The research area that I affiliate myself with is “Computer–Supported Collaborative Work.” We are, by and large, very interested in ways that people do work together, and we spend a lot of time talking about synchronization and coordination.

In contrast, this project requires you to simultaneously not be online: it’s a collaborative screensaver, which takes advantage of computer location to run F1 cars between desktops. (To see the concept, skip the talking heads, and jump to about 3:50).

(via GizModo, via Richard Banks)

In our upcoming Infovis 2008 paper on “Effectiveness of Animation in Trend Visualization” (sorry, not online yet), my research group explored the idea of animation to show changes in data over time. (Not unlike, say, Gapminder). Not to give away too much, but we found that users found animation disorienting. Still, it seems from watching Rosling’s much-discussed TED talks that when you have a really-well-managed visualization, and you know where the data is going, you can make it comprehensible. (He does a few things: he transforms movement in four dimensions into a linear horse-race: can we get from the top-left corner to the bottom-right?).

Last year, George Robertson and Jeff Heer published a paper on Animated Transition in Statistical Data Graphics. They found that staged animations could help users track changes in visualizations more easily than linear transitions could. Past work, from Robertson (e.g. Polyarchy; Cone Tree) and others—has also found that appropriate animation can help ease transitions.

All this by way of prelude to thinking about applications that don’t properly show transitions. One easy example is Excel. Traditionally, many calculations in Excel are very simple, and so I can see that if I change A1, then A3 will change with it. But spreadsheets are being used in ever more-complex ways. When I run a computation in Excel, numbers everywhere can change—and I have no notice. What just happened? The screen blinked; the old version is gone. The longer the causal chain—the more steps of computation—the less likely I am to be able to decode what just occurred.

Could animation help here? One handful of colleagues have suggested a system called Phosphor. Phosphor shows users what has just happened on their screen by adding visual cues: small highlights that help show what has changed on an interface. It’s not animation, but it’s highlighting where a transition happened. Indeed, in many ways it’s better than animation, as it doesn’t force a user to understand the change as it is happening. It’s worth checking out the linked demo to see just what is happening.

While the demo requires a fairly arbitrary and bizarre set of changes, I suspect that Phosphor-like techniques for the Excel scenario could help this situation. We might even be able to go a step further: for Excel, lots of results are numbers, which (gosh!) move up and down. Could we make semantically meaningful change, perhaps just showing a little arrow next to numbers that shift after computation?

I realize that you can do this yourself. Simply run all your calculations twice—once on the new value, once on the old—and then compute the difference. But visually highlighting change seems like a useful trick.

(For what it’s worth, I could also see ways to apply this to anything where data is changing in parallel. MS Project, when you change one constraint and the rest of the project recalculates? Web pages that automatically refresh?

But we’re left with a second problem. In the Phosphor example, all the changes were happening on-screen. How do we cue a user when a change happens off-screen? In Excel, formulas can span sheets. In Project, a single plan can span hundreds of rows in a hierarchy. Let’s stick with the same goals: I want to allow the user to know where to look for changes, and to give them some sense of what the change was.

There are several ways I can think of offhand. First, you can just point to the changed data. “Wedge” suggest ways that we can cue distance on things like maps; perhaps we can do something similar with a spreadsheet? (This is roughly what some collaborative editing systems do when they put little marks inside the scrollbar showing what has changed). An alternate plan is to try to show the entire space at once. You can zoom way out—showing only change across the whole span—but this forces you to pull the user far away from their original context (which may have been the thing that created the change.) Alternately, you can try to compress the parts you aren’t looking at, by compressing them. This Fisheye Calendar is a pretty good idea of how this could work.

But these are the ideas that are at the top of my head tonight. I’d be very curious about other approaches to this problem. What other ways of visualizing change have you run into?

To cancel my newspaper subscription, I called an 800 number, gave them my phone number, and was done. To cancel most other subscriptions, I do the same.

About six months ago, Golds Gym closed the branch nearest my house. I found that I went to the next-closer one less and less often and eventually, not at all. So in the last few days, I visited the local 24 Hour Fitness, and, while it is a Big Evil Chain Gym, it's a closer one to my house. And then I decided to cancel my Gold's membership.

1. Call Gold's. Get the number for ABC Financial, the fine organization that manages the account
2. Get lost in ABC Phone tree. Call back Gold's, get my membership number
3. Try ABC again, this time with my membership number. Find an actual human... who explains to me that the way to cancel my membership is to send a handwritten, certified mail request to their main office in Sherwood, Arkansas requesting a cancel. They will bill me for the last time within 30 days of that request, which means that the gym membership will conclude another month later. (Sherwood is a few minutes drive north of Little Rock.) I'm not impressed. Clearly, this is not arranged for our convenience--instead, it's meant to discourage closing the membership.

So let me add to the cacophany of voices pointing out that getting yourself into one of these contracts is on the same order of annoyance as getting a cell phone. (For what it's worth, when I was a member of the vastly-more-expensive Pro Club, a Microsoft benefit, it was easy to cancel. Pro Club obviously figures that at their prices, they are charging for customer service. Golds ... not so much.)

I'm back in Seattle, and am starting to dig through my travel pictures. for the first time, I've really dived hard into using photo management software, and I've been enjoying Windows Live Photo Gallery. I'm finding myself using the tagging and sorting extensively, and really enjoying being able to do things like grab all photos from a particular day tagged with a particular term. More, I'm liking the fact that they finally integrated MSR's image stitcher, which is what bumped it over the line for me from Picasa--I select a sequence of photos, right click, and let it go in the background while I keep working. 

This precarious bit is from the Brenta Dolomites, in Northern Italy. The ladder is a 'via ferrata', or iron road, put up for those times when you don't feel like scaling the mountain yourself. 

I also have been taking a little time to play with Photosynth. It's now publicly available, and a lot of fun. I had about five minutes to kill in a piazza on Verona one cloudy afternoon, and decided to see what burst mode on my camera could do for Photosynth. The answer turned out really well, here. 290 photos came out 98% synthy. (Which means that only a handful of photos were not well-integrated into the synth).

Until last week, I was traveling internationally. I spent two weeks doing research at INRIA, just outside of Paris. Microsoft Research has a series of e-science programs that co-fund projects at universities; INRIA is one of our collaborators. I then took some side trips, and made my way to Trento, Italy, where I worked with another of our grantees, COSBI.

The route that we took runs through Switzerland by way of Geneva, and then up and over the mountains. The train ride from Switzerland to Italy took five connections and six different trains, including some regional trains through tiny towns as well as big international trips. Some of the connections were as tight as five or six minutes; others, a half hour. Not a single one of them was long enough to make you feel like you could make the connection at an airport--or, for that matter, any form of American transportation at all. Our timetables just aren't that precise.

So it was a complete surprise that the tightest few connections went off without a hitch in Switzerland. We hopped off one train, crossed the tracks, waited four minutes, and got on the next. The online trip planner for Swiss Rail had the greatest confidence that we'd make the connections, and helpfully printed out the incoming and outgoing platform. It was completely true until we hit the Italian border. The first town that our train stopped at in Italy--I no longer remember the name--was a brief scheduled stop that got extended, twice, and ultimately lasted for forty minutes. Apparently, Italian scheduling really is different from Swiss precision.

(How do they do this, by the way? It's the same train. The same crew. What changes?)

When I asked Italians about this, they laughed. Apparently, Italian trains are notoriously late; Swiss trains notoriously timely; and as such, cross-border trains from Italy to Switzerland get a built-in delay at the border. Sometimes they need the delay; often, they don't. But it allows Swiss Rail to keep their time schedules precise on their side of the line.

i'm actually in Paris now, yes, blogging away at 1 in the morning. I'm out here visitng the AVIZ group at INRIA for a two week visitng researcher shindig, before I go to Trento, Italy for another two week researcher gig. (I'll spend a week getting from one to the other, and doing a little tourism.)

I've been meaing to blog abuot that, but what I actually want to talk about is the fact that my email address was just posted to Craiglsist. This classified ad was posted by xanyelstaples@gmail.com_nospam, while my own email address is xanyel@gmail.com_nospam. Replacing the x with a d. Turns out that the ad then suggests that resumes should be sent to ...

... me. xanyel.

So here I am now with a dozen or so very eager hairstylists from Culver City, CA who feel they could make the best hair assistant ever. Can't delete the ad, although I have written to the other address to complain about it. Maybe this is the same person as tried to sign me up for a dozen email lists last week? Maye there's a disadvantage to having a short name at gmail.com these days...

For the last month and a half, I've been buried under a few tough deadlines, but I'm resurfacing now. I'm at the International Conference on Weblogs and Social Media ("ICWSM"), here in Seattle; I'll be presenting two projects: BLEWS and some qualitative experiential work from my "community consultant" hat.

We're hearing an opening talk from Brad Fitzpatrick, who is talking about the Google Open Social  API. In short, that API is designed to allow sites to expose facts: "Danyel Fisher is a friend of ...", "Danyel Fisher has his home page at ...", etc. He concluded that building a large, interconnected network--where you can go to Twitter and it discovers your LiveJournal friends and helps you subscribe to their feeds--would be much less frustrating. "And then," he said, "we'll see users much less frustrated with these systems, and we'll see a social network that will last more than five years."

Needless to say, people jumped down his throat. The objections came down to two issues:

• How do you plan to handle different parts of my life? How do I separate my motorcycle friends from my blogger friends?
• Is that really the problem with social network systems?

He had answers, but no one seemed really satisfied. And so, it's time for a bet.

Today, Facebook has on the order of 60 million users. Half of whom watch Grays Anatomy. Two years ago, we loved MySpace. And before that, Friendster and Orkut and LiveJournal and Xanga and ...

I believe that this curve is fundamental: that at some point, users get sick of seeing their pictures from the party five years ago (and their friends from five years ago), and go create a new themselves somewhere else. Obviously, Brad disagrees, or thinks that users will be able to answer that with a combination of policy and system hacking.

So. On the one side, you have the intuitions of most people I've talked to. On the other, you have the Wall Street admiration (and valuation) of companies like Facebook and MySpace, Clarke's First Law, and the hopes of a starry-eyes future.

The Bet: In ten years from now--April 1, 2018--we will not have seen a social networking system (in the sense that we understand it today) with a population that will have maintained 60 million active users and continued increasing (or not dropping) for five years or more. I will not count MMORPGs; I will count acquisitions, but only of unique users. (If Facebook buys MySpace, and uses that for growth, I'd count it.) There might be other terms; let's discuss.

Note that this is somewhat different than the question of whether these aggregation systems are a good idea. (I think the answer may be yes: if nothing else, it's annoying to keep checking with Dopplr to see if my friends are using it.) But that's a question about whether how we use these large, centralized address book systems, not whether any of them will last for five years.

After something like a year and a half of preparation, HCI Remixed has emerged from the ether out at MIT Press, and is now available for sale. Mostly for those who care about such things, it's a collection of essays from luminaries (and others) in the human-computer interaction field talking about research they have found inspiring and old (ten years or more) work that they continue to address.

I'm a contributor, so take my words with a grain of salt, but I'm really enjoying kicking back and reading this. Bill Buxton periodically wanders around the MSR campus; it's really interesting seeing how he got his start. (It involves borrowing a motorcycle.) Other researchers contribute stories about things they've found impressive.

As a result, my reading list is a few texts longer than it might have been, and I've got something to chat about with Dan Russell next time I see him. It's also nice to read these people in some form other than the eight-page/two-column/hypothesis-methods-conclusion paper. Some of my colleagues are impressively articulate writers who make their points beautifully.

If you aren't an HCI researcher--if none of the names on the list mean anything to you, and you don't really want to know why reading the Apple Human Interface Guide might have been an inspiring experience, then this isn't really a good choice. But if you want to know which books fell off the shelves to inspire HCI research as it is happening today, then this is a fine text.  Definitely worth checking out.

In the New York Times Book Review from a week or so ago, the special holiday buying guide--for the first time, as far as I can tell--required me to get up from my armchair, technology-free ease and go to a computer. A review of Michael Palin's diaries suggested that we review the classic Monty Python "Department of Silly Walks sketch to remember who Palin is.

Perhaps even more surprising, a review of a biography of the dancer Nureyev (clip) criticizes the biography for under-reporting his appearance on the Muppet Show--not for his pas de deux in "Swine Lake" (apparently the book mentions that) but for the following hot-tub scene afterward.

While I'm accustomed  to the Times Book Review assuming casual familiarity with people I've never heard of, artists I didn't know about, and parts of New York that aren't on maps, I am new to the audio-visual embed. (But maybe I'm just behind the time: a quick search of the New York Times website finds over a thousand hits--and over 400 articles that don't mention "Google" or "video". Is YouTube where we go now for this? If so, it redefines the "in passing" cultural reference: when members of my generation depart, we can now just point each other to links. This might be even a stronger generational identifier than calling out "have fun storming the castle" (proof positive that one was born in the mid-70s to early 80s).

This week and last have been information visualization weeks at Microsoft, as I've been busily meeting with various people to talk about projects I've done, projects I'm going to do, projects they are going to do, and so on. On the way, I've seen a lot of pie charts from various people. A lot. Also things that were 3D, glossy, glowing, and appeared to be made out of Infographics plastic.

And I ran into a marvelous quote (via Stephen Few) ranting about pie charts from Coda Hale.

Piecharts are the information visualization equivalent of a roofing hammer to the frontal lobe. They have no place in the world of grownups, and occupy the same semiotic space as short pants, a runny nose, and chocolate smeared on one’s face. They are as professional as a pair of assless chaps. Anyone who suggests their use should be instinctively slapped.

(He then follows it up with some useful discussion of why.)

TechFest is a big annual thing at MSR -- it's one of the ways that researchers show their work to the product groups. (My manager gives a pretty good overview in this video, although I need to disagree that MSR is "top secret"--we talk a lot about our work.) Right about this time of year, we start submitting our projects into the hopper; MSR Program Managers start their "world tour" in a few weeks to decide which projects will make the cut.

I'm shuffling through my photos from last year, and I ran into this pair that--to me--summarize how you can tell it's TechFest time. This first picture is a Microsoft fridge. You've heard of these. Free Coke, free orange juice, etc. The middle section of the case used to hold Talking Rain cans, but it's now empty.

Why are all the cans gone from this usually well-stocked fridge? Is it that talking at TechFest is thirsty work? Is it that there's been a run on blue and green seltzer in the last few weeks?

No. It's that cans are about the right weight to hold a poster hanging flat.

• Information Visualization Researcher. I need to find a student with background in information visualization, or related work, who will work with me on visualizing some really interesting, very large data sets. These will be used by a real team to solve some really tricky problems, so we'll be working closely with our future users to make sure that they are happy. Expect to spend a week watching them work in their workspace; expect to iterate a lot on various ideas.
• Human Computer Interaction Researcher. Meanwhile, a very different project is just about done--and we need to know whether it does what we think it does. We're going to grab a bunch of volunteers and have them work with our interface, and with the state of the art, and see whether we're any better. The student for this position will have some idea of what a classic CHI lab study feels like--the best choice would be a second or third year grad student.

Both of these are winter or springtime internships--I expect to have other positions open (and it's never too early to get an application started), but what I'm looking for right now is starting in December or January. (Seattle has an incredibly mild winter, plus amazing nearby skiing: it's not a bad place to spend a winter.)

If you think you'd be a good match for either of these, or know someone who would, drop me an email.

[R]ural areas account for only 19 percent of the U.S. adult population, but they also account for 26 percent of the casualties.

It is fairly well known that enlistment is heavily rural (e.g. this article, from the conservative Heritage foundation, has statistics suggesting that recruits are about twice as likely to come from a rural area then an urban one). The naive hypothesis--the one I imagined before the news article started talking-- was that dying in Iraq is pretty much zip-code-blind. So the article must be pointing something else out. There must be some odd little causal thing here. Perhaps enough to be worried about: recruits from rural areas are being trained for more hazardous jobs. Recruits from rural areas are clumsier on he battlefield. Recruits from rural areas are being pushed to more dangerous positions. Something.

Note that the linked article doesn't manage to explain this. It gives us death figures in Iraq per state population--but not per enlistment. Again, the man-on-the-street, this-is-not-a-trick-question answer should be "if you join the service, you have a one-in-so-many chance of dying, whatever your role."

The original source article is deceptive: it's written to be picked up by the press. It talks about the "rural disadvantage" and the "disproportionately high price", although it admits:

The high death rate for soldiers from rural areas is linked to the higher rate of enlistment of young adults from rural America.

And that's when I started shouting at my radio. This is not news. The news story would be if that weren't true--if rural Americans, despite enlisting in far greater numbers, were dying at rates proportionate to their state's population. ("Kansas produces more Supermen: Rural soldiers less likely to die in Iraq.")

I'm a few (or a lot) steps behind. The last time  you got a post from me, I was recovering from a day of travel to Kunming. Now I'm recovering from half a day's bus ride from Dali to Lijiang. In summary:

• Kunming was rainy. I got my head into several museums (posts and more to come), dealt with a transportation fiasco involving a train station, was guided around by a very helpful friend, and ate hotpot. (That's it. No more hotpot.)
• I took a minibus from Kunmng to Dali. Dali is a lovely little town, although it has been discovered by tourists who also enjoy and admire its many charms. I spent one day on the youth hostel's "Round the Lake" tour, which I was surprised to discover was actually very good. The next day, I climbed up into the Cangshan mountains and hiked around the "Jade Belt"--lovely.
• This morning, I hopped another minibus to Lijiang. Lijiang is famous for having become Chinese Disneyland, and I'm afraid it's quite true. Fortunately, you can get out of town easily by bike, and the area around town seems to be very pretty.

The air is breathable out here and the weather has been mild--perhaps mid-60s to 70s, with people sweating and cursing at days when it gets up to 85. (Compare Beijing, where everyone got excited about the heat breaking when it got down to 85.) Of course, it's also been rainy every other day; I fully expect to spend my next few days shaking off water in between bike rides and hikes.

Ok, time to go explore town... more later as I begin to work on my photo backlog.

Travel: it's definitely an experience, and that's not a bad thing--I just have to remember to step back every once in a while, look at it with a good sense of humor, and laugh loudly before plunging in again. When I forget to laugh, I let myself get stressed, and that's not as fun. And so, for example: my cell phone connection to a local tour guide was poor, so I tried to dial from the hotel phone. 9 + the number. 9 + 0 + the number. The works. I couldn't do it. So I called "0", the direct dial operator, and asked. She couldn't get what I meant. So I called "5", the manager, and asked. He couldn't understand the question.

So I walked to the "business and travel center" and showed them the number, and the girl behind the desk looked at it, very confused. Note: 138 is the local cell number prefix. Everyone with a cell phone in Kunming's number starts with 138. This should not have been mysterious. She helpfully dialed it and then charged me an outrageous amount, 10 Y (about $1.50 for a ten minute local call.)

I asked, "If I want to call again, how do I do it?"

She didn't get it.

"Can I call from my room?"

No idea.

I hold the telephone to ear and point to the number and start dialing. She looks confused: why am I doing that?

"I want to call a friend in Beijing. How do I do it?"

"This number?"

"No. What about different number?"

"What number you want to call?"

Ok, forget it.

"Let me try something else. I would like to buy a train ticket." Panic. Flurry. Phone calls. And a guy from downstairs comes up with the train schedule to Biejing, carefully written out. Apparently, the only two words in this conversation that have registered are "ticket" and "Beijing."

The guy from downstairs wants to be helpful. "What number would you want to call?"

"I don't know. I know how to call local. I know how to call USA. How do I call China, but not local."

"Not local?"

"Call Beijing. Call Shanghai." Certainly someone has done this before. "What if I wanted to call it in the morning?"

"Ah. we will call this number for you in morning, then ring your room. What time?"

I suddenly realize that the subjunctive mood has failed me, entirely. Humor is a better choice, and I laugh.

1) When the travel book comments that Kunming is the Seattle of China, you may have thought they meant--indeed, it said--that it was the laid-back attitude and proximity to mountains that you can't see. This may be true. But the rain and pervasive gray might be vaguely reminiscent, too. (Although Kuming's rainy season is Seattle's dry season.)

2) If you have bought your hotel reservation the night before with eLong, and so didn't get a chance to print out the Chinese taxi driver directions, then make sure that the person at the airport information desk writes down Green Lake View Hotel for the driver.  You will be initially pleased by the decor of the place--am I really staying there for $40 a night?--and then it will dawn on you that you are at the Five Star Green Lake Hotel.

3) The desk clerk will laugh pleasantly. "Happens all the time," he'll say, and you'll agree, as he writes down a new name in Chinese. You'll bring it to a cab, drop another 10 Yuan, and feel surprise--and then a sinking sensation, when you realize you just found yourself at the Green Land hotel.

4) The staff will think it's hilarious, of course, when they see your taxi sheet now has both those names on it; they will, however, very empathetically get you to the Green Lake View hotel, which is quite pleasant and was expecting you. As you come into your room, a staff person will even follow you up, bearing a little fruit basket with a crisp pear, a couple of bananas, and some mangos. This will, in fact, make it all better.

5) Dang, but that's some good mango.

So maybe my last entry gave it away. One too many hand-waves. One too many trips back and forth between officials who didn't understand me. One too many times frantically flipping through the phrasebook at the back of the Rough Guide trying to make my point. And one too many meals of strange stuff mired in goo.

Today I gave up. At the advice of the wise and elegant Kim, I canceled all my afternoon plans, and checked into a fairly good hotel, slept for much of the afternoon, took a long bath, and watched a James Bond movie. I went to dinner at "The Wine Library", which serves decent Australian wine and pretty standard pasta, and plays American Jazz classics. Listened to Louis Armstrong and ate pesto with a fork. It helped. I'll sleep well tonight; tomorrow to Kunming.

The group in the restaurant was interesting, by the way. Two completely different populations: tables of Chinese people, eating exotic. Some of them even using a fork. And a few sets of Westerners eating comfort food: a jetlagged pair of Australians sat down at the other end of the bar from me.