On News Visualization, Part II
This week’s Loose Wire column in WSJ is about visualizing news. Researching the column I had a chance to interview Craig Mod, the guy behind the excellent Buzztracker. Here’s an edited transcript of our chat:
Craig Mod: We have over 550,000 articles in the DB now, spanning back to Jan 1st 2004. “Buzztracker” went from 750 hits on google the day before the launch to now … 39,000+ which was suprising
Jeremy: when was the launch?
Craig Mod: About 3 weeks ago
Craig Mod: got slashdotted within 12 hours
Jeremy: could you walk me thro how you think people might use it, or derive benefit from it?
Craig Mod: sure. the project started about 2 years ago as a pure art project .. some of the original output was just the dots, with no map .. but the closer you looked, suddenly land masses began to emerge and you started forming associations
Craig Mod: I’ve obviously tried to make it a lot more pragmatic and functional now
Craig Mod: fundamentally it’s supposed to get people thinking about why these connections exist — why is Shanghai and Canada connected (during the SARS outbreaks)?
Craig Mod: How did the virus spread?
Craig Mod: What sorts of checks can you preform to prevent that sort of spreading?
Craig Mod: Is it possible?
Craig Mod: etc etc
Craig Mod: and from there begin to explore how these events are being covered
Jeremy: interesting.. is there a page for the SARS stuff in the archive?
Craig Mod: clicking on the locations obviously gives you a list of the articles they appear in
Craig Mod: unfortunately the SARS stuff happened when I was building the beta 2 years ago .. so it’s not in the current DB
Craig Mod: but the recent demonstrations in China have popped up a lot
Craig Mod: there’s a China-Tokyo-Jakarta triangle that appeared during the summits
Craig Mod: and you can click the “tomorrow / yesterday” buttons and see just how long these stories linger in the collective media conscience
Craig Mod: which is kind of fun
Jeremy: is there a danger the external links die off?
Craig Mod: There is .. and we orignally had links to our internal cache but .. obvious copyright infringements issues scared us away from keeping the feature on new articles
Craig Mod: although, we still have all the data, of course
Jeremy: yes, the copyright thing is tricky…
Jeremy: how do you plan to deal with that?
Craig Mod: By not publicly offering the articles
Craig Mod: And by keeping advertising off the site .. keeping it as pure an art project / public service project as possible
Jeremy: tell me a bit about you.
Craig Mod: I’m 24
Craig Mod: Born in Hartford, CT
Craig Mod: graduated from UPenn 2 years ago — degree in Digital Media Design (BSE in Comp. Sci with a very strong Fine arts component)
Craig Mod: Came to Tokyo 4 years ago for a year abroad, came back 1 1/2 years ago to run the Tokyo component of a small publishing company I helped start
Craig Mod: So a total of 2 1/2 years in Tokyo
Craig Mod: 2 years of which was spent at Waseda University in the intensive language program
Jeremy: how’s your japanese now?
Craig Mod: Extremely functional but I still can’t “relax” with a novel (although I just finished Murakami Ryu’s Almost Transparent Blue in Japanese)
Jeremy: so what are your plans for buzz?
Craig Mod: Right now I’m working on re-writing the drawing routines in a more power language .. the plan is to produce super-high-resolution prints for gallery display
Craig Mod: but being the only guy working on this + running sales / pr for CMP in Tokyo means it unfortunately takes a while to rewrite components
Jeremy: when you say hi-res prints, you mean of the maps?
Craig Mod: Correct
Craig Mod: There is a lot of information being lost in the low resolution of comp. screens
Craig Mod: especially Buzztracker connections (the thin, light lines get lost)
Jeremy: with thinking gap donned, where do you see this kind of thing going? do you think as people turn more and more to the net for news, these kind of visual displays will catch on?
Craig Mod: I don’t think traditional news delivery will be subverted anytime soon, but I do think that as digitized nformation increases (digital photographs, journals, etc) people are going to need clean, effecient methods to engage with the data / find what they want
Craig Mod: Something like buzztracker is an attempt to both clean up the delivery of a tremendous amount of information while also brining to the surface patterns otherwise invisible — missing the forest for the trees, etc.
Craig Mod: but what I’m hoping … what I had in mind as I was designing and building the information structure of buzztracker was that things need to be as clear and simple as possible
Craig Mod: this isn’t meant to provide an incredibly exhaustive set of news mining features — it’s meant to be highly accessible by anyone
Craig Mod: I haven’t seen any of the other newsmap interfaces but perhaps unlike Marcos’ work or, hopefully, mine, their information architecture wasn’t as transparent
Jeremy: transparent meaning?
Craig Mod: meaning, they innundated the user with superfluous interface elements, cluttered typography, illogical hierarchies .. I don’t want anyone using buzztracker to be concerned with how they engage the software/site .. the focus should, I hope, be engaging the data, the news
Craig Mod: (although I don’t know if they did that since I never saw any of them 🙂 )
Craig Mod: on the tech side of things, there was a point where I was debating between flash and pure html .. in the end, I think going with html made sense for those exact reasons — quick loading, standards based, etc
Craig Mod: There’s also, I suppose (to a small degree) a sense of bias being eliminated in these sorts of ways of navigating the news ..
Jeremy: very true.
Craig Mod: But almost unavoidable .. but those biases are also interesting ..
Craig Mod: buzztracker being completely rooted in anglophone news sources
Craig Mod: you start to see things like .. Africa doesn’t exist in the mind of enlgish speaking sources .. most all news takes place on a thin line just above the center of the map
Craig Mod: Animations are also comming .. along side the high-res output ..
Jeremy: how would the animations work? evolution of a story over a period of time?
Craig Mod: you could follow certain keywords — allowing you to follow certain stories .. You could also map the news on an hourly basis — interpolating the rise and fall of events smoothly ..
Craig Mod: the thing with the animations is that, I believe, by watching repeated time lapses you’ll start to see “news rhythms” erupt ..
Craig Mod: which begs the questions — if you map these animations to sound, can you decern other patterns that you were missing visually?
Jeremy: what about some of the criticisms that you’re leaning towards datelines, and so stuff like the tsunami wasn’t represented properly?
Craig Mod: There are some events (like the tsunami) which appear after the day they happened .. one of the best and worst parts of Buzztracker is that it’s fully automated so if something doesn’t appear when it “should” that’s representative of the media in some ways
Craig Mod: The spain explosions last year are incredibly represented
Craig Mod: I think some — such as false results, or skewed distrobution in the wrong ways — could be corrected by simple human intervention .. Looking for, spotting these “errors” in calculation, and adding rules to fix them
Craig Mod: but at the same time, that takes away from a bit of the purity of the automation of Buzztracker .. it’s always about balance I suppose