The Ballmer Prize Finalists

The short list for the Ballmer Prize is in, and as usual the projects are remarkable. You can see them on special display at the Smithsonian, although as noted below most of the projects require a degree of audience participation that makes it difficult to appreciate them fully there. This year the finalists for the $1 million prize are Alan, Calista, Dave, and Sarang. The theme this year was "the human condition," and the prize will go to the project that best expresses that theme. I don't get a vote anymore (thanks Donna Tartt!), but I will describe the finalist projects for my loyal readers.

Calista's entry, "More Evidence," is deceptively simple. It consists of a giant crossword puzzle spanning hundreds of rows and columns but otherwise following the usual conventions. It is intended to be solved collaboratively, with two or more people starting in different places on the puzzle and eventually working toward each other's completed squares.

The trick is that there is no unique solution—each clue can be solved with at least two possible answers. This ambiguity tends to be resolved based on earlier answers, which lock in certain letters. Those answers, in turn, depend on the happenstance of which answer occurred to the solver first. The puzzle captures how our knowledge of the world comes to us in path dependent ways. Depending on our formative learning experiences, we process later information differently and try to fit it into our existing world view with minimal disruption to its coherence.

In "More Evidence," this can play out in different ways. Some couples solve the puzzle with no discord and are surprised to find the project garnering rave reviews as a work of art. This is because they were serendipitously working toward the same solution the whole time. Other players find that when their filled-out areas started to overlap, they give incompatible answers (that is, each has been working toward a different solution to the overall puzzle). They are forced to reconsider at least one player's work and start over. Of course the problem is deciding which player's work to cast aside, and it neatly captures something deep about the way that people form incompatible but equally valid beliefs about the world. How do you decide whose truth is acceptable and whose must be discarded? And on a psychological level, how do you cast aside so much hard-earned "knowledge" all at once? The resulting fights have been legendary, and at least two spectacular breakups have resulted.

Dave's entry, "Calling All," is perhaps the most ambitious of the finalists. It is a procedurally generated computer game consisting of billions of records—grainy photographs, newspaper stories, emails, transcripts of telephone calls, and so forth. There is no stated goal or win condition, but players have realized that the game invariably depicts the last few weeks of human life on earth around the time of an enormous catastrophe. It's widely assumed that the point of the game is to piece together what happened.

It tends to be a lonely enterprise because a new scenario is generated each time someone plays the game. You can't apply anything learned by anyone else to your particular iteration of the game. (Some players have managed to play collaboratively on a single "run" of the game, though.) Sometimes the world ends in a nuclear cataclysm, sometimes humanity is wiped out by infectious disease. Sometimes the world's institutions simply collapse and we lose the capacity to feed or protect ourselves.

My game happened to involve a military conflict culminating in nuclear war. The nature of the game is that I had more than enough information to piece together what happened (in fact, more records than I could possibly review in a lifetime), but it was nearly impossible to tell whether any particular piece of evidence hid some important meaning. Here's a picture of an Israeli jet flying over Turkey. (Or is it a Russian jet disguised as an Israeli jet?) Here's a transcript of a frantic radio transmission as a Japanese army outpost is overrun. But then, here's a prewar memo from the U.S. Secretary of the Interior about wolves in national parks—can that possibly have any bearing on what happened? (The Secretary, for what it's worth, survived the initial nuclear strike and briefly led the Coalition relief efforts in Alberta.)

The information is voluminous and fragmented and often contradictory. A few players have published compelling timelines for their games, but when they've opened up the data to the public, as often as not the initial timeline gets picked apart as people turn up countervailing evidence. Part of the problem is that with so much information to review, it's impossible to know whether you've reached the truth or whether there is some email or photograph lurking in the archives that will completely change your understanding of events. It's difficult to describe the feeling this creates, but it's a kind of uneasiness paired with frustration and boredom. Most people give up trying long before they form any stable, coherent understanding of the world of the game.

Sarang's entry, "I Want a Lawyer," is a movie following 8 characters over the course of a 24-hour period in Philadelphia. I should say "movies," because "I Want a Lawyer" actually consists of 8 movies, each of which is exactly 24 hours long and depicts everything that happens to one of the characters over the course of that day. The project can only be screened at multiplexes with at least 8 screens, and each member of the audience must choose which screen to watch at any given moment (viewers are allowed to go from screen to screen throughout the viewing). The only way to see every frame of the film is to view it 8 times, and even then you can't take any breaks to sleep or use the restroom. (This is alleviated somewhat by the fact that the camera stays with each character even when he or she is sleeping, so you can catch a break during those downtimes.)

One interesting aspect of "I Want a Lawyer" is what happens when two of the main characters meet. In some cases, Sarang used one shoot (with two cameras) to capture the scene, so that the viewer is seeing the same action take place on screen. This does not mean that you would necessarily come away from the two depictions with the exact same understanding of what happened, of course, but it limits the extent of any divergence.

In other cases, Sarang shot the scenes separately, and fairly significant differences were introduced, presumably reflecting the characters' subjective experiences or memories of the events in question. This makes it extremely challenging, even after viewing all 192 hours of film, to say exactly what happened in the movie. Perhaps the title of the project is a reference to the fact that the way events are understood is inherently up for debate and therefore subject to advocacy. In other words, comprehending the "fact of the matter" is the beginning, not the end, of our journey toward truth.

Finally, Alan's untitled project is in some ways the simplest. It is an old brown dog with a graying muzzle. It lies on a well-worn hair-covered bed and occasionally licks itself or farts. It is friendly enough, and certainly doesn't bite, but it tends not to respond much to human visitors unless they bring cheese.

Sarang's The Angel Critique

Sarang's latest effort, The Angel Critique, is an odd but timely book. Set mostly in Michigan, the book follows T'Quan Phillips, a star football player in his first season for the Detroit Lions, with periodic flashbacks to his time at the University of Michigan and, earlier, playing high school football in his hometown of Flint. As the novel opens, Phillips enjoys something approaching hero status in Michigan, where he led the Wolverines to a national championship and where he is the Lions' only hope for a successful season. It barely catches the reader's attention when Sarang casually mentions that the University of Michigan's computer system has been hacked and its academic records breached. There are far more exciting events to consume our attention, like Phillips' outrageously good 17 touchdowns with only 1 interception during the first five games, all of which the Lions win.

But things fall apart quickly for Phillips once one of his undergraduate papers is made public between game 5 and game 6. The paper focuses on students in inner-city schools and finds that they learn best when they play an active role in deciding how the topic they are studying is presented (I'm simplifying—the actual finding is hard to encapsulate). At first, the paper is downloaded by football fans eager for any insight into their star player, but soon it attracts broader attention on social media, especially Twitter.

That's where it comes to the attention of Murray Angel, an NYU statistician who writes a blog post pointing out flaws in Phillips' methodology. The problem is not so much that Phillips engaged in "p-hacking" (tailoring the numbers to reach a conclusion that crosses the 0.05 "significance" threshold for p-values) or "fishing" (trying many theories until one reaches the significance level, like this)—Angel doesn't find any indication of these things. But the study suffers from what Angel calls the "garden of forking paths": Phillips had too many degrees of freedom. That is, he could have analyzed his data in many different ways, and the particular way he chose to do so was probably shaped by the data itself. This undermines the strength of his conclusion and calls into doubt whether it has any real value. Nevertheless, the paper earned Phillips a B+ and helped him pass the class and ultimately graduate from the University of Michigan, all of which is now cast into doubt in the minds of the public.

Once they've absorbed Angel's bombshell critique, outraged Lions fans turn on Phillips with a vengeance, even as he delivers stellar results on the field. After each victory, during postgame interviews, sports reporters only want to talk about the Angel critique. Ultimately, after a 9-1 start to the season, the Lions have no choice but to bench Phillips, and the team finishes out the season at 10 and 6, barely missing the playoffs. Although his contract will last for two more years, Phillips knows that he will never play another game for the franchise. His career in professional football is effectively over, and because of the ignominious way in which it ended, he is unlikely to get a job in sports writing or broadcasting.

The Angel Critique is unsettling because it shows how our heroes so often fall short of our expectations. Whether it's performance-enhancing drugs, domestic abuse, or the garden of forking paths, our athletes often disappoint us off the field. So why do we keep investing our hopes in them? Why do we insist that they share our values and our politics? Why are we so crushed when they turn out to be deeply flawed?

These are questions that are roiling the world of professional sports, and so Sarang's work is yet again essential to understanding the times we live in.