In the field of safety science, we have stopped competing empirically. The theorists fight each other with keynotes and editorials, the empiricists tinker within the boundaries of existing theory, and the practitioners use neither theory nor evidence to determine their activities. As a result, safety science is advancing very slowly, despite a high and increasing volume of research activity and publication. The journal Safety Science alone has published over a thousand articles in the past five years and has rejected over five thousand. Some of those articles were the capstones of PhD projects. Some were the outputs of publicly or industry-funded research. Most represented hundreds of hours of intellectual labour, and substantial emotional commitment. Taken together, this is a massive program of work that has had a marginal impact on moving existing theory or improving safety practice. Whilst it is tempting to believe that this is just the normal grunt-work of science – small steps, dead ends, and occasional breakthroughs – a close examination of the work being produced makes clear that the unproductive effort is not necessary swarf from the machine-work of making knowledge, but waste caused by poorly directed or poorly designed research. Such squandering of energy, talent and resources makes us furious. This paper, targeted at the Special Issue on the Future of Safety Science, is a proposal for how we should frame our empirical contributions so that safety science (and the journal Safety Science) has a positive future. For a field of research to move forward, each new project or paper must strive to change what has come before – adding, synthesising, testing, tearing down or making anew. Not every piece of work will be successful in creating lasting change – but every piece of work must genuinely try to advance current theory. The paper frames and justifies a set of commitments by the authors in order to find a brighter future for safety science and invites readers to share those commitments.