It is best to start thinking about how extended nuclear deterrence (END) might work in future by looking back at its Cold War origins. It is, of course, an American concept. America deliberately promoted expectations that it would respond with nuclear weapons to a Soviet attack on its allies in Europe and Asia. This policy had two aims: to deter the Soviets from attacking US allies, and to keep the allies loyal by reducing their incentive to build their own nuclear forces.

For END to work, America needed both the evident nuclear capability to strike the Soviets hard enough, and the evident willingness to use that capability if its allies were attacked. The capability bit was never in doubt, but it proved harder to persuade both Moscow and America's allies that it would actually use its nuclear forces to defend others. 

The proportionality of a nuclear response to a conventional attack was one problem, but the real question concerned costs to the US once the Soviets had the ability to strike back. Would Washington risk nuclear retaliation against the US itself to defend an ally an ocean away?

Much of US nuclear strategy in the Cold War was devoted to persuading both friends and foes that it would. Ultimately, the US succeeded because it convinced others that it saw the loss of a European or Asian ally as posing a direct threat to the US, because they feared that would lead to Soviet domination of Eurasia, which would make Moscow strong enough to overmatch and dictate to the US.

What does this tell us about END over the next few decades? The capability element seems to me pretty clear. The US can easily maintain nuclear forces able to devastate any adversary, and – speeches in Prague notwithstanding — I think there is little doubt that it will do so. But America's ability credibly to threaten nuclear attack to defend other countries is much less assured. It depends on whether Washington can persuade others – adversaries and allies alike – that it would be willing to go ahead and launch a nuclear attack if its bluff was called. 

There is no reason to doubt American willingness to use nuclear forces against powers that cannot retaliate directly against the US, at least in relation to nuclear attacks when proportionality is not an issue. It is, for example, entirely credible that Washington would launch a nuclear attack against Pyongyang or Tehran if either of them used nuclear forces against a US ally, because neither have (for the time being) an assured capacity for nuclear retaliation against the US. 

But these are not the tough cases. The real question about the future of END is whether the US can credibly threaten nuclear attack against an adversary that could retaliate against the US – which for present purposes means Russia or China. 

Take Europe first. If Moscow launches a conventional invasion of a NATO ally – say Latvia or Estonia – could America credibly threaten nuclear attack to force Russia to withdraw? Could it persuade Moscow that the independence of the Baltic States is so important to America that it would accept nuclear attack on the US to preserve it? That seems to me very doubtful, because Russia today does not have the potential to dominate Eurasia the way it did in the Cold War. The stakes for America are simply not as high – and not high enough.

What of China? As General Xiong once asked, is Taiwan's independence worth Los Angeles? More broadly, is anything on the western side of the Pacific important enough to Washington to convince Beijing that America would accept nuclear attack on its homeland to defend it? Even Tokyo?

This is the core question for the future of END: what is so important to Americans today that they are willing to suffer nuclear attack to defend it? And can the US persuade friends and allies that it is willing to make this sacrifice? Many would suggest the maintenance of a US-led world order is important enough, but I doubt that, especially when the non-military challenges to that order are so strong anyway.

I think the costs of nuclear attack are so high that nothing except the defence of America's own territory and independence would justify them. In the Cold War, when it seemed one power could dominate Eurasia, America's own security appeared to be at risk from a Eurasian hegemon. Today, with at least two and maybe four major Eurasian powers, that risk seems very remote.

So against any power capable of delivering nuclear weapons onto American soil, END is an anachronism. Neither America not its allies yet accept this. The sooner they do the better for everyone.

The Nuclear Reactions column is supported by the Nuclear Security Project of the Nuclear Threat Initiative, as part of a wider partnership between the NSP and the Lowy Institute.

Photo by Flickr user -Alina-.