Yeah... the incentives are so lopsided for mostly positive studies/hypothesis.
I have been told many a times that I will have no career prospect if I publish, "everyone else is seeing phenomenon X in these experiments, but I do not in similar experiments." (or) "phenonmenon X is an artefact".
Aside from the publishing, the replication issue concerns me deeply as someone who straddles both theory and experiments. I have conducted long complicated experiments, get robust results etc., (and I am happy I see interesting effects in my data), but I always fear if the experiments and methods we engage in are becoming so complicated, not easy to record (or easy to overlook certain things one might take for granted) that replication becomes impossible.
jonny