I kinda feel like that 'safest' option, even when including the 'no nukes' option, is for stable and rational powers to have lots of nukes, and probably a 'fail-deadly' doomsday machine type setup. If they're all rational, and all know about each other's doomsday machines/deadman's switch, they wouldn't attack each other as it would be tantamount to suicide.

Of course, things start to get a bit dicey when you add not so rational nuclear powers in to the mix (e.g. North Korea), or diversion of nuclear bombs to not so rational 'non-state actors'.

> If they're all rational

I feel you, but that assumption is where this argument diverges from reality.

I've read that many people consider the "both sides have few nukes" dangerous. One side may think they can strike first, either conventionally or with some of their own nukes and disable their opponent's ability to retaliate. With thousands of warheads that isn't possible.

Similarly, the complete removal from the earth of nuclear weapons could potentially make the world much more dangerous. Without the threat of nuclear annihilation conventional war between large nations becomes more conceivable.

It would take more time than a nuclear exchange but it is possible to inflict comparable levels of destruction through conventional warfare if you have multiple nations sending millions of humans into battle as part of a war that drags on.

A perhaps unforeseen consequence of the development of nuclear weapons is the marked decrease in war deaths since their introduction at the end of world war 2 [0]. The Long Peace (as it is called in [1]) may be due to a number of other causes, but nuclear deterrence between the major powers has to be included as a factor. Of course, this decrease in war deaths could be reversed instantly in just one bad nuclear day.

[0]. https://ourworldindata.org/war-and-peace-after-1945/ [1]. http://www.fallen.io/ww2/

I think the source is lacking larger context. For example, postwar reconstruction in Europe ushered in an unprecedented age of European states not going to war with each other all the time. So if you measure absolute numbers of war deaths declining, you can't attribute them directly to nuclear weapons.

Also, advances in medicine have significantly reduced the number of deaths in war. It's worth noting that in the American civil war, the vast majority of deaths on both sides were attributed to non-combat disease.

So maybe you have a point, but I doubt that it's as simple as more nukes = less deaths.

I recently watched "Command And Control" (http://www.commandandcontrolfilm.com) on a whim. I hadn't read up about the film before seeing it, so I understood "nuclear missile accident" to mean, presumably, almost accidentally firing a warhead -- a more-or-less standard nuclear nightmare.

This, of course, is not what happened. The Damascus Titan accident did not involve a nuclear weapon being mistakenly, but intentionally, armed. The very real fear was, instead, that the explosion of a damaged missile could accidentally detonate the warhead as well -- or, in a case only slightly better, blast it into radioactive dust.

I'd never considered the risk entailed by the simple existence -- intentional use entirely aside -- of nuclear weapons. Reading the list of (public!) broken arrow incidents is harrowing.