[bob1029]

I agree.

How long until subcritical experiments need to turn into supercritical ones? At some point you have to do an integration test or no one can say anything of certainty.

Proving a sample of plutonium emits the right dose of radiation when whacked appropriately is a great starting point, but it doesn't validate the rest of the weapon system.

You could test the bombs with inert cores to prove everything outside the physics package is valid, but there is still the "but sometimes" bullshit space where perhaps the core seems good on paper but the neutron initiator has an undue delay related to aging circuitry and we lose 80% of the expected yield. Whatever the case - in isolation both system elements might test OK but they could still fail when combined.

I feel like the dial-a-yield devices are most precarious (e.g. B61). How could you really know you aren't going to over/undershoot massively? What is the range of uncertainty on that system after 3 decades?

[jjoonathan]

> undue delay related to aging circuitry

Instead of a dumb inert core they use a smart inert core that broadcasts precise measurements of the implosion using fiber optics and triboluminescent capsules that flash as the implosion wavefront crushes them. The computer has to broadcast the data before it itself gets crushed microseconds later. But yes, I still share your overall concern even though we can push the question marks one step later in the delicate process.

https://www.youtube.com/watch?v=FYdAT0v4DHs

[D-Coder]

> What is the range of uncertainty on that system after 3 decades?

On a NUCLEAR BOMB? Who cares if it's 50% bigger or smaller than expected? These are not precision weapons.

[defrost]

There are multiple ways in which nuclear weapons are 'precise'.

The yields are known to precise amounts - with more than 2,000 nuclear weapons already detonated both below and above ground there's a mind boggling array of technical data available.

There's data on the physical impulse of various types of nuclear detonation, and data on how much or little direct radiation is created by doping the core and surrounds.

The effects of those yields is known for a wide range of scenarios; very high in the atmosphere, above the ground such that the blast radius barely touches the ground, at ground level and below ground.

Data exists to show they essentially suck for technical engineering and canal building, they irradiate vast amounts of debris at ground level and create massive clouds of fallout, and that they flatten cities with minimal side effect if detonated high enough for the shock wave to do all the work of crushing buildings.

The precision is sufficient to setup in advance to create 'Dixie Showgirl' bad ballet photoshoots:

https://blog.nuclearsecrecy.com/2012/05/18/friday-image-the-...

FWiW I'm not pro nuclear war here, but it pays to be realistic about what has been achieved in the nuclear weapons domain.

[jjoonathan]

Problems don't make it go to 50%, they make it go to 0.1%. Not only are these precision weapons, these are the most precision weapons.