Everything about Ohio v. American Express was wrong and the adoption of “two-sided platform” reasoning into American antitrust law might be one of its worst, most regrettable wrong turns in decades. That is not because the original theoretical model of two-sided interaction has anything wrong with it at all. It is rather that nothing could be gained by incorporating it that could be worth the result in the American Express case itself, or the difficulty that has likely been invited into antitrust litigation. The consequences are hard to predict, but they may be severely limiting to our already moribund antitrust enforcement.

I will offer two major responses in this Article. Part II states a simple theoretical argument to demonstrate an important mistake in American Express. It was crucial for the American Express majority to characterize its decision as simply a problem in product market definition, but for the strictly strategic purpose of requiring plaintiffs to prove that where markets are two-sided, the challenged conduct reduces the quantity of the jointly demanded product. However, quite contrary to the majority’s self-assured confidence, the anti-steering rule at issue indeed could cause serious harm without reducing the quantity of the jointly demanded card-swipe transactions. The harm is dynamic and depends on the fact that, contrary to platform theory’s presumption, elasticities are neither fixed nor exogenous. Part III then explains how a deeper epistemological problem inherent in cases like American Express, and in conservative antitrust law more generally, poses very serious risks for the policy.