I've been there. This is not an easy problem to solve - they do have vetting procedures and sometimes reject bids because they think the contractor can't do the job. But companies are pretty good at making sure the boxes are all checked during the bidding process.

I had a relative in the Air Force, and his group was in charge of salvaging programs that failed because the contractor couldn't deliver. They would get half-finished satellites and boxes of parts arriving at the loading dock and then try to figure out what to do. Is someone criminally culpable? Can we do a contract to finish this, or do we need to start over?

According to him you could just never tell - sometimes the contractors you were sure would fail actually delivered, and sometimes the old reliables went belly up or just kept asking for more money.

Yeah I'd say the issue there is that traditional defense contractors are highly silo-ed, quality can vary drastically from program to program. I've worked on programs that I'd pit against Google and programs where the culture was a bunch of math majors who learned to code in the 70s and haven't updated since the 90s. Just last year I had to review a 500 line FUNCTION in NEW CODE.

Thankfully I'm no longer attached to that program, had to continuously bite my tongue and pick my battles in code reviews to avoid getting a reputation. Even nicely telling people 15-20 years my senior "go spend two days refactoring this crap" was a non-starter. To be fair the code did work, I doubt a user would ever notice. But I felt bad signing off on the review all the same.

I sdhould point out that not all the old-timers are a problem. I'm currently working with some 20-year guys who are VERY up to date and open to new ideas; and the company is trying to modernize from the top-down, but some programs have taken to it better than others and some have outright resisted. Given that these programs are only staffed after a bid is won, I'm not sure what kind of vetting process could catch these kinds of inefficiencies.

Is there some reason we can't make the penalty of cost overruns that the contractor eats it, with the provision that the government is not allowed to change the specs once committed, after a preliminary speccing phase where they work with a few bidders to see their proposals? I imagine that we'd start seeing much more realistic bids in a hurry, with potential overruns built into the bid.

Periodically we try fixed price contracts and it never really works out. It's fine if you're ordering hammers or assault rifles - something that gets built all the time in quantity.

But when you're building an aircraft carrier there are just too many unknowns, so you run into trouble in the very beginning when nobody wants to bid on it. Then if it looks like the overrun will be too large the contractor goes bankrupt and you end up with a half-finished ship and a gaping hole in the capability it was supposed to provide.