Like almost everything, testing suffers from diminishing returns. The difference between no testing and delta testing is often huge, but the difference between a lot and a lot plus delta is often pretty small. But each delta of testing you add costs the same amount in terms of the prices you listed.
The optimal thing to do would be to keep adding more testing until the benefits of the latest delta drop to equal their costs. However, all the economic research on this topic that I've read suggests that while we need some testing we're way over the optimal point. Its the same sort of research that lets you publish papers on how the optimal copyright length is 14 years.
The same sort of ratchet effects are involved in both places too. The only people who care passionately about copyright lengths are those who have copyrights that are about to expire. The people who could have benefited from a work going out of copyright are often only dimly aware of it. Likewise when the FDA approves a drug that causes problems its obvious and there are a lot of people complaining, but when they don't approve a drug that could have helped someone - or when drug developers are scared off - people aren't as aware of it.
The optimal thing to do would be to keep adding more testing until the benefits of the latest delta drop to equal their costs. However, all the economic research on this topic that I've read suggests that while we need some testing we're way over the optimal point. Its the same sort of research that lets you publish papers on how the optimal copyright length is 14 years.
The same sort of ratchet effects are involved in both places too. The only people who care passionately about copyright lengths are those who have copyrights that are about to expire. The people who could have benefited from a work going out of copyright are often only dimly aware of it. Likewise when the FDA approves a drug that causes problems its obvious and there are a lot of people complaining, but when they don't approve a drug that could have helped someone - or when drug developers are scared off - people aren't as aware of it.