Coates is pointing out that there are good reasons
    to suspect strong sample bias.

If that was all he did then there wouldn't be a problem, but it wasn't. He also made unsubstantiated, anecdotal claims, and he presented the whole thing as fact.

    OkCupid didn't address this -- they just assumed an unbiased sample.

The OKCupid post very explicitly addressed how their user demographics differ from the norm.