In the ideal world, everything would be replicated and retested at least a few times. Practically, we do not have the resources, and sometimes other interests come into play. For example, those of some industry to fund studies that refute some wild idea that threatens their business. How many studies on safety of sugar are paid by Coca Cola?
That was not my intention. My intention was to say that a single study might not be good enough as a refutation. BTW one of the largest proponents of re-checking studies that were once considered reliable is a medical doctor, John Ioannidis. It is not just soft science that suffers from the problem.
Given that we are really short on money (and, with regard to ideas, always will be - it is cheaper to produce ideas than to test them even cursorily, much less thoroughly), every proposed mechanism will have large downsides. I do not have a proposal.