Short is still sweet.

Shortly after taking over as Chief Editor of my former journal, EMBO Reports, I penned an editorial outlining the kind of research articles we were seeking as submissions [1]. The message was straightforward: we were looking for manuscripts reporting single key findings of note, which were backed up by multiple lines of evidence. Although EMBO Reports embodied the idea of short-format papers in its very name, I’m trying to bring the same philosophy to Fly. A common fault of many submitted ‘in-brief’ papers is that they present the results of just one experiment, often accompanied by exorbitant claims as to their significance. Reviewers, unconvinced that what was presented is a substantial advance, or even doubting its veracity entirely, then propose other experiments that they themselves dreamed up, which go off on a tangent, even if it’s an interesting tangent. The end result is that the paper becomes an unwieldy composite of two, three, or four disparate experiments, none of them backed up by independent evidence. When authors carry out an experiment demanded by a reviewer, which produces the anticipated result (‘the authors must demonstrate that ....’), the satisfied reviewer is often inclined to disregard the fact that the finding is of small magnitude, statistically questionable, or lacks independent repeats which the journal’s revision timetable made impossible. The published paper therefore ends up full of holes, is fundamentally unreliable, meanders all over the place and is rarely cited because nobody knows what it is really trying to say, even if the data are believable. At the opposite extreme, junior investigators often try to bend the parameters of a manuscript so that it includes every possibly relevant experiment they have ever done. The submitted draft thus resembles a ramble built from scraps demanded by reviewers, but from a much earlier stage. If the reviewers are awake, they will often seek removal of extraneous materials until, finally, little of substance is left, and it risks actually becoming the one-experiment paper that lacks evidentiary rigour; unless the editor is also awake. To me, the perfect short-format paper should have a laser focus on one simply formulated scientific question, that is then tested by different experimental approaches. Beyond that I would not seek to be prescriptive. The various streams of verification could be, for example, a bioinformatic prediction of an interaction, plus a biophysical test in vitro, plus a set of pulldown assays from a cell-line grown under different conditions and, finally, a phenotypic analysis of an in vivo model. Or any of hundreds of other possible combinations. The experiments should not simply be linked by a keyword, but should constitute parallel or sequential experimental tests of the specific hypothesis, where the outcome could be either ‘positive’ or ‘negative’, but should be coherent and statistically sound. Note that negative findings can provide at least as much mechanistic insight as positive ones. Negative results formally exclude a proposition, whereas positive ones tend simply to ‘not exclude it’. Only rarely do they provide such compelling evidence that they can be considered to have proven its validity. Asserting inappropriately that they do so is the basis of most of the exorbitant claims referred to above. How can you test whether your manuscript really fits the mould of a short-format paper? Many highprofile journals nowadays insist on short titles, such