More On Competition April 20, 2006Posted by gordonwatts in physics.
Tommaso's and my posts on the Bs rivalry got picked up by the Cosmic Variance blog in, ironically, a post talking about sociology (I have got to update the look on my blog; ugly compared to everyone else!). One of the things both Tommaso and I discussed is what it takes to get a result out of a large experimental collaboration. The comments to the Cosmic Variance post picked up on some of this. In particular, there were several comments that could be summed up by this one by Scott O:
The SNO collaboration goes a step further still. It is collaboration policy not to show any result in public unless it has both gone through extensive internal review and has been submitted for publication to a refereed journal. In other words, there is no such thing as a “SNO preliminary result”. The attitude is that if it’s not ready to submit for publication, it’s not ready to show in public either. Obviously this slows down the publication process, but personally I think there’s a lot of be said for it as well.
His point is that no result that couldn't make it past a prestigious journal's referees (i.e. it is ready for submission) is fit for public consumption. Ok, I'm being a bit extreme here, but you get the point.
This is a valid point: the preliminary results that we release are lower quality than the results we submit to the journals. For example, if we measure, say, the top mass. The result consists of a central value — the mass of the top — and an error. The difference between a preliminary result and the result submitted to a journal for publication is the number of cross checks and (thus) the understanding of that error. It usually takes several more months to take a preliminary result and get it up to paper quality.
Some preliminary results never make it into a paper. Sometimes by accident: the person doing the analysis leaves before publishing, etc. But usually it is because the preliminary result is an update of a previous result. Since it is an update, it isn't necessarily worth another publication. But it is worth getting the result out there so others can see how we are doing, update inputs to their calculations, etc. Sometimes the result, while interesting to a small community, isn't generally interesting and so may not be suitable for publication. Cross checks between Monte Carlo and data will often fall into this bin.
In the end, however, the published paper is the matter of record. Take the recent Bs results I mentioned above. Had we released our preliminary result, and then CDF released theirs and immediately submitted a paper (as I think they have) — and D0 never did submit a paper… well, 50 years from now CDF would be the one that did it first. For big results like Bs, the top quark discovery, the (future) Higgs discovery… there will be no preliminary result. The first result the world sees will be the same one submitted to a paper. These things are too big to make a mistake on; all cross checks have to be done up front. In that sense, Scott O has it dead on.