When I arrived at the BMJ in 1979 the journal was set in hot metal, there wasn't a computer to be seen, and it took three months for copies of the journal to reach Australia. As I depart a quarter of a century later, many more people access the journal through their computers than on paper, and Australians are the first to read each issue because the British are abed when it hits their screens. Yet my overwhelming impression is that change has been slow. If resurrected, Thomas Wakley, the founder of the Lancet who died in 1862, would instantly recognise both his journal (despite its recent makeover) and the BMJ. We are still at the beginning of the electronic revolution, and Drummond Rennie, the deputy editor of JAMA, has castigated editors for neglecting their craft and failing to innovate.1 2 We have been an instinctively conservative crew.

It took me many years to realise that I completely misunderstood what journals did. I imagined that doctors opened their BMJs on Friday mornings, read of some innovation, and used it on the next relevant patient. Many still seem to cling to this naive view of the function of journals. In fact words on paper rarely lead directly to change—and thank goodness they don't, considering the rubbish that journals often publish.3 What journals do best is what the rest of the media do best: stir up, prompt debate, upset, probe, legitimise, and set agendas. They are good at telling readers what to think about but not what to think, and theme issues may be particularly successful in putting important but neglected subjects to doctors. Increasingly I wonder as well if there isn't something fundamentally misguided in sending ordinary clinicians, …