This is a problem that cannot be solved by hard and fast rules.
There needs to be a decision making process that can guide the R&D project.
Unless of course goals, progress, etc. can somehow be quantified. (But these are complex projects, complex goals, even coming up with relevant metrics is hard, and somehow coming up with one utility function is kind of impossible - without spending a lot of money on R&D for that, but then you're back to square one.)
That said, yes, risk should be better spread, but also the following are all critically related to public spending/procurement: Personal accountability and motivation, too big to fail, cartel/oligopoly, agent-principal problems, corruption, institutional deadlock, too much bureaucracy, bad responsibility-authority allocation, etc.
There needs to be a decision making process that can guide the R&D project.
Unless of course goals, progress, etc. can somehow be quantified. (But these are complex projects, complex goals, even coming up with relevant metrics is hard, and somehow coming up with one utility function is kind of impossible - without spending a lot of money on R&D for that, but then you're back to square one.)
That said, yes, risk should be better spread, but also the following are all critically related to public spending/procurement: Personal accountability and motivation, too big to fail, cartel/oligopoly, agent-principal problems, corruption, institutional deadlock, too much bureaucracy, bad responsibility-authority allocation, etc.