While discussing the paper I’m working on, John Baez (from The n-Category Café) and I ended up discussing the use of web-based references in academics. He said
I cite online stuff all the time, and include hyperlinks in my papers. Of course stable locations like the arXiv are best… but we can’t continue pretending the web doesn’t exist.
[This is] one of my minor crusades: gettting academics to take the web seriously.
I do see a bit of a point to the traditionalist view. Stability is not just a benefit, it’s half the point of academic papers. Journal references are one stand-in mathematics has to replace repeatable experimental results. I’m a little edgy citing anything that a reader can’t follow chase down (as I’ve had to a couple times) for precisely this reason.
The arXiv is valid because the mathematical community as a whole has decided to trust that it will be there in as much perpetuity as print journals will be. I don’t think the community believes your site will be maintained for quite that long a time. In effect, there’s some unspoken idea of how “solid” or “verifiable” a reference is, and the level of publication dictates a minimum level of solidity. I can write a weblog post based solely on conversations with colleagues, but I can’t have a bibliography entry reading, “John Baez believes it and that’s good enough for me.”
The other side is giving credit where it’s due. See, I come up with at least half the stuff I do on my own, but I’m not the first person to do some parts of it. In particular here, I know I’ve seen John cover the fact that spans of finite sets decategorify to matrices of natural numbers. But he doesn’t know who did it first either. And so it slips into the cultural background and nobody gets the proper credit.
On the one hand, some day some referee will come screaming that I need to cite this or that, or be jealous that I didn’t mention his (or his student’s) tangentially related work. I really don’t mind putting in such references, but I’m really horrible at knowing what to cite. On the other hand, what happens if something I think I’m breaking new ground on — like the use of spans to extend classical invariants, or the “covariant” definition in this paper — slips into the mists like that? People use it and like it, but nobody points to me. I’m really not jealous in the long run about my work, but I’m sort of at a vulnerable stage in my career here, and I need to play the game a little cutthroat.
So what does this have to do with the web? Stability comes back to the forefront. A stable reference includes reassurances that the credit that was given today will still be given tomorrow. A generic weblog might ascribe a result to Stroppel today, but edit it to Sussan tomorrow, and which can I believe? I might trust the author of the weblog, but how does a referee know to trust my trust? In a way, it ties back into the notion of “Common Knowledge” that’s been burning up Terry Tao’s weblog for a while now.
The web makes a great place to disseminate information, but it’s just too unstable from a common knowledge viewpoint to refer to most of it in a bibliography.
