Christopher Deppmann

A single, well-designed experiment can overturn theories. I don't see this often today (what with Cell papers and their 99 figures), but there are many papers from before 1970 that seem to fit this description. In 1943, for example, Luria and Delbrück tested whether bacteria became resistant after encountering a virus or were already resistant beforehand. (The widespread belief, at that time, was that cells became resistant to viruses by touching them.) The dup grew separate bacterial cultures, exposed each to phages, and counted the survivors. If exposure to a virus *induced* resistance, the cultures should produce a similar number of survivors. Instead, the counts varied a lot, just as you'd expect if a resistance mutation arose early in some cultures, but late or never in others. In the 1860s, Louis Pasteur used swan-neck flasks to dismantle spontaneous generation. He boiled broths in flasks with long, curved necks that would allow air to enter and yet trap dust. In flasks where dust could reach the broth, contamination followed. In open-air flasks with a bent neck, the broth never became contaminated. This experiment suggested that broths did not "spontaneously" generate life; they became contaminated by microbes clinging to dust in the air. Can we bring back these "classic" experiments? I think so. Perhaps there should be a research journal that publishes short papers built around a single, extremely well-designed experiment. If this journal were well-funded (a stretch of the imagination), it could host little workshops where researchers discussed an open debate and decided, together, what experiment would resolve it one way or another. It's effectively a sort of pre-registration. Then, the journal could fund scientists from each group to actually do the experiment and resolve that debate, and then publish the experiment, null or positive. I'm skeptical that this will ever happen, of course; there are too many crooked and perverse incentives in scientific publishing. But it's probably how science ought to be done. The journal could be called PARADOX.