To this point we have used the Asymmetric Escalation Game to assess the usefulness of all-or-nothing and limited-war deployment policies in extended deterrence relationships. As some might expect, we found that all-or-nothing policies are largely ineffective deterrents, unless Challenger is very likely Soft. When the odds are that Challenger is bluffing, Defender can rationally plan to move to the brink of unlimited war by escalating a crisis unilaterally, as Eisenhower and Dulles did with some success in the 1950s. Of course, Defender could plan to escalate even when Challenger is likely to be Hard. In this case, though, the beliefs that support Defender's intention to escalate first are implausible. Thus, while credible strategic-level threats may sometimes be useful for deterring direct attacks (see chapter 5), this prophylaxis is not so easily transferred to third parties.

To be sure, the prospects for peace are enhanced, and the chances of extended deterrence success are increased, when highly credible strategic-level threats are buttressed with credible and capable sub-strategic threats. But even here deterrence success is no sure thing: other rational possibilities always exist. Not even the most efficacious Flexible Response deployment policy, No-First-Use, reliably sustains the status quo. In extended deterrence relationships, therefore, there is no quick fix, no obvious or transparent way to guarantee a third party's security.