1. We have learnt not to trust untested equipment
2. We have learnt that under pressure well meaning people will demonstrate equipment that while kind of representative of the reality are honed, improved and generously helped to perform well on test
3. Only repeatable performance under active competition is a real measure. It’s why sane people like open well designed markets. And it’s why weapons that get battle tested tend to be reliable
I am sure that overall if we read the red button nukes will fire and land and kill millions.
It’s just the day after I expect that the auditors will find a large number failed to fire, or failed to leave the tube, went the wrong way or simply otherwise went wrong
The point of credible threat is not one shot one kill, but so many shots we can basically guarantee a kill.
In other other words, both sides hold enormous number of nukes ostensibly to ensure they have some to fire after a first strike - but the effects of an enemy first strike are indistinguishable from poor engineering and beauracracy