This was in the context of someone complaining people don't publish null results. It raises the interesting question of how to define a "significant null result", where "significant" is the publication-worthy sense not the statistical one