The Tech - Online EditionMIT's oldest and largest
newspaper & the first
newspaper published
on the web
Boston Weather: 29.0°F | Mostly Cloudy

MacGregor Lounges Expecting Crowding

By Karen Robinson

Crowding in the MIT housing system during the 1999-2000 school year will be at about the same level as last year, according to Director of Residential Life Phil Bernard. This year, however, Bernard hopes to create a more even distribution of crowds across dorms.

Last year’s situation was unusual in that pledging started a day later than in previous years, and the Residential Life Office was unable to predict how many people would pledge after receiving dorm assignments.

“After we ran the lottery we had about another 50 to 60 people pledge,” Bernard said. “That opened up spaces in some places, for very uneven distribution.” In previous years, about one-fifth as many people would pledge after the lottery, he said. “People didn’t pledge out of MacGregor” last year, however.

“I hope not to crowd any lounges at all,” Bernard said. “It’s not a good situation.” Tentatively, however, he expects about ten lounges to be converted into doubles. Complaints about this type of crowding do not generally come from students in the crowded doubles, since they are large rooms with nice views, he said.

However, students have complained when forced to decrowd out of their MacGregor entries.

New Lottery Optimizations

The lottery is run by Eliot Levitt ’90. His algorithm “optimizes a ‘happiness’ function that tries to capture the appropriate balance of assignments to preferences in a single value,” Levitt said. Last year the algorithm examined possible switches between two or three students, but this year, larger groups will be considered.

This year’s algorithm also uses some genetic programming to optimize the code, Levitt said. This is still in trial, however; last year’s program will be run simultaneously, in case the latest version underperforms. Levitt does not “expect any better results to come out of version 2 than we saw last year,” but the algorithm will be closer to the “right way” of solving the problem, he said.

“And of course if, in the meantime, the new techniques strike paydirt and improve the assignments even one or two percent, I know those four or five freshmen will be very happy we took the time to do it right,” Levitt said.