In fairness, the original source gives a calculation that's basically a more quantitative version of Gowers's dimension-of-intersection-of-Cantor-sets argument,

In fairness, the original source gives a calculation that's basically a more quantitative version of Gowers's dimension-of-intersection-of-Cantor-sets argument, and yields the same conclusion.