I did the calculation for Ohio-class capabilities assuming it was only loaded with the smaller 100kT warheads (W76), as the 450kT ones (W88) are relatively rare, according to Nukemap's estimates for a 840m airburst a single sub can cover 1500sq km (~1/3rd of RH, 10x DC) in 20psi (severe damage or complete destruction of concrete buildings, ~100% fatality), 1800 in 1000 rem (~95% protracted mortality) and 7700 sq km (Delaware, half of Connecticut) in 5 psi (~universal collapse of residential buildings, widespread fatalities).
But at the same time, we're moving past deterrence. The USSR fell 26 years ago. There's "deterrence" and there's "we can wipe out the entire planet". It's especially disconcerting when you consider that in most nuclear armed countries, US included, the authority to launch a nuclear strike lies solely with one person.
We are in no way moving past deterrence. SALT, START, ABMT, etc essentially codified it.
And deterrence is predicated upon creating the absolute certainty in your enemy's mind that you will leave their country a smoking, radioactive hellscape if they attack you.
This requires convincing them that the decision will be made, were it to come to that. Not "Oh, but a mid level officer might ignore the command."
In nuclear game theory, the less certain your opponent is of your resolve and capability, the more likely nuclear armageddon is.
I'd imagine it makes it less stable simply from a complexity = narrower bounds of stability standpoint, but fundamentally it doesn't change anything.
That's why nuclear fingerprinting and global monitoring have been so heavily invested in. "Safety in numbers" isn't a defense if the retaliatory strike still hits all the belligerents.
MIRV are fucking terrifying.