For perhaps ten years now, the future of scientific publishing has been hotly debated. The traditional models are often thought to be badly broken, although convergence to a consensus of what a better model should be is not apparently close. But to my mind, much of this debate seems to miss one important point, how to publish data.