What if the same experiment gives different results? Scientists have this problem sometimes, and now they have a new weapon against it.
Each generation of scientists knows more and more about less and less. As research delves ever-deeper into the fascinating, bewildering details of biology, physics and other disciplines, research papers become more complicated and, let's face it, harder to understand.
This is good - we need to pursue new knowledge on all fronts. But ever-narrower specialisation means that other scholars, not to mention the general public, are ill-prepared to check the data and assess the conclusions of scientific papers. And scientists are human; they make mistakes.
The problem has been studied in scientific circles for years, because it is growing serious: more and more experiments, their results proudly announced and guiding further research, are proving to be "unreproducible" - science-speak for wrong, or at best suspect.
Enter Science Exchange, a California-based firm with a plan. Outside experts will check and repeat an experiment before results are published; if the findings are greatly different, it will be back to the old drawing board for the whole project. This is a sound idea. We put a lot of trust in science and we need to be sure that we're getting the right answers.