Michael Leventhal
preferred email: (my first name).(my last name)@(my current employer).com
alternate email (might get lost in the spam): michael@textscience.com
Cell: +1 408-386-2608
Senior Director, Tarari, San Diego, California June 2003 - Current
Tarari is a venture-funded (~$45Mil) Intel Capital company which makes hardware and software products for network and application acceleration.
To see my resume at http://www.textscience.com/cv.html click here
These papers are, in my view, among my most important. Taken chronologically one can readily trace the evolution, sometimes by conscious steps and sometimes through the hazards of earning a living, of my thinking about the Internet. XML is certainly the common thread that unites them, in part a hazard of earning a living, in part because I had the good fortune to find something I was good at, and for the third part, because XML proved, contrary to all reasonable expectation, to be a topic of increasing importance as time went on.
My very first interest when I stared to play with the World Wide Web was to create applications that would use it to enable global-scale collaboration. By global scale I mean not just distributed around the globe, but tasks of such grand scale that only a world-wide collaborative framework would be able to accomplish them. One idea that I came up with was to create a massive body of context-specific analysis of literary works and one direction of that analysis would be to support students of foreign languages. Hence the prototype Web application for annotating Chinese language texts for the use of students of Chinese as a foreign language. The wiki is something along the lines of what I was thinking, although with the use of XML I conceived of a much more profoundly structured annotation interface. I believe we are still in the infancy of what we can do to enable collaboration through the web; the highly structured interface is one possibility which may merit further exploration.
The Perl parser I wrote for this article was nothing at all, but it was clever in a small and fun way. This article and the next appeared in the first publication ever on XML, published by the W3C, and played some part in making XML accessible and winning a large following for it among programmers early on.
The second article I had in the first publication on XML, it proved to the perfectly innocent beginning to what was to be central to a long and painful debate and troubled trajectory as XML technologies developed. At the time Web browsers did not support XML and certainly not XML with CSS but at the French company Grif we brought out the first XML publishing tool with the first use of XML and CSS.
I wish now that this had a more serious sounding title, but I remember the presentation of this paper as the best and most dramatic I ever made - and in the fairy-tale setting of Grenada, Spain. This was the first presentation of the mozilla-based browser, DocZilla, I had worked with Finnish company Citec to create - that is, the first not-mozilla mozilla-based browser, actually far in advance (maybe to this day in some things) of the original. We were able to show some amazing interactive web capabilities, very much like what AJAX is trying to do, entirely based on XML, javascript, and the DOM. One of the applications I demonstrated, an application for the annotation of texts, was based on concepts I had developed in 1995 with the CELL.
Probably the most infamous XML article ever written. I argued against the idea of using XSL in Web browsers to make applications interactive, promoting instead the use of the DOM, javascript, and XML-based techniques I had demonstrated with our mozilla-clone. It was a very good article; today, with the wild popularity of AJAX and the extremely rare use of XSLT for interactive web applications, I feel that my position has proven with time to have been quite on the mark.
At the time when Web Services was understood to be a painfully verbose way of serializing RPC, I led the team at Commerce One that created a very early Web Services platform and the only one that was designed expressly for document-centric web services - i.e., to handle large business documents like those in UBL. We came up with many unique solutions including a manifest which solved problems with SOAP with attachments, an XML schema to Java class generator which was far ahead of everything available at the time, and an architecture, designed for pipelined document processing which was several times faster in this use case than anything else.
The tech downturn proved, for me, fortuitous, as it brought me to San Diego to work on XML acceleration in silicon for Tarari and this has led to some of the most interesting and challenging work of my career. Unfortunately, I cannot, for business reasons, write about most of our discoveries and pioneering work and I have no substantial paper on the topic. This paper is extremely thin, having been generated and only slightly filled out from a PowerPoint presentation but there is still much of value in it. It covers the rationale for RAX, an XPath-based programming methodology we invented, stimulated by the need to come up with approaches to XML which are profoundly acceleratable by special-purpose hardware.
Most interest in silicon-based XML acceleration comes from network and security equipment vendors as XML begins to bring profound changes to the demands on the infrastructure and the opportunities for the vendors to offer new types of services. This article is my long meditation on the topic - actually it is four topics in one paper: the impact of XML traffic on the network stack, XML steganography, the potential impact of XML on problems and approaches to security in the network, and the Net Neutrality debate as it might relate to the increasing dominance of XML traffic.