Archive for May, 2011

Algorithms listing the dead

May 17, 2011 Leave a comment

Later this year, New York will unveil the memorial to the victims of the 9/11 terrorist attacks in 2001, a memorial which lists the 2,982 names of all those killed (in New York, in Washington, and in Pennsylvania).  How to order the  names became an interesting sociological and computational question, as Nick Paumgarten describes in The New Yorker:

By the end of that year, the foundation had received twelve hundred requests for adjacencies (and these didn’t include the self-contained adjacencies, such as, say, Ladder Company 7 or Cantor Fitzgerald, which, with six hundred and fifty-eight names, represented the biggest, and most challenging, adjacency block of them all). The reasons for these requests were varied. Sometimes the victims were cohorts, or best friends. In other cases, the families knew, from last phone calls, whom their loved ones had been with in the end—in an elevator, on a ledge—and wanted those people listed together. A same-sex couple and their three-year-old son all perished on Flight 175; their names, certainly, belonged together. One woman, Angela Houtz, died in the Pentagon, in a conference room with seven others coördinating a response to the attacks in New York. Her mother requested that Angela be listed with these seven. Another woman, Abigail Ross Goodman, lost her best friend, who’d been on the ninety-sixth floor of the North Tower, when Flight 11, with her father aboard, crashed into it—a meaningful adjacency, to be sure.

.  .  .
At a certain point, the foundation recognized that this job could use the assistance of a computer. Even so, the first few computer scientists and statisticians the foundation got in touch with said that it couldn’t be done. “It really did seem insurmountable,” Daniels recalled. But then his chief of staff called Jake Barton, the principal at the media-design firm Local Projects, who took on the assignment, and, with a data artist named Jer Thorp, designed an algorithm that could sort the names in keeping with all the overlapping requests. Before long, they had a distribution designed to please everyone, including Arad.

“It was a computer-science problem, but it was also a big, crazy typography problem,” Barton said last week. As a spatial puzzle, it also owed a little bit to the so-called “knapsack problem” in mathematics, which involves trying to optimize the fit of irregularly shaped or weighted objects in a backpack. Their solution was really a combination of algorithms, which they called the Names Arrangement. A graphic representation of the computational armature, color-coded on a laptop screen, brings to mind Tetris, but the sight of the names themselves, inscribed in bronze, linked together by happenstance and blood, calculus and font size, is a little like the faint silhouette of a cosmic plan, or else of the total absence of one.

Categories: Algorithms
%d bloggers like this: