That's true to a point, but replicability really is the cornerstone of science. A replicable result, even if it's ridiculously implausible, at least needs to be properly investigated to see where the mistake lies. There's something there that is investigable (is that a word?). Conversely you can have all the controls and peer review you like, but if the result isn't replicable it's garbage. Just nicely presented, plausible garbage which frankly is the most insidious and dangerous kind.
There have been some interesting studies into replicating results from published research that have found that only a very small fraction of peer reviewed papers present actually replicable results. That's shocking.
With replication, somebody has to actually try and duplicate experiments.