ah. good to know that i can replicate myself :-)

Tom, my experience is that its very difficult to identify key conclusions; and SPIN

Do all of these large replication studies select highly 'impact' studies where the 'winner's curse' is strongest or do some replication projects choose experiments randomly?

Responding to Tom, At the Brazilian Reproducibility Initiative we are working with a random sample. That said, it’s always filtered by “experiments we have the capability of performing”, so the final sample is never truly random.

(see https://doi.org/10.7554/eLife.41602)

Effects are exaggerated by a factor of 6, on average?

Have you considered sharing your results with journal editors or as a PubPeer comment on papers that were not replicable? This could notify future readers of non-replicable studies, who might reconsider citing these studies.