Monday, November 01, 2004

Game to go overtime? 

I don't often delve into international affairs and the politics of terrorism -- there are far too many blogs with writers more knowledgeable, including most of NARN -- but I think some people are going to overlook an intelligent point Craig Westover makes today. Regarding the diatribe/no-detonate behavior of bin Laden, Craig observes:

There was a lot of speculation that there�d be a terrorist attack before the election, which hasn�t occurred with a day to go. We, and I number myself among the �we,� have credited the President�s policies, which I still believe are the best available course. But think about it. What would have a more powerful effect if you were Bin Laden. An attack before the election or after?

An attack on U.S. soil after the election creates instant buyer�s remorse on the electorate and deepens the divide of the American people. If Bush wins, an attack �proves� his policies have failed to keep us safe. If Kerry wins, an attack �proves� the terrorist have been emboldened by Kerry�s weakness. In either case, an attack virtually ensures four more years of division and partisanship. Even for a new President Kerry, there�d be no unifying spirit that Bush experienced.

I was trying to sketch this out in game-theoretic terms. Suppose OBL/AQ have one attack on U.S. soil that is still undetected and with a high chance of success. The nature of the attack is unimportant, but suppose it would be of similar magnitude to the Spanish attacks. The only decision available is timing of the attack: before or after the election? And of course there are two outcomes, Bush and Kerry. Think in terms of payoffs A, B, C, and D.

Many assume that Kerry is Osama's candidate. The Spanish story says that to get that outcome you should attack before. But we should recall that the Aznar government appeared to be rolling to victory before the attacks. AQ may have seen it plausible to tip the election by the attacks before the election and did so, but only because pressure from the attack could change many votes. Suppose, however, that you could not predict in the U.S. whether or not an attack before the election would move voters towards Bush (out of spite, or out of perceived Kerry weakness) or towards Kerry (out of a sense that Bush had failed.) Suppose as well -- as many believe, but not me -- that the electoral outcome is close. You thus need to meet three conditions to make A or B the most desired outcome:
  1. You know which way voters will move;
  2. Votes are responsive or elastic to terrorism on U.S. soil; and
  3. Without your attack, your less-preferred candidate will win.
I think that amplifies the reasons Craig discusses for why the post-election attack may make more sense.