When is the critical rate of discoveries that warrants any further experiment?

We were recently dicussing that problem too what Nature writes about the Encode project:

The question is, where to stop? Kellis says that some experimental approaches could hit saturation points: if the rate of discoveries falls below a certain threshold, the return on each experiment could become too low to pursue

As always – the scientific method once invoked – creates beautiful results but when it comes to justification of programs or methods it’s all about personal preferences, irrational beliefs, common misunderstandings, conformance to general trends, and whatsoever non-scientific influences.