Thursday, December 19, 2013

Tracking down the Harvard Non-Bomber

This year, (allegedly) a Harvard student performed the modern equivalent of pulling a fire alarm in order to avoid a final exam, in this case by sending an e-mail claiming that there were bombs in several building throughout the campus.  (One of many Crimson stories here.) 

I am proud to say that this student, who was apparently a psychology and sociology major or a prospective psychology major (according to Crimson reports), was (allegedly) using TOR and Guerrilla Mail to try to cover his tracks.  (See, for example, this article.)  I think it shows how Harvard has made it as a computer science/engineering school, now that even our psych majors know how to set up and use tools like this.  Years ago, before CS started taking off at Harvard, you would be hard pressed to come up with a student from a liberal arts major who could use tools like this.  It just goes to show how the place has changed for the better.  I like to think that, if he was a computer science major, and would have correspondingly more understanding of what tracks he was leaving (hint:  don't use your own computer through Harvard's wi-fi when sending a bomb threat...), he might have gotten away with it, or at least been a lot harder to track down. 

[Just to be clear, this is very tongue-in-cheek;  I in no way support or even really want to make light of what this student did, it's utterly reprehensible.  And as several colleagues of mine and I have noted, he knew just enough to be dangerous-- mostly, in the end, to himself.

Also, I was (again, along with several of my colleagues) 95+% certain right off the bat it was a student trying to escape finals.  Besides the timing, the 4 buildings named as where bombs might be hidden included 3 big lecture buildings where exams were taking place... and a freshman dormitory.  (In fact, MY freshman dormitory.)  It seemed unlikely that the dorm would be on any real bomber's radar, and seemed to me to be a clear signal that one or more students were behind it all.]   

Monday, December 09, 2013

Lesson of the Day

Saturday I took two of my daughters to see a musical at Harvard.  Amazingly, in the small theater, we were in front of a pair of students who seemed intent on talking throughout the performance.  (One male, one female;  the male did seem to be doing more of the talking.)  The volume seemed to increase until by the end of the first act they seemed to be talking at normal conversation level.

As soon as the curtain hit I turned and as nicely as I could (which was, probably, still with a snarl) that there were several bars and cafes available in Cambridge if they wanted to talk, but we were here to watch a show.  I got several approving nods from around the nearby audience;  in fact, about a minute later, an usher for the theater came over and appeared to be telling them to be quiet or get out, so others had clearly complained.

To their credit were apologetic and stayed quiet for the second act.  I can only hope that I helped teach these students the important lesson that conversing in a theater is a very bad idea -- probably more important that most of what I ever teach in class.  (Although how they managed to get this far without absorbing that lesson somewhere is, I admit, beyond me.)  My older kids already know that, but they got some useful reinforcement.   

Wednesday, December 04, 2013

Algorithmic Growth (Class Size)

Pre-term planning numbers are in for Harvard, and it looks like the undergrad Algorithms and Data Structures class has about 175 people planning to take the course.  That's a huge jump again over the last few years (where it's jumped from the 50s to well over 100).  I imagine the growth is spurred by our ever-increasing enrollment numbers in the intro classes, as well as the fact that it's being taken over by a younger, dynamic, new faculty member.  (Go Jelani Nelson.  I can't help but think some number of students were waiting for me to go on sabbatical...)

These numbers are usually within plus-minus 10% of the final, though there's higher variance when new instructors take over.  If 175 became the steady state class size, it would mean a bit over 10% of the students at Harvard would take Algorithms at some point.  I don't think I ever expected that when I started. 

If we can get the people resources, at some point we'll probably want to start breaking this up.  One direction is to make an "honors" class that would cover more/harder material more rapidly.  (We're thinking of making this an "honors theory" course, that would cover both complexity and algorithms -- 2 classes packed into 1.)  The Math department here has done this very successfully, separating out the future Putnam winners from other students early on.  A benefit is it leaves the remaining students happier as well, as the curve-breakers remove themselves from the regular class.  Another possibility is to do an algorithms style class designed for non-majors, that would bring in more people not currently taking algorithms as well as some of those in the current course.  There are "topics" classes like this -- the Easley/Kleinberg book Networks, Crowds, and Markets: Reasoning About a Highly Connected World is algorithmic and seems to allow an instructor to choose a desired mathematical level from a broad range -- but I don't really know of examples of something more like a standard algorithms/data structures courses designed for a broader audience than for CS majors.  I'd be interested in pointers to and anecdotes about experiences in such classes if they exist.