stéfan sinclair :: scribblings & musings

Jeff Han has posted an amazing video compilation of demos based on Multi-Touch Interaction Research, a project that he directs and NYU. Much of this just feels like the natural next step in the evolution of the graphical user interface, which of course has habituated us to interacting directly with icons and other graphical objects on our screen. The examples show a wide variety of objects that we could manipulate directly with our hands (rather than with our mouse); reminiscent of the futuristic interface in the movie Minority Report. Some say the next generation iPods will incorporate similar features (this is plausible based on recent patent applications, but of course rumours about Apple products are often wildly speculative).

Thanks to Sean for sending me this link.

ZDNet News has a good article about the ENIAC, the “first” modern digital computer to operate, 60 years ago this month (as the article explains, the ENIAC wasn’t really the first computer, but it’s generally recognized as such, partly because of the attention it received). Of course, the ENIAC is unrecognizable by today’s standards, not least of which because it used a decimal (10 digit) system rather than the usual binary system (even then binary was more common). The article includes a nice inset table for comparison:

clear=‘both’/>The Canadian Symposium on Text Analysis in 2006 will be held at the University of New Brunswick. The Call for Papers is available; see CaSTA 2006: Breadth of Text for more information. This year’s edition of CaSTA bills itself as a Joint Computer Science & Humanities Computing Conference; it promises to be very interesting!

Yesterday (January 31st) The Toronto Star published on its front page a photograph accompanying an article about an altercation between a motorist and a bike courier. The executive summary is that the motorist allegedly threw some garbage out of his van, the courier opened the van door and threw the garbage back in, the driver threw some coffee at the courier and then got out of his van, at which point the altercation between the male driver and female courier became more physical (the details may different according to perspective, but I think the basic idea is there). It’s a remarkable story in itself, but I’m even more intrigued by two other aspects:

• The photographs were taken by a bystander who then posted an article on citynoise.org. The article was apparently noticed by the Star, which proceeded to publish its front page article; a great example of community-based blogging/journalism feeding mainstream journalism.

• The photographs were printed by the Toronto Star without the photographer’s permission, a clear contravention of copyright (in fact, the citynoise.org editor claims in the article that the photographer explicitly denied the Star permission to use the photographs). The Star attributes the photographs properly, but that’s not sufficient, as we remind our student regularly in the Multimedia programme. As someone who likes to use images in my blog posts, my class presentations, etc. (without always the due diligence that could become a full-time job), I won’t be the first to cast a stone, but I find it truly remarkable (and a bit disheartening) that a commercial publisher like the Star seems to blatantly ignore copyright laws when it comes to digital materials found on the web.

Wired has a very brief article about a Canadian study that indicates that users form meaningful and determining opinions about the visual appeal of a website within one-twentieth of a second. That’s fast. Interestingly, the study suggests that there aren’t necessarily obvious principles that determine what makes a site appealing or not. I’m a bit skeptical about this last point: I suspect the researchers may have lacked some of the visual design expertise needed to assess similarities and differences between the sites. However, it does suggest that good design isn’t limited to superficial aspects, and that there’s an interplay between different factors.

After being available only briefly and then disappearing for over a month, the beta of Google Earth for Mac is available again. If you haven’t seen it or played with it yet (on PC or Mac), I highly recommend that you do; it’s like Google Maps on steroids.

Users are able to select what kind of data sources to superimpose over the geographical representation – a form of Geographical Information Systems (GIS) for the masses. I’m especially fascinated by the potential for community-oriented information to be created and disseminated.

I’ve been ruminating lately about how different kinds of web-based text analysis tools might be enhanced (or not) with AJAX support (AJAX, as I’ve described elsewhere, is a combination of web technologies that allows information to be transferred from a server to a browser after the page has loaded, and without doing a page reload).

There’s little doubt that AJAX can be a very useful way to provide a progress or status report during a long text analysis operation (though it’s possible in most browsers to provide dynamic feedback to a user without using AJAX, as my HyperPo tool demonstrates during the initial parsing of a longer document). But I’m more interested in dynamic, real-time interactions where user input affects the response.

For example, a student sent me an interesting prototype of a prototype AJAX translation tool (thanks Chad). Another good example is this spellchecker, which – unlike Google’s – does checking-as-you-type. This got me thinking about whether or not such an approach would be useful for the LePatron project I just mentioned. I’m sure it would be useful, though it would require us to completely rethink the way we log user errors and the improvements that are made during subsequent submissions (this database is invaluable in helping us improve the tool).

As I think more about text analysis tools, I’m convinced that the potential lies not with tools that provide static results (like lists and charts), but ones that are more focused on interactive visualization of texts and data. I’m thinking especially of where the amount of data to process or present is potentially enormous – what is generated and displayed is determined by the user, just as the user only explores one region of Google Maps at a time. What kind of tools correspond well with this vision? I’ll get back to you, more daydreaming is needed…

“LePatron”:http://lepatron.ca/ is a web-based French grammar checker that I’ve been working on for several years with my colleague Terry Nadasdi at the University of Alberta (and he was working on it for several years before I became involved). The tool is designed to serve two major purposes: 1) to help learners of French improve their writing by having them actively participate in correcting their texts (friendly and accessible explanations are provided about various types of errors, but – unlike many commercial packages – users must make corrections themselves); and 2) help instructors of French as a second language avoid having to correct the same mistakes repeatedly.

I’m very pleased to see that LePatron is getting significant use (the site averages several thousand visits a day) and considerable media coverage (on Radio-Canada and Europe 1, for instance; see this article). It’s especially rewarding to be involved in a project that’s getting significant use by the larger community outside of academia.

The Power of Words is a visualization of a handful of speeches by political leaders at moments of crisis – the texts are semantically tagged to enable words to be displayed in different colours and sizes depending on their association and frequency. The underlying methodology isn’t terribly sophisticated, but the presentation is effective; I see both of these as positive traits for an important breed of humanities computing projects. (I should mention for full disclosure that the project was created for my Text Analysis and Programming course, but all credit is due to the students who worked in a very self-directed learning format).

Those interested in visualizing political discourse might also want to visit Parsing the State of the Union, which I’ve mentioned before.

Typorganism is a site that provides several noteworthy examples of interactive kinetic typography (letters and words that move based on input from a user). Even if I find many of the visualizations aesthetically pleasing and generally cool, I can’t help but feel a bit of frustration at the apparent lack of commentary and explanation that accompanies them. For instance, on the first page, what determines where the circles are placed and how they move around? On the “weight” page, what determines the weight of each letter? I don’t think my desire for more documentation (somewhere) is unreasonable, even for primarily experimental and artistic productions.

Thanks to Chad for pointing out this site to me.