A Contest Story

Posted by Helen Chen,

Matthew Simoneau, matthew@mathworks.com
November 8, 2002

There are a thousand stories in each MATLAB Contest. This scatter plot tells one of them. It is a plot of the leading entries on Thursday night and Friday morning, each entry’s score against its time of submission. The area of each marker is proportional to the number of non-whitespace characters in the entry. The red line shows the running top score (lower is better). Blue markers show entries that call CONV and green represent the rest. All submissions by Kevin McGill, Carlos R. Colon, and Cory Sharp are labeled with last names.

Notice the small blue dot in the upper-left quadrant of the graph. This represents the first entry of Kevin McGill. According to his web page (thanks Google), he is a Biomedical Engineer at Stanford. Called “just pleats”, it is the first entry to use convolution to determine the best places to fold the protein strand. Although it didn’t perform quite as well as the top entries, it ran two orders of magnitude faster and only contained about 1/10 the code. If it had been submitted two hours earlier, it would have easily taken first place. He made two more submissions, one contained an error and the other timed out. We haven’t heard from him since.

Apparently, Carlos R. Colon also spotted this remarkable entry buried in the rankings. Notice his slightly larger blue dot to the right. Then notice the progression of blue dots with his name on them, each growing and pushing downward. It looks like he adds more and more code from the Frankenstein entry currently leading the pack until it breaks into first place. From that point forward, every 1st place entry contains this innovation.

It looks like Cory Sharp also knew which way the wind was blowing, but Carlos R. Colon beat him to the punch.

This sort of competitive collaboration is one of the aspects that makes the MATLAB Contest “programming as a spectator sport”.

Comments are closed.

These postings are the author's and don't necessarily represent the opinions of MathWorks.