- From: Stan Devitt <stan.devitt@gwi-ag.com>
- Date: Tue, 1 Apr 2008 08:05:49 +0200
- To: <m.kohlhase@jacobs-university.de>, "Bruce Miller" <bruce.miller@nist.gov>
- Cc: "Henri Sivonen" <hsivonen@iki.fi>, "David Carlisle" <davidc@nag.co.uk>, <ian@hixie.ch>, <public-html@w3.org>, <www-math@w3.org>
I have been watching this debate for sometime now and have been somewhat surprised by a number of things. 1. Math presentation and semantics are somewhat unique in that we are exposed to many to one mappings in both directions. An incredible (to me) amount of this discussion appears to have ignored that. We only saw the tip of the iceburg when the topic of notations for integers came up. 2. A relatively unique but apparently little understood aspect of MathML is that it is actually a crucial enabling technology. It allows (but does NOT force) the author to explicitly state the semantics that is meant by a particular picture (presentation). It should be no surprise that implementation the implementation of that is somewhat more complicated than an attribute value. TeX/Latex did not really formalize that. Nor did any of the other predecesors. This formalization of multiple representations of an object - each potentially enhancing understanding is important. No other widely adopted markup for mathematics has made that possible in any (semi) formal way and the difference between possible and not possible is enormous. It actually allows for notation to be specified and re-used - in just the way we do Mathematics. 3. This technical possibility for the author to "clarify what they mean" is extremely valuable as we move to a semantic web. Reverse engineering from presentation can only go so far, and is especially dangerous in the context of mathematics. The basic concept applies to all aspects of the semantic web - not just mathematics. Only the author can really know. 4. Given that HTML5 is intended to be a replacement/evolution of HTML in an era when the semantic web is coming of age, I am very surprised that there has not been more focus on enabling that (no doubt another forum). I think there is a huge obligation to put in place the mechanisms for more precise semantics given the huge effort that is needed to update such infrastructure. By all means explore other mechanisms, but it would be a travesty to go another ten years with no infrastructure in this arena. I understand that at this point in time only some tools and a small percentage of the user base actively participate in using such mechanisms. However, providing mechanisms the data to be automatically identifiable and consumable should be fundamental to evolution of the web - not just mathematics. By all means provide a better way, but provide something. To adopt the "One semantics fits all model" is to role back the clock 10 years and ignores mathematical reality. --- The other aspect of this discussion that has been somewhat troubling has been the reluctance to treat mathematics as anything special. Historically, text, mathematical notation and annotated images have been cornerstones of communication and mathematics as one of the core languages. Surely that is no accident. Of course the challenge is to strike a balance, but I am deeply concerned by what I see as a tendency to ignore or or down play these structural issues. Of course LateX authors were/are mostly worried about presentation - no one had made it possible before for the data to be automatically consumable. That can take some getting used to, and it raises all these issues. Either way, it is dangerous to argue requirements here by historical precedent. Designing a product solely by user surveys and without presenting the possibilities never leads to very satisfactory product design. I have great sympathy for the desire to make simple things easy to do, but that doesn't need to mean closing off the richer possibilities - possibilities essential to our growth. Stan Devitt (speaking for myself)
Received on Tuesday, 1 April 2008 06:06:30 UTC