Wednesday, January 28, 2009

Jury Math

For the first time ever, I got called for jury duty. Yet another place in which better use of computers would really expedite the process (did we really need to answer questions with our assigned number so that it could be written down by hand? Isn't there some automatic way to do this? Maybe involving answering questions in advance, all at once? Etc.). The jury selection process was the most interesting part. From what I could gather, the rules for the selection were as follows (after jurors were asked individual questions earlier):

  1. Randomly order the potential jurors.
  2. Fill the jury seats with the jurors at the front of the queue (for criminal trials, that's 12 jurors plus 2 alternates or 14 seats). Some jurors in the queue are skipped over at the discretion of the judge (presumably based on the earlier questions). The judge has no limit on the number of jurors they may skip over.
  3. Each attorney is allowed 10 dismissals. These dismissals are done in rounds. In each round, each attorney may dismiss a juror or pass. I'm assuming that the attorneys don't try to dismiss the same juror. When jurors are dismissed, new jurors from the queue are inserted into the empty seats. I also assume that if an attorney passes "too many" times in a row then the judge gets annoyed and makes them choose.
  4. When all 10 dismissals are made, the process is over and all remaining jurors are dismissed. The jury are the people left in the seats at the end of the process.

So then, the questions are: What's the optimal strategy for each attorney? What if the desirability of a juror is not independent of the other jurors chosen? Ignoring the random ordering, what's the best position for a potential juror (assuming that they don't want to be chosen and assuming that they can't chose to be near the end of the queue) - how much do the surrounding jurors matter? How does this change if attorneys are allowed an unlimited number of dismissals, but have to pull the jury from the limited queue?

This is how I entertained myself while waiting. Well, that and thinking about how the white noise switch that the judge used when the attorneys were talking at the bench and we weren't supposed to hear them was sort of cool. Really the only mildly high-tech piece of the whole experience.

Wednesday, January 21, 2009


I was at the national mall yesterday for the inauguration. It was crowded and amazing. I made it there without too much trouble, including awhile spent on a metro platform trying to get out of the station with thousands of other people, singing "Na Na Hey Hey Kiss Him Goodbye." I ended up just about as close as I could have, considering that I didn't have a ticket. I was in that fourth clump back from the capitol reflecting pool, just past 7th street. And I could see one of the jumbotrons pretty well. The inauguration was moving and of course historic, but I think the most touching thing was the turnout and the joy among those of us watching, happily shoved up against each other. We were a happy mob - peacefully jumping over fences to get into the inauguration or chanting "move that bus" as we tried to leave en mass. I'm glad I went.

I also went on Sunday to the concert at the Lincoln Memorial. It was, if possible, more moving than the inauguration. There's just something about a giant crowd singing "This Land Is Your Land" while swaying and meeting each other. And since it was longer, there was more time to set the history of the moment through presidential quotes, famous singers, and shots of the memorial itself. They also showed clips of past important meetings at the memorial - Marian Anderson and the MLK March on Washington.

There were about 750,000 people on Sunday and about 1.8 million yesterday, making yesterday the largest gathering ever on the national mall. And some people were turned away (though it looks from the satellite photo like there was room, I believe it was all fenced off). The police, army, national guard, metro police, etc. were visible, but not always so good at directing people to the right route. Still, I think the city did a good job, considering. I wonder if understanding the Natural Algorithms that caused the crowd to create those clusters would help in the future.

My city has been celebrating in a unified, gleeful way. And now I'm excited to see what President Obama will do. And excited in general every time he's called "President Obama."

Wednesday, January 14, 2009


For the past week I've been frantically making intricate animations for the talk for my proposal (this is the problem with doing work on moving points - if you know of any good software, please leave a comment). The proposal was today, and I passed! It's strangely anticlimactic, since it's really a mid-process deadline, but exciting nonetheless.

Wednesday, January 7, 2009


I got back from SODA last night amidst the snow (NYC) and rain (DC). I suspect that somewhere in the middle it was ice, so I hope the folks that were driving made it home safely.

Chazelle's talk on "Natural Algorithms" was a highlight. I also have no desire to focus on minute changes in run times, so I enjoy the talks that introduce new problems the most. By that measure, Chazelle's talk was especially good. He discussed the convergence time of a flock of birds, but more generally introduced (or probably re-introduced) the idea of examining natural phenomena through an algorithmic lens. One of the problems I've been thinking about also involves bird flight - can we calculate the migratory path of birds in an online manner as they fly? I suspect that looking at it from the birds' perspective instead of an all-seeing perspective would also be interesting. So perhaps my enjoyment of the talk was really because it related to my work, but I'd say that it related to many problems and that's what made it good. (Of course, he's an excellent speaker... his talk on "How to Win the Best Paper Award" was given at the business meeting.)

I also enjoyed the talk and paper by Yi and Zhang, "Multi-Dimensional Online Tracking." The talk was scheduled right after lunch on the last day, so if you missed it, that might be why. They considered the problem of online maintenance of an approximate value for a changing function. Again, it's related to my work so I'm biased, but I appreciated the problem and thought the 1D solution was especially clean. It's worth a read, or at least a skim.

The proceedings this year were distributed on a CD instead of printed. Abstractly, I think this is a great idea, since after the conference I would be unlikely to look at the hard copy of the proceedings... but I missed them during the conference. Still, if they're keeping the proceedings online and free permanently, it's a net win.

Overall, it was an enjoyable conference. It was good to see/meet some of you. I'm sad to return to "the real world" in which I have a lot of work to catch up on...