16 September 2005

Lo vs. Wang becomes Lo vs Wang vs. arxiv.org

Every day, people post new papers on the arxiv.org preprint server, and they can update old papers. It's mostly unrefereed, but yesterday a paper titled "A brief history of the decoy-state method for practical quantum key distribution" (quant-ph/0509084) was replaced by the single line "This paper was removed by arXiv admin due to content not conforming to the standards of academic communication."

Fortunately (depending on your perspective perhaps), the arXiv still saves old versions of papers, and the original version is still online. Even non-quantum people should be able to appreciate the human drama contained within. (Quick summary: Lo is widely credited with being the first to do something, but Wang says that Lo's first implementation didn't work and/or didn't contain new ideas and that either he (Wang), or another guy, Hwang, was the first to do it, depending on how you define "it.")

I mention this a) because it's kind of funny when dirty laundry is aired in public, but b) because there are some semi-serious issues at stake. It's kind of tacky when people argue about priority, but let's be serious here - we do care about it, often we care about it a lot. And what exactly are "the standards of academic communication" when it comes to arguing about priority? And of course, who decides?

Right now, I think decorum may be preventing some useful clarifying discussion. For example, who should I cite for the result that N qubits can specify a reference frame to an accuracy of O(1/N^2)? quant-ph/0405082 and quant-ph/0405095 appeared on the same day. The acknowledgments in the published versions don't really clarify things. Do I cite both to avoid offending either group of authors? Or should I just cite quant-ph/0407053, which was definitely later, though also done independently? I might do this partly because I'm more familiar with its methods and partly because, being later, it references all the related papers, so that I only need one reference to explain this (for my purposes) rather tangential point. Frustratingly, it's hard to tell if a consensus emerges when people a) don't directly talk about it, and b) instead often err on the safe side by citing every paper that has made some contribution to the final answer.

Of course the problem is that people doing the citing (like me) care about things like not offending anyone, making the references useful to the uninitiated reader, and keeping the total number of references under control; usually in that order. The only people who care about clearly establishing priority are usually the authors of the papers in question, and of course they're always self-interested. This post is starting to have a pointless feel to it... But at least quant-ph/0509084v1 is a good read!


ahren said...

i'm against citations having any importance beyond their use as a reference, for this very reason -- it just seems to end up as a waste of time and energy for everyone.

aram harrow said...

to which i can only say, "clearly you're not a golfer."

or an academic.