I agree that recent results have been negative. The reason why they were negative when other previous ones were more positive is, however, undetermined.

With a process that seems to be highly variable, and that the original paper had a only 1 in 10 chance of any success PLUS one that seems to be highly change to produce similar results (but unconfirmed if the same) I’m not too surprised at the high variability of the tests so far.

Frankly, I think it’ll be several years before we get anything close to more conclusive results in either direction. The replication results so far have been… enthusiastic but not exactly stringently controlled.