Shortlog - a log of everyday things



Got back from UIST last night. Storytime! (Note: post-dated. I'm really bad at getting these entries written on time.)

Sunday: What a terrible morning. I managed to hit my snooze button 9 times unconsciously before finally waking up 70 minutes later than planned. I was supposed to wake up at 7:15, so I'd catch the 8:08 BART to Oakland and catch the 8:50 Coast Starlight Amtrak train (which leaves once per day).

I looked at my alarm clock, and the last digit was right. Unfortunately, the other two were too large. I had missed the BART. Next train left Berkeley at 8:38, and arrived in Oakland at 8:51 - too late. ****.

Turns out I wake up very very quickly and become quite alert when I'm dosed with adrenaline like that. It was a good thing I had packed my bags the night before - I threw on clothes, looked up driving directions to Jack London Square on Google Maps, and snapped a picture of the directions with my phone. Google estimates the route at 17 minutes. Sounds doable. I drove as quickly as safely possible to my destination.

I arrived by the station, but didn't know where I was supposed to park. There were garage buildings, but they all appeared to require permits. Driving around frantically, I finally saw a sign that read "Amtrak Parking" and followed it. As I crossed the tracks, I saw the train I'm supposed to be boarding waiting. I got quite nervous as I had to drive further away from it.

I turned into the nearest parking lot...which happened to be owned by Fox News, and gated. I exited that parking lot and entered the next one, which had a pre-pay meter where you buy a permit for X hours and display the receipt on your dash. Great, except that it only lets you pay for 72 hours in advance. I was going to be gone for ~86. I decided I would worry about that later, and right now I needed to catch my train. I bought the 72-hour permit, unloaded my backpack and suitcase, and sprinted as fast as I could.

I all-out ran a football field and a half to the four-story skybridge that would take me back over the tracks. I reached the base right as the elevator started going up. Guess I'll take the stairs. I was breathing heavily by this point, but I couldn't stop - I had to catch that train. I took the stairs back down on the other side, and ran for the end of the train where I saw a uniformed someone standing. He asked if I had a sleeper car ticket. Nope. Wrong end of the train.

Which was fine, because I didn't have my ticket anyway. The woman at the other end of the train informed me that 1) I needed a ticket, not a confirmation code, so I'd need to go inside and print one off, and 2) She had to leave. So I left my bags right there, ran in, printed my ticket, and ran back. I collapsed into an empty seat, lungs burning, asthma attack setting in. Within 30 seconds, the train started moving. I was exhausted. But I was on board.

The rest of the train ride was fairly uneventful. I used my N900 to set up a wireless hotspot for myself and Valkyrie, and I did some reading. Lunch was expensive and the burger was a bit dry, but I've had worse.

We arrived so early that the previous Amtrak train was blocking us at the station in Santa Barbara. I checked into my hotel and dropped off my bags, and we walked the ~2km along the beach to the conference hotel to collect our registration materials. All the food was gone, so we went to El Torito for dinner. I told everyone about the origins of the CD-ROM boot standard by the same name.

Monday: overslept again, but it just so happened that the session I was late to was the crowdsourcing session, so I wasn't too heartbroken. UIST is a single-track conference, which threw me off, initially - there's not a whole lot going on at any given time other than the paper presentations or breaks. On the upside, you get to see all the papers. On the downside, the long sessions get somewhat grueling.

I particularly liked Leila Takayama's talk about the (in)effectiveness of sidetone for telepresence and the MIT folks' Real-Time Collaborative Coding in a Web IDE talk. I want better tools for collaborative reverse engineering.

Björn, Wes, Meredith Morris from MSR, Thorsten and Jan from Aachen, Valkryie, and I all went to the nearest restaurant for lunch (which happened to be that same El Torito). It was a pleasant discussion - I really enjoyed talking with Thorsten about differences between the US and Germany/the rest of Europe.

Monday's keynote by Ge Wang on his iPhone Ocarina and ChucK and thinking about time-based concurrent programming was fascinating. I thought back to the time when I was chatting with Dave Thomas about it - he found it neat and was tinkering with randomly generated melodies. I wound up using ChucK to make TDMF tones - at some point, I'd like to be able to identify phone numbers by ear.

We presented our project - the Moussage mat - at the Student Innovation Competition. We take input from a Microsoft TouchMouse, transmit it over the network, map the touch points into touch intensity within 12 zones, and send 12 values to a microcontroller, which drives 12 vibrating motors embedded in a sheet of foam that you're lying on for a human-controlled massage experience.

It was very well-received - we wound up with a bunch of people claiming they would just lie down and sleep there for a while. (Nice try.) By the end of the night, people were coming over, saying that their peers had told them to come see our project. People were amused that we had used bullet vibrators as motors. They're cheap (the ones we got were ~$4 each), small, well-encapsulated, and reasonably powerful. In other words: the perfect off-the-shelf part for our project.

Sadly, we managed to burn out the power supply that we were using to drive all the vibrators about 70 minutes into the three-hour demosession, but that was after the judges (and most of the people that would visit our demo) had come through, so it turned out okay.

We're in the process of preparing a demo video; I'll probably make a decent writeup of the whole thing and post it as its own subfolder on my site. Later.

Tuesday: More fine talks. I liked the Stacksplorer talk. Julia Schwarz likened dealing with multitouch APIs as trying to do a lot of things at once, like juggling - and then proceeded to juggle 5 balls on stage. The round of applause was merited. We talked after the session about other approaches people have taken and how all "potential action" previews need to be free of side effects. Other cool talks:

After the talks and a brief town hall, it was dinner time. I had several helpings of the delicious salmon. The keynote speaker, Dan Jurafsky, gave a great talk that explored language and what we can learn from its history. We traced through the history of "ceviche" and "ketchup", explored the difference in words used to advertise cheap and expensive potato chips, and learned some decent predictors of interactions from a corpus collected from a speed-dating session. I will be hopelessly analyzing all my conversations henceforth.

After dinner came thanks and awards. Our Moussage mat won 2nd in the "Best Implementation" category of the Student Innovation Contest! I was thrilled. After the level of positive reception from the previous night, I somewhat expected we'd win something, but it was exciting anyway. I wound up chatting for a good while with Juan, the guy from København whose project beat ours. Maybe I'll see him during my Europe trip this winter.

Since it was our labmate Shiry's 32nd birthday, Bjoern invited us back to his hotel room for birthday festivities - the party included Bjoern, Shiry, Maneesh, Wes, Peggy, Valkryie, Nick, Manas, and myself, all from the BiD lab. Manas brought along Jen Fernquist. Fun discussion was had - politics, economics, and all sorts of things I usually don't chat about with my labmates.

Manas, Jen, and I went to the Student Volunteer party after that. Katrina, the head SV who I'd met in May at CHI, had told me to come and that I should feel free to "bring all the cool people." Most declined. I volunteered to be the designated driver - I've never been much of a fan of alcohol. Others at the party were.

Wednesday: I was originally going to take the Amtrak back on Wednesday, but then I found out that the 3D interaction session wouldn't happen until after the train left. That session included the much-anticipated KinectFusion demo. Fortunately, Manas had driven down and was kind enough to offer a ride in his car back. So I got to see all the fancy demos!

Since relatively few people get to go to these conferences, let me say very clearly: KinectFusion works just as well in person as it appears to in the video - 30 FPS with all the bells and whistles. It crashed a couple times, but I'm not going to complain about it. Very impressive work.

I liked the concept from 1 Thumb, 4 Buttons, 20 Words Per Minute: Design and Evaluation of H4-Writer of making Huffman trees with four branches at each node. It makes me want to learn all the DTFM tones by ear and use them as a higher-data-density Morse code.

The Toucheo demo was just awesome. Multitouch + 3D display, with somewhat sane surface manipulations. If they throw in a Kinect for direct 3D interaction, they'll be golden. I loved it.

Manas and I drove back to Berkeley, stopping in San Luis Obispo for dinner. We found a place called Thai Boat with unusually enthusiastic staff. They were all just so happy to serve us! And you can never go wrong with pineapple fried rice. We stopped at a gas station, and the owner of the convenience store greeted Manas in Hindi and they had a nice conversation. Shared language and culture gives most people a sense of connectedness, it seems.

We got back to Oakland shortly after midnight. We found my car, which to my great surprise was neither ticketed, towed, nor impounded. I drove it back to my apartment in Berkeley, relieved, and crashed.