For instance, let's say China takes military action against Taiwan. Clarke asks, "What president would order the navy into the Taiwan Straits … if he or she thought that a power blackout that had just hit Chicago was a signal and that blackouts could spread to every major American city if we got involved?"Clarke then recommends a variety of steps to be taken---most importantly that the US President should take a close look at what the Cyber-generals are planning, since that could constrain his options in a crisis. More importantly, Clarke raises a bunch of questions that, as he notes, are obvious but hard to answer.
In this sort of crisis, the nuclear era's basic concept of "deterrence"—a threat of retaliation in kind—would have little effect, because America's power grid (like everything else electronic) is much more dependent on cyberspace and thus more vulnerable to a cyberattack.
What we need, Clarke writes, is "a credible defense," designed to cast doubt in the minds of potential attackers that their cyberattack will knock us out or paralyze the president with fear—at least enough doubt to dissuade them from launching the attack to begin with.
This stuff is far out of my expertise. My only experience, as I have mentioned before, is that during my year on the Joint Staff, the unclassified computers with access to the internet were incredibly slow because they were under attack all the time. The past year's events, with Google and China, clearly raised this issues' visibility. We all have much to learn. Clarke's book sounds like a good place to start.