly* what one of the US wanted to do with it or another system, when they pitched the system as client server).
Things would probably go easier if customers accepted 3 things.
1)No software is bug free.
2)Suppliers will release software with bugs. The question is what is an *acceptable* level, what should be done about them and who is responsible (IE who pays for the work).
3)Suppliers can (and *have*) botch the requirements analysis and know they will make any money they lost low bidding the contract by charging through the nose for change requests. CSA system anyone?
More creative mapping of government tasks to COTS software would probably save quite a bit. Looking for the right *class* of COTS software to work on and being prepared to *change* your existing (non computer) procedures to improve the fit.
The trick is to avoid excessive "customization." ending up with effectively yet another proprietary system that cannot leverage later upgrades to the core software (a perennial screw up of in house modified MRP systems back in the day).