In at least the original landmark replication study, the studies actually replicated very well, but (~60% of the time) the signal wasn't as strong in the replic

In at least the original landmark replication study, the studies actually replicated very well, but (~60% of the time) the signal wasn't as strong in the replicated study as in the original reported results. That tells us a lot both ways. Also, while they are valuable, the error could be in the replication study.