The Frozen North
One of the weirder byproducts of having started this blog was being invited to speak at a conference on the Evolution of Mathematical Communication … in Minneapolis … in December.
Not being one to shrink from mere weather, I was enticed by the prospect of meeting Roger Sidge, the man behind Mozilla’s MathML support, Robert Miner and Neil Soiffer of Design Science, and a lot of others whose names I’d heard, but would now be able to associate with a face.
It was quite a bit more fun than I expected.
T. V. Raman gave an inspiring talk. I think he’s more optimistic about the XML-based (AJAX, XForms, …) future of the Web, and what it means for delivering mathematical content, than most. His bottom line seems to be that, at least for our purposes, plugins (whether this one or this one) are no longer anathema and they enable us to deploy MathML and SVG (and whatever fancy technologies come after them) without waiting for a certain dominant browser-maker to catch up.
Andrew Odlyzko gave a beautiful historical talk about the adoption rate of new technologies. Lots of entertaining facts. But the main point, and the fact that astounded me, was that —even today — only about 10% of papers published in Math are available in some open-access form (either from the arXivs or elsewhere). Wow! What could the other 90% of authors be thinking?
Robert Miner gave a demo of MathDex, their forthcoming Mathematical Search Engine. If you’ve ever tried to search for something mathematical on Google, you know that it’s a basically hopeless task. First of all, you need some mechanism for entering the mathematical expression you’d like to search for. Second, just as Google needs to know that “Göttingen”, “Goettingen”, gottingen”, etc. are all the same search term, a mathematical search engine needs to normalize queries for mathematical expressions. It’s a tricky problem, but they’ve made considerable progress (not that you’ll be able to see much on the publicly available website, yet). So far, they’ve just been playing with a database consisting of Wikipedia articles and arXiv eprints. One of the next things they’ll index are the blogs here on golem (a rich source of MathML on the web).
In my talk, I made a pitch for various ways in which what we do hereabouts could be made easier. I talked about the prospects for embedding MathML and SVG in HTML(4/5)1, for being able to copy/paste2 MathML formulæ. My laptop died3 rather unceremoniously before my talk. So, instead of a fancy Keynote presentation, I was relegated to giving a totally retro (and, in the context of the conference, wonderfully ironic) blackboard talk. Surprisingly, it went quite well.
A large fraction of the MathML Working Group was in attendance. And, it appears, they really are serious about trying to come up with some sort of recommendation about input syntaxes. Superficially, that makes a lot of sense. There are, for instance a bunch of TeX-like input syntaxes, and it would be nice if, at some core level, they were interoperable. That is, however, trickier than it seems. Blahtex is designed to be compatible with texvc, as used on Wikipedia. itex2MML is based on WebTeX. There are also a bunch of other converters (TeX4ht, Hermes, …) that are designed to operate on complete (La)TeX documents. Deciding what would be valuable as a compatible subset is not straightforward.
Anyway, I heard some great talks, had some great conversations, and the weather wasn’t even that cold …
1 See, also, Peter Jipsen’s earlier attempt to get something that works in MathPlayer as well.
2 Right now, you can do something really kludgy in Mozilla/Firefox:
- Highlight the text you want to copy (including the MathML formula).
- Right-click (Ctrl-click on a Mac) to bring up the context menu.
- Choose “View Selection Source”.
- Copy/paste that.
Surely, we can do better.
3 Later resuscitated. A word to the wise: a static discharge can turn your laptop dead as a doornail. But, before giving up and replacing the motherboard, try doing a PMU reset. Worked for me.
Re: The Frozen North
Hearing that people are working on these problems makes me feel a little better about everything. I have often wished for interoperable math-description languages and math-savvy search tools, but the current state of affairs is so abysmal that merely being able to type equations in Mark Chu-Carroll’s Good Math, Bad Math would be a great step forward. Just a few days ago, I discovered that superscript and subscript tags do not work in the user comment threads. And we’re supposed to be able to discuss math like this?