Is there a journal or news site that exclusively publishes on studies that have been replicated?

There should be a whole library of such studies, perhaps even with more than one successful replication.

All signal, no noise. Entire disciplines would be absent from it

Imagine if academic success were measured by having studies in such a library. Not just the original, but also the replications! Somehow we'd also need to incentivize publications of non-successful replications just as much though.

I wonder what an action plan to make this happen would look like.

It would appear that many researchers are aware the system is suboptimal but are resigned to it, or even benefit from it.

Maybe we should pitch a tech billionaire to fund a journal like this.

96% seems a bit high to not be mildly suspicious.