Opponents of the Barack Obama administration's Iran negotiations have long insisted that the Tehran regime has cheated, is cheating and will cheat. After recent reports that Iran has been increasing rather than reducing its stockpile of enriched uranium, even supporters of the negotiations now concede that Iran will cheat if it can.
I am no fan of the administration's announced framework, but I do think some clarification is in order. In particular, as I often point out to my students in courses on both contracts and the ethics of war, the well-founded belief that your adversary is likely to cheat does not by itself mean you shouldn't make a deal. As Thomas Schelling and Morton Halperin put the point in their classic work "Strategy and Arms Control," "It cannot be assumed that an agreement that leaves some possibility of cheating is necessarily unacceptable or that cheating would necessarily result in strategically important gains."
Put otherwise, the question isn't whether your adversary will cheat. The question is whether the cost of your opponent's cheating will outweigh your gains from the treaty. To understand the point, let's examine two propositions from the literature -- one that accords with common sense, and one that does not but is probably true.
Proposition 1: Countries will cheat if they can.
Let's see why. When Country A enters into an arms-control agreement to ban behavior P, it is promising not to do what it otherwise would do. Put otherwise, in the absence of any agreement, A would still be doing what the treaty bans. This in turn suggests that A has calculated that doing P is in its best interest.
The only effect of the treaty, then, is to make doing P harder (that is, more costly) than before. It is still in A's interest to do P, as long as the chances of getting caught are low. More precisely, Country A will cheat as long as the perceived gain from doing P outweighs the discounted cost of being caught.
What does this mean in practice? Iran perceives the value of doing nuclear weapons research as very high. The critics of the framework think that the discounted cost of being caught is low. The discounted cost is a function of two variables: the likelihood of being caught multiplied by the cost of the penalty that would be imposed. A high penalty makes little difference if the likelihood of being caught is low, and a high likelihood of being caught makes little difference if the size of the penalty is small.
Critics think that the framework offers the worst of all possible worlds: The likelihood of being caught is low, and the size of the penalty is small. If they are right, it is easy to predict that Iran will cheat.
Whether the critics are right depends in part on Proposition 2.
Proposition 2: Sometimes, a stronger verification technology actually decreases the likelihood that cheater will be caught.
This controversial hypothesis, supported by some but not all of the literature, proceeds from an estimate of human behavior. Evidence of cheating is not an absolute. Rarely will one discover a decrypted transmission that reads, "OK, time to start cheating." Fragmentary evidence must be evaluated. A relatively weak verification technology may cause the evaluator to be overly suspicious, and a relatively strong verification technology may make the evaluator complacent.
When the technology is weak, the evaluator may think, "I don't like the look of all these little clues. Country A must be cheating." And so you get intelligence analysts insisting on thin evidence that the presence of weapons of mass destruction in Iraq is a slam-dunk.
When the technology is strong, the evaluator may think, "They can't possibly be stupid enough to cheat. They know we can see what they're doing. We must be looking at something else." And so you get intelligence analysts assuring the White House that clear evidence of the construction of intermediate-range ballistic missile sites in Cuba in 1962 isn't what it appears to be.
The administration's position is that the West will know if Iran cheats. This assumes both a strong verification technology and correct interpretation of the evidence. But if evidence of cheating is likely to be open to question -- or if the process for determining compliance is likely to be lengthy and involved -- then the cost of cheating decreases, and the incentive to cheat increases.
That's why the administration continues to insist that sanctions will "snap back" should Iran cheat. Unfortunately, in the real world, there is no way to craft or enforce such a sanctions regime.
But that hard truth doesn't mean that it would be wrong to reach an agreement. Again, one would have to calculate the West's gains from Iran's compliance and subtract the costs of the likely cheating. That's the real battle between the administration and its critics. They're not fighting over negotiation versus war. They're fighting over the arithmetic.