Yes. But physics has a layer of abstraction on top of the replication problem, if you will.
If you were to go through the physics literature, you would find studies that are hard to replicate. Maybe even my dissertation. ;-)
The "studies" that are hard to replicate in the social sciences are testing what I call isolated factoids, such as: If you give children a chance to eat a sweet snack, they will tend to become criminals as adults. There is no way to test this factoid except by carrying out more identical studies and hoping that statistics will converge in favor of an answer.
Instead, physics can attack a problem from multiple angles, especially by connecting different factoids with a web of relationships based on theory. If it works out, then you end up with a web whose structure remains reliable even if you knock out even a large number of studies. For instance I don't think we'd consider abandoning Maxwell's equations if we discovered that much of the early experimental work done in support of developing those equations was faulty.
This is why I actually think that replication is not actually the gold standard of science.
Perhaps I should have clarified. What I should have said is that replication is certainly important, but replication alone can't drive science forward or produce a reliable scientific knowledge base. At least, it hasn't done so yet.
And a science that's bolstered by the development of robust theory can tolerate a certain amount of replication failure.
It's fascinating to me that social psychology can have a replication crisis, and somehow gets turned into a replication crisis in all of the social sciences.
