Fair enough--I suppose this one should go into the "maybe" column, given the different results. The study you cite has a problem with circularity of logic: people went to the hospital because they felt chest pain (predominantly--90+%) and then, of the small fraction who had elevated cardiac troponin I levels (16%) indicating a high likelihood of an actual heart attack, well, yeah, they basically had chest pain.

But this could have been a selection effect--people with chest pain go to the hospital because they know that if they have chest pain it might be a heart attack and they should go to the hospital. This is ironic, given that the article criticizes the large post-hoc studies for having selection effects. (Also entirely true. It's just that we have no reason to assume they have any less of a selection effect.)

It's not a very definitive paper. Neither were the other ones, though. Putting it in "contentious" is reasonable. Putting it in "mythical" is unwarranted.

Anyway, this mostly a tangent, and largely irrelevant to the overall point.

The problem is the absolutism, not the reality of a substantial (but not complete) dichotomy.

"Look how many clustering algorithms there are which I didn't apply, to contrast with Stock's one ad-hoc hypothesized clustering-whatever which she also didn't apply" really doesn't lend itself to any substantive conclusions about the clusters.

The problem is the absolutism. Just attack that.