Yglesias is simply pointing out a dynamic which he feels may be under-appreciated: the impact of cannabis sales on the purchase and consumption of other intoxicants. He isn't saying it will go one way or another, but merely drawing attention to this dynamic and its potential implications.

A lot of useful information follows the format of "X may affect Y in one way or another, and we need to be cognizant of how these potential interactions could impact relevant issues." Your dismissal of Yglesias conjecture as useless is, in my opinion, missing the point. Just because we don't yet have data on a specific measure doesn't mean we can't discuss why that measure is significant and worth our attention.