The real issue is that citation count is a terrible metric and always has been (when I was doing research the magic you wanted was the low but not no citation count paper which went into a huge amount of detail on exactly what they did).

Replication failures disproportionately hit high-profile journals. The journals' editorial processes favor "surprising", counter-intuitive claims: life based on arsenic, stem cells created by treating skin cells with a mild acid, cold fusion, etc. The prior probability these are true is lower (hence the surprise). At the same time, this also makes it more likely that someone will attempt to replicate the results--and that there will be interest in a negative paper if they can't.