Note:These pages make extensive use of the latest XHTML and CSS Standards. They ought to look great in any standards-compliant modern browser. Unfortunately, they will probably look horrible in older browsers, like Netscape 4.x and IE 4.x. Moreover, many posts use MathML, which is, currently only supported in Mozilla. My best suggestion (and you will thank me when surfing an ever-increasing number of sites on the web which have been crafted to use the new standards) is to upgrade to the latest version of your browser. If that's not possible, consider moving to the Standards-compliant and open-source Mozilla browser.
David Berenstein joins the blogosphere.
I gave a talk, entitled Blogs, Wikis, MathML: Scientific Communication in a New Century at the Science in the 21st Century conference. The sound quality is a little dodgy, and the synchronization of the video with the slides could be better but, all in all, it’s not half bad.
Posted by distler at September 10, 2008 12:03 AMTrackBack URL for this Entry: https://golem.ph.utexas.edu/cgi-bin/MT-3.0/dxy-tb.fcgi/1788
Good talk, Jacques! And thanks for mentioning MathPlayer!
Paul
Thanks for the link, Jacques.
By the way, one thing you didn’t mention in your talk sufficiently is that even though the wordpress rendition of Tex is not perfect, it can be seen on any modern browser.
Hi Jacques,
how was you talk with garret?
It was astounding – I sat discreetly nearby them at lunch, and have have never before heard such a resonance of great minds, almost like Pauli and Heisenberg together again for the first time. Complex representations, quaternionic, octonionic, senidenionic, all in a semi-orgasmic flurry too fast to follow.
They continue spending much of the meeting holed up together at a blackboard in one of the many PI expresso machine nooks, scrawling equation after equation and transferring to tiddlywiki, and I’ve counted at least one generation of fermions, possibly three (will try to post a photo of blackboard in follow-up, but still waiting for a rare moment to sneak by when they’re not busily working there).
The local rumor is that they’re planning to write a seminal article together, and even planning to go surfing together in Austin. It’s all very exciting, all very 21st century science, as the meeting advertised.
Who says there are no happy endings?
“surfing together in Austin.”
So, that means what everything you said is sarcasm? Unless surfing together means collaborating…
A nice talk.
A clarification regarding accessibility: as far as I can see, all LaTeXRender pictures automatically come with a title tag written in essentially “plain LaTeX” which I imagine is as easy as would be feasible for a blind person to ‘read’ mathematics.
This also means that the LaTeX images in Wordpress may not be perfect but they are just as indexable as your MathML and more “mathematically searchable” than PDFs.
Also the ‘background’ issue is surely trivially circumvented by changing LaTeXRender to use transparent PNGs. (In fact, it seems that this is what happens - for example, using the different Wordpress themes). Also images do work well with CSS.
You are absolutely right when it comes to scaling and the baselining problems. The other main problem with images is of course bandwidth. And lack of copy/paste is the single biggest annoyance (though this is user-circumventable with some GreaseMonkey).
My feeling is that you will never win the war on Wordpress-LaTeXRender scientific blogs until someone provides a complete hosted solution. Even just installing software is a step too far - blogs should “just work”. Also it is annoying that the necessary fonts are not included with browsers “as standard” but nevertheless it is true, and this is a barrier to more widespread acceptance.
Incidentally, I find it very strange that prominent blogs (Noncommutative geometry/God Plays Dice/Lubos Motl) still use Blogspot which seems incredibly ill-equipped for equations without serious fiddling.
One area which can be developed further is the online equivalent of “standing next to a blackboard, scribbling” (or, LaTeX support for instant messengers). MathIM has made excellent steps in this direction but seems to be quite unknown.
I think there is considerable scope for some very interesting mathematical applications of Ubiquity (the ‘web command line’).
PS - a unrelated ‘usability’ request here and on the n-category cafe - would it be possible to change default behaviour and not make the default ‘subject’ the article title - it is redundant. It would be better left blank by default (or even better, removed - does each comment really need a subject line?).
A clarification regarding accessibility: as far as I can see, all LaTeXRender pictures automatically come with a title tag written in essentially “plain LaTeX” which I imagine is as easy as would be feasible for a blind person to ‘read’ mathematics.
Reading the LaTeX source is better than nothing, but it’s not nearly as good as getting a spoken version of the equation, such as MathPlayer provides.
Still, you’re right, the output of LateXRender is not completely inaccessible to the blind.
This also means that the LaTeX images in Wordpress may not be perfect but they are just as indexable as your MathML
Fascinating! Both the search bar on Terry Tao’s blog and Google will return appropriate hits for the search term
z \mapsto \sqrt{z^2 + 4t}
That’s pretty cool.
What still won’t work is the “Find on this Page” function (Cntrl-F/⌘-F in Firefox). But, admittedly, that’s not nearly as important as having your content indexed by Google.
Also images do work well with CSS.
I think you misunderstood what I meant. You can’t turn an equation using CSS, or change its font-size, or … And you can’t manipulate its content using DOM scripting, etc. Even something as simple as making a term in an equation a hyperlink (which you can do in itex, using the \href{}{} command) is not possible in something like LaTeXRender.
The other main problem with images is of course bandwidth.
I didn’t mention the bandwidth issue because most of us have nice broadband connections most of the time. But, if you ever find yourself at the wrong end of a low-bandwidth link, you really begin to appreciate the problem with having to download dozens of little image files in order to read a page.
Also it is annoying that the necessary fonts are not included with browsers “as standard” but nevertheless it is true, and this is a barrier to more widespread acceptance.
I think that the STIX fonts (which is all that is needed now) are/will be offered as an automated XPI download in Mozilla/Firefox.
PS - a unrelated ‘usability’ request here and on the n-category cafe - would it be possible to change default behaviour and not make the default ‘subject’ the article title - it is redundant. It would be better left blank by default (or even better, removed - does each comment really need a subject line?).
I wish the “Subject” line was more “discoverable” and better used.
When you have a long comment section, it’s really useful to have comments tagged by an appropriate subject line.
Perhaps not every comment needs a subject line. But having one present by default make this feature somewhat more discoverable (since it is not common on other blogs).
I think you misunderstood what I meant. You can’t turn an equation red using CSS, or change its font-size, or … And you can’t manipulate its content using DOM scripting, etc. Even something as simple as making a term in an equation a hyperlink (which you can do in itex, using the \href{}{} command) is not possible in something like LaTeXRender.
Excellent point.
What still won’t work is the “Find on this Page” function (Cntrl-F/⌘-F in Firefox). But, admittedly, that’s not nearly as important as having your content indexed by Google.
My own dirty hack for such situations is to search the source using something like FireBug or any other of the myriad extensions which lets you scan the source ‘in sync’ with the rendered page. Of course, this isn’t a satisfactory solution - and moreover searching LaTeX is fraught with problems - the first which comes to mind is dealing with collisions like g_1 and g_{1}.
When you have a long comment section, it’s really useful to have comments tagged by an appropriate subject line.
Perhaps not every comment needs a subject line. But having one present by default make this feature somewhat more discoverable (since it is not common on other blogs).
I disagree with the first point - I think a far simpler and more effective approach is to use thread numbers - the vertical lines on the left already serve to mark threads out from one another. But that is probably just personal preference.
I was unclear about the second issue. I was advocating simply removing all subject lines for comments (which, of course, requires my disagreement with your first point as a prerequisite). Informally scanning n-category cafe, it seems genuinely new subject lines are used fairly rarely. At the very least, to improve the frequency and discoverability, I would set the default subject to be “blank” and then if still blank under preview have a nonforcing polite sign in red letters alerting the commenter.
Incidentally, I was wrong about transparent PNGs. It in fact uses a hack to try to manually match background colour. But I’m sure transparent PNGs would be easy (easier?) to implement.
It is interesting from a sociological perspective that despite all efforts the web grew fast with hacked together malformed HTML. The proportion of standards compatible sites out there is miniscule. And there is a reason! I fully support your efforts to do things the right way, but it is annoying! I had 4 malformed XHTML errors during the typing of this comment, caused by lack of automatic paragraph tags when copy/pasting into a blockquote. If I didn’t know how (or couldn’t be bothered) to fix it you would have lost a commenter.
I think the history of software, the history of the web, or even just the history of the LaTeXing blogosphere makes it clear that “works easily” is more important (for most purposes for most people) than “works well under the hood”.
Incidentally, I was wrong about transparent PNGs. It in fact uses a hack to try to manually match background colour. But I’m sure transparent PNGs would be easy (easier?) to implement.
Indeed, LaTeXRender does not use transparent PNGs. I believe the reason has to do with Internet Explorer. But the upshot is as I said in my talk — a patterned background, like the one used in my talk, would leave you completely in the soup.
I fully support your efforts to do things the right way, but it is annoying! I had 4 malformed XHTML errors during the typing of this comment, caused by lack of automatic paragraph tags when copy/pasting into a blockquote. If I didn’t know how (or couldn’t be bothered) to fix it you would have lost a commenter.
I think the history of software, the history of the web, or even just the history of the LaTeXing blogosphere makes it clear that “works easily” is more important (for most purposes for most people) than “works well under the hood”.
You would have been interested in the two slides that I skipped in my talk. They were about precisesly this issue and the things I have learned since I settled on the method for ensuring well-formedness, used here.
In Instiki, I think I have something absolutely as effective, but much more user-friendly. And, yes, the software is much easier to install/set up.
LaTeXRender uses LaTeX and dvips to create a postscript file and then leans on the ImageMagick convert utility to turn that PS image into a PNG. At this stage, one can turn a chosen colour transparent using the -transparent switch. A line like the following in class.latexrender.php will produce light text on a transparent background (see, for example, here).
$command = $this->_convert_path." -density ".$this->_formula_density." -trim -transparent \"#FFFFFF\" -negate".$this->_tmp_filename.".ps".$this->_tmp_filename.".".$this->_image_format;
This still has the other problems of maths-as-images, in particular the poor scaling behaviour.