Well Chris, I'm still mad that you got your picture in there and I didn't.

But I was pleased with the depth in which Jim Giles analyzed the issue, and I also felt that all of us expressed fairly similar views. The Medawar quote was a decent (if a bit overwrought) summary statement. Too many researchers waste far too much time—often in parallel with many other of whom they're not even aware—struggling to reproduce claims.

'Replication' does need a social component beyond the occasional scientific meetings and e-mails and phone calls. As you alluded to, within a given subfield there's almost always an informal understanding of what's turned to gold and what's rubbish. But placing this discussion in a more formal (but not overly), err...format and getting scientists to be aware of it, to be able to easily access it if they're looking for it, and even just stumble across it sometimes is what needs to happen now. Throw out whatever cliché you want—'paradigm shift' comes to mind—but that's what needs to occur in the way scientists approach publication and the subsequent analysis and re-analysis (and so on).

What I'm hoping to do at CSH Protocols (and I don't mean for this to be an advertisement) is to create this on a small scale with a (relatively) non-controversial subject—the technical aspects of individual experiments. Hopefully efforts like this and PloS One will, over time, help change the way people look at scientific research, from a set of self-contained, smoothly flowing publication units to a set of messy little experiments that don't have to have grand, sweeping consequences. in other words, something that better reflects the reality of the bench.

There will still be lots of dead ends. But why spend years travelling them?

Michael Ronemus