Agreed; it is difficult to build a product when there is only one likely customer - it is a bespoke solution (bespoke to that one customer) and I probably view it as consultancy rather than 'product' - but everyone will vary in their opinion of that. The crucial difference as to whether 'private sector' will build a product at risk might simply be the size of the potential market in this respect - if there is only one potential customer probably that customer should bear the risk of financing it.
With regard to expertise in government, my <personal> view is that the expertise of governments is in governing, not production. Its been a long time since government had a built roads, defence equipment or town halls on any scale. Same is true of IT systems as well. I don't see that as a particularly bad thing - the funding of a thing is seperate from the production of a thing, and its quite correct that the government has a unique capability to fund large scale works. However this does veer into the realms of politics and thats more of a belief than a technical matter, so I'm happy to drop this angle.
On your last comment, I agree that if the government or any other body actually commissions a piece of work they are at liberty to tell the supplier to build it out of pink cheese - its their cash, and as you point out, their product and IPR at the end of the day. What happens to this argument where the government is not comissioning the work ? They can certainly have a procurement policy on what COTS packages they might wish to purchase with tax payer money, but it seems to me that a policy enforcing Free software might just drive up the price of a piece of product rather than reduce it as expected. If supplier X invests £10k building a commercial product and wishes to sell it in the open market, and one half of the market insists it will only purchase if they can change and resell that software, or give it away, and the other half of the market is happy to purchase a closed-source system, the supplier might well conclude they are likely to make a thumping loss supplying to the Free sofware purchaser. This is a hypothetical scenario, but to look at it from another angle, if you as a software supplier needed to raise capital for your startup company to speculatively build a piece of software, and you were in the position of trying to convince a bank or Angel to fund this, what actual value do the investor get in return for their risk - it seems the IPR generated by the capital would just be distributed gratis to the open market. I can't quite see how the financing of this would work. If you have an example or a hypothetical scenario that rewards the investor for their capital, I'd be happy to get to understand it better.