I think this problem has to be tackled at the level of the journals. Journals should not reward "interesting" or catchy results with publication. Instead, exper

I think this problem has to be tackled at the level of the journals. Journals should not reward "interesting" or catchy results with publication. Instead, experiments should be accepted for publication before they are performed, and then published regardless of result. The criterion for evaluation should be methodological rigor and proven replicability, not potential for journalistic headlines. Change the incentives and you change the outcome. If scientists are rewarded for rigor (and they currently are NOT), they will bring rigor to their experiments. If they are rewarded for cutting corners, corners they will cut.If they don't they themselves will be cut from the profession as their CVs lag behind their less scrupulous peers.