A lot of our nuclear weapons are old, and it’s not clear that they still work. If we still did underground tests, we’d know for sure (and could fix any problems) – but we don’t do that. We have a program called stockpile stewardship, that uses simulation programs and the data from laser-fusion experiments in an attempt to predict weapon efficacy.
I talked to some old friends who know as much about the nuclear stockpile as anyone: neither believes that that stockpile stewardship will do the job. There are systems that you can simulate with essentially perfect accuracy and confidence, Newtonian gravitational mechanics for example: this isn’t one of them.
You had two approaches to a problem that was vital to the security of the United States: option A was absolutely sure to work, option B might possibly work.
The Feds picked B.