Well, I think one of the major factor to problem being discussed here and in the link provided is also the fact that publishing negative results is not very rewarding in any field of science. I mean let's say I read paper XYZ and I try to incorporate it into my research. If it works great, if it doesn't work it is highly unlikely that I will bother myself with publishing that certain experiment in XYZ paper doesn't work or doesn't work in context of my research. I will rather move on and focus on what will give me the best paper in context of my research.
The only instance where people will update that following paper and its experiment is all wrong is when -
- the paper makes an extraodinary claim and claim isnt reproduced in other labs - example being STAP cell paper and GFAJ1 paper.
- when their research is in direct contradiction of published results, oh well that makes good papers as well.