XTech 2005

At first glance, XTech could be described as "WWW Lite", but a better conference for the development and user foci. A reasonable subset of the normal cast of characters are here, including John Wilbanks and Ivan Herman (W3C), Dave Beckett (on Redland), Jean Broekstra (Aduna/Sesame). Alistair Miles (on SKOS). Libby Miller is here, chairing a session. The Mozilla Foundation is well represented by Mike Shaver and Ben Goodger (lead engineer on Firefox). Dominique Hazael-Massieux is presenting on GRDDL.

Michael Kay has founded a new company, Saxonica, creating both the free Saxon processor and a new commercial one. He presented a comparison of XQuery and XSLT (2.0) in which he claimed to be evenly biased. Interestingly, he suggested that Saxonica (and other companies) would start to produce products supporting a fusion of the two languages for commercial advantage.

XQuery has been designed to be small and optimized for database queries. XSLT has more features, such as templates, formatting, regexps. XSLT is a flexible, dynamic language, which is exactly what you don't want in a database query language like XQuery.

XSLT is stronger on:

• rendition


• up-conversion


• documents

XQuery is stronger on:

• optimization


• structured data

Ben Goodger gave an introductory talk on XUL, Mozilla's XML-based UI language. XUL uses GTK's Flexible Box Model, which I have always blamed for GTK's widget/panel initial sizing issues but he raved about it. The coolest thing I heard was that one can write apps as if one were scripting a Web app, but bind it to XUL. I will have to try writing a XUL app.

He recommended two books for XUL development: Rapid Application Development with Mozilla (for full XUL apps) and Firefox Hacks, which includes a couple of chapters on XUL extensions for Firefox.

The Most Important Meme?

Hmmm. New research seems to suggest that human settlements predated agriculture, perhaps by a long time. Permanent hunter-gatherer settlements at Ohalo II in Israel (23,000 years old, with huts made from brush plants), as well as the Natufian settlements in Jordan and Israel (14,000 years old) strongly predate the first agriculture (currently dated 11,500 years ago in the Levantine Corridor).

If hunter-gatherers were living in permanent settlements for a long time before agriculture, we have to ask why. Agrilculture has generally been presumed to be the answer, not some Johnny-come-lately.

Ian Hodder, dig leader at Çatalhöyük in Turkey has suggested that the residents there settled due to religious convictions. That is particularly interesting, since Joseph Campbell has told us that those societies were almost certainly maternal and worshipers of the lunar bull cult (a flavor of earth goddess worship).

Could it be that women caused the first human settlements to occur and agriculture came later, to help feed people who were already stuck in one place? That would certainly explain a lot, especially the population explosion dates and the lack of significant climate change in the Fertile Crescent. If so, human settlement would be the first, and perhaps most important, of the Big Memes.

FOSSTEC is a Farce

Wow. I started the process of submitting my accepted paper to FOSSTEC 2005, but started to get suspicious when the online registration seemed geared toward paper authors. I Googled it and found this post claiming the conference is, in fact, a scam. *Sigh*.

At least I have a paper all ready to submit elsewhere :)

More on Tiger

The 10.4.1 update is out. Yay! I'm told I should be able to sync now.

Why doesn't Safari report the Slashdot RSS feed? It is certainly there at http://slashdot.org/slashdot.rss. Safari will render it if clicked, but does not report it in the bookmark summary. I wonder if Wired, CNET, NYT, etc are hard-coded somewhere?

I upgraded to iClock version 2.1.9, which fixed a critical Tiger bug (the multi-timezone clocks in the menu bar were not updating). It is a handy utility, which I recommend for anyone dealing with multiple timezones regularly.

Interview with Really Strategies

Ed Stevenson, Director of Content Strategies at Really Strategies interviewed me last month for their "5 Questions" series. The results of that interview are here.

Cancellation of Smart Content Tutorial

I am sorry to say that the Smart Content Tutorial I was to participate in at The Gilbane Conference on Content Management in Amsterdam, The Netherlands on 24 May has been cancelled. I hope to have the opportunity to work with Wernher Behrendt, Mills Davis and Steven Newcomb sometime in the future. I will still be going to Amsterdam for XTech 2005.

Steve Newcomb's Versavant, a way of describing Topic Map Applications in accordance with the draft ISO 13250-5 Topic Maps Reference Model has been released on Sourceforge.

Named Graphs at WWW 2005

Jeremy Carroll from HP presented his paper on Named Graphs, Provenence and Trust. His co-authors were Chris Bizer (of D2RMap fame), Pat Hayes of IHMC and Patrick Stickler of Nokia. I am pleased to see that Tucana's decision to name graphs (in 2001) has finally been given some support from people who have done the math. Many of the potential applications that Jeremy suggested, such as access control, passing of graphs, referencing of graphs in RDF, etc, have been implemented for some time in Kowari.

WWW 2005 has not yet put the paper on the Web, but an older version of the paper appears to be here. This paper was (wrongly, I think) originally rejected by ISWC 2004.

It is interesting to note that a couple of students from Italy had a poster showing how one could define a minimally self-defined graph (MSG) and digitally sign it. They did not resort to naming graphs to allow the definition of the graphs. Their paper is here.

Paper Accepted at FOSSTEC 2005

I had a paper accepted at FOSSTEC 2005, the International Symposium on Free/Open Source Software, Technologies and Content. The conference will be held in Orlando, Florida from 10-13 July.

I had submitted an extended abstract for a paper entitled, "Open Source Software Strategies for Venture-Funded Startups". Now I have to write the paper and submit it by 23 May. Hmm... It is a good thing I have some airplane time coming up.

WWW 2005

Today was the first day of WWW 2005 in Chiba, Japan. The panel I convened ("Can the Semantic Web be Made to Flourish?") went well, but I learned a lesson. I had organized some questions in advance, distributed them to the panelists and used visual aids during the session, in accordance with guidance from the panel chairs and discussion with the panelists. After all that, Jim Hendler told me after the session that the way to do it is to simply introduce the panelists and take questions from the floor! Next time I'll know :)

I had expected the conference to be primarily attended by Japanese and had prepared accordingly. I suppose I had that impression due to the cost of traveling to Japan and the fact that many people I know from W3C didn't come. Again, I was wrong. WWW 2005 is truly international, with over a thousand attendees.

Bernadette and I spent four days prior to the conference in the lovely mountain village of Nikko, about 100 km North of Tokyo. It is a very historic place, where the masoleum of Tokugawa Iyeasu is located. He was the first of the Tokugawa shoguns in the early 1600s and the basis for James Clavell's novel Shogun. The Shinto and Buddist temples are amazingly ornate. I'll post some pictures when I get home since I forgot my USB cable and can't dump them from my camera :)

More Tiger Problems

Can anyone get iSync 1.5 to work in Tiger? It just crashes for me. I also lost the ability to print to my home printer, an HP Officejet d145 reached via JetDirect socket (the dreaded Error -9672). Grrrr.

A Whimpering Tiger

Hmmm. I have Mac OS 10.4 installed and a Tiger in my tank. Dashboard is useful (of course, the KDE guys had this one figured a long time ago, although not as nicely packaged). I have yet to be impressed with Spotlight. My internal drive took about 10 hours to index. Searching for something simple (I chose "snow") worked well. I immediately found pictures, etc, related to snow. Next I tried a "real-world" search: patents. I have several patent applications on my machine and lots of supplementary data, including many mail threads. This time, I was not so lucky. Although Spotlight returned the right Folder, indexing of mail seems to be both slow and very incomplete. After an hour of searching, I had to rely on more manual means of finding the appropriate email message. That is not encouraging. Did you know that Spotlight will, by default, carefully index Mail's Junk folder? That's just not thinking. All four mail messages regarding patents that it did find were spam and filed as such in Junk. Sigh. And, while I'm ranting, who decided to index fonts? Why is that useful? I turned it off pretty quick.

I think we all knew better regarding Spotlight. Right? It is technology, after all, and not magic. We know how they do it and we know they don't really have the data to make the kinds of decisions they marketed. It couldn't stop us from hoping, though :)

Tiger also broke the three mail bundles that I used (GPG, Mail Prioritizer and Mail Appetizer). I had to uninstall all of them before I could get Mail 2.0 to work. That's OK, I can live with Prioritizer now that Mail supports mail priorities natively. The lack of GPG/PGP mail could bother me. I hope the authors catch up to Tiger rapidly.

Desktop Manager, a nice virtual workspace manager for OS X, thankfully supports Tigers (mostly). Transitions no longer function, but at least it works.

I haven't tried Automater yet, but plan to. I hope it lives closer to its hype than Spotlight.

Reading RSS natively in Safari is cool. I'm still not sure it is as cool as Apple have claimed, but it is useful.

In summary, don't coming running to Tiger if you want stability or have a need for every app that you are currently using. But if you are a gadget junkie, the wow factor is enough there to be worth the upgrade. I won't go back, but am definitely looking forward to the next few OSS upgrades. At least Terminal is still fully functional!

Tiger is Here!

My copy of Mac OS X 10.4 ("Tiger") just arrived by FedEx. It looks awesome. Unfortunately, I am off for a long weekend with my family in under an hour. Let's see. Family...new OS. Hmmm. This is harder than it should be...

Speaking Engagements for May 2005

May looks to be a busy month! I'm only taking two trips (Japan and The Netherlands), but will have multiple speaking engagements in each place. Here's the list:

WWW 2005 in Chiba, Japan:

• 10 May, 10:50-12:20. Panel, "Can the Semantic Web be Made to Florish?" (convener).


• 13 May, 10:30-12:00. W3C Session on the Semantic Web. I will be presenting some information on existing SemWeb applications.


• 13 May, 16:30-18:00. Questions & Answers to the W3C Members and Team.


• 14 May, time TBD (morning). Kowari demonstration at the Dev Day.

The Gilbane Conference on Content Management in Amsterdam, The Netherlands:

• 24 May, 13:00-16:30. Smart Content Tutorial (with Wernher Behrendt, Mills Davis and Steven Newcomb).

Vrije University Amsterdam:

• 26 May, time TBD (afternoon). Seminar on a SemWeb topic to be determined.

XTech 2005 in Amsterdam, The Netherlands:

• 27 May, 09:45-10:45. Paper presentation: "Kowari: A Platform for Semantic Web Storage and Analysis" (written with Paul Gearon and Tom Adams). A preprint is available here.

Doxygen for Kowari v1.1

I have been playing with the generation of software relationship graphs (collaboration- or call-graphs) for Kowari. I have pretty much decided to use Doxygen and GraphViz to generate them. Along the way, I generated Doxygen documentation for Kowari v1.1, thinking that others may find it useful.

Preserving XHTML 2.0 Link Relationships in HTTP

Those who know me understand (and learn to live with) my insistence that the Semantic Web will not flourish until simple means to use it are put in the hands of HTML authors. I have tried again to push this viewpoint through a paper I just submitted to Extreme Markup Languages 2005. The basic idea is to pass the link relationship information from XHTML 2.0 via HTTP headers, thereby preserving that information for processing by a Web server. An RDF triple may thus be formed at the Web server from the REFERER information, the link relationship and the link requested.

I don't plan to publish the paper until it is accepted somewhere, but I will be glad to give a copy to anyone who asks me directly.

Star Wars: Revelations

Wow. I just saw the professional-grade fan movie Star Wars: Revelations (Direct download: here). We often talk about the PC Revolution and how big corporations no longer have the lock on large creative works, but I was very impressed. The movie is 47 minutes long and was made in and around Washington, DC by volunteers.

Bushonomics

I was utterly stunned today to hear that the President of the United States of America said that the Social Security System was broken and required repair because it was backed by US bonds instead of cash. This was according to an NPR newscast. Wow. The President thinks that US bonds are just paper? The only security backed by the very existence of the US Government? It is not that I am liberal. I'm not even a Democrat. It is just that such a thought is, um, stupid...

Where does G.W. Bush think your money goes when you put it into a bank? Does he think it sits there, in cash? Or just maybe does he realize that it gets circulated, including investments in bonds?

For those of you who think this is an anti-Bush blog, my advice is to wait until the next Democratic president does something stupid. I'll coment on that, too :)

TKS/Kowari Non-Commenting Source Statements

TKS/Kowari NCSS,
originally uploaded by prototypo.

This image shows the Non-Commenting Source Statements (NCSS) for the TKS and Kowari projects from 2002-2004.

TKS versions 1.0 and 1.1 are shown together under the TKSv1 label. TKS versions 2.0 and 2.1 are shown together in the accumulated bar graph to the right.

Non-Commenting Source Statements differ slightly from Source Lines of Code (SLOC). SLOC counts physical lines (and is therefore more impacted my formatting), whereas NCSS counts source statements regardless of formatting. So, NCSS will always be less than or equal to SLOC for a given code base.

NCSS for TKS/Kowari was generated using JavaNCSS for Java files only.

The # vs. / Argument Continues

The W3C Semantic Web Best Practices & Deployment Working Group formally weighed into the Technical Architecture Group's httpRange-14 issue last week. I sent this message, noting the unanimous decision by the SWBP to stand against Tim Berners-Lee's position on the issue. As expected, TBL fired back. My response, hopefully clarifying the SWBP position, is here. The full thread is in the SWBP mailing list archives for March, 2005.

It looks like httpRange-14 will be with us for a while yet :)

Playing the Odds; Venture Capital as Roulette

A recent survey of start-up companies in the Mid-Atlantic region of the United States by Core Capital produced some interesting results. It seems that only 1 in 27 companies which successfully raise an initial ("Series A") VC investment will survive to get a subsequent ("Series B") round. I heard that the number being used in New York City this year is 1 in 30.

Those odds contrast quite severely with the traditional rule of thumb of 1 in 20 (or more markedly with the rule being used in the Dot Com days of 1 in 10).

Let this be a lesson to budding entrepreneurs: This is a tough market. The odds are that venture capital should be avoided. It is expensive money. Only those who don't need the funds can possibly afford to take them. A better avenue for most entrepreneurs is clearly to shift business models, find new customers and wait for better times to bring a new type of product to the market.

That is not to say that my advice is better for VCs. It is not! VCs make money when the 1 in 30 hits. They need to invest in order to have a chance at the occasional return. No, I am making a suggestion solely for the benefit of entrepreneurs. After all, it is your company which will go out of business at the end of the year, not the investment fund.