TITLE: Open Mind, Closed Mind AUTHOR: Eugene Wallingford DATE: April 28, 2007 12:55 PM DESC: ----- BODY: I observed an interesting phenomenon working in a group this morning. Another professor, an undergrad student, and I are at Duke University this weekend for a workshop on peer-led team-learning in CS courses. This is an idea borrowed from chemistry educators that aims to improve recruitment and retention, especially among underrepresented populations. One of our first activities was to break off into groups of 5-8 faculty and so a sample session PLTL class session, led by an experienced undergrad peer leader from one of the participating institutions. My group's leader was an impressive young women from Duke who is headed to Stanford this fall for graduate work in biomedical informatics. One of the exercises our group did involved Soduku. First, we worked on a puzzle individually, and then we came back together to work as a group. I finished within a few minutes, before the leader called time, while no one else had filled in much of the grid yet. Our leader asked us to describe bits about how we had solved the puzzle, with an eye toward group-writing an algorithm. Most folks described elements of the relatively naive combinatorial approach of immediately examining constraints on individual squares. When my turn came, I described my usual approach, which starts with a preprocessing of sorts that "cherry picks" obvious slots according to groups of rows and columns. Only later do I move on to constraints on individual squares and groups of squares. I was surprised, because no one seemed to care. They seemed happy enough with the naive approach, despite the fact that it hadn't served them all that while solving the puzzle earlier. Maybe they dismissed my finishing quickly as an outlier, perhaps the product of a Soduku savant. But I'm no Soduku savant; I simply have had a lot of practice and have developed one reasonably efficient approach. The group didn't seem interested in a more efficient approach, because they already knew how to solve the problem. My approach didn't match their own experiences, or their theoretical understanding of the problem. They were comfortable with their own understanding. (To be honest, I think that most of them figured they just needed to "go faster" in order to get done faster. If you know your algorithms, you know that going faster doesn't help at all with many, many algorithms! We still wouldn't get done.) Dr. Phil -- How's that workin' for ya? After making this observation, I also had a realization. In other situations, I behave just like this. Sometimes, I have an idea in mind, one I like and am comfortable with, and when confronted with something that might be better, I am likely to dismiss it. Hey, I just need to tweak what I already know. Right. I imagine Dr. Phil asking in his Texas drawl, "How's that workin' for ya?" Not so well, but with a little more time... When I want to learn, entering a situation with a closed mind is counterproductive. This is, of course, true when I walk into the room saying, "I don't want to learn anything new." But it is just as important, and far more dangerous, when I think I want to learn but am holding tightly to my preconceptions and idiosyncratic experiences. In that case, I expect that I will learn, but really all I can do is rearrange what we already know. And I may end up disappointed when I don't make a big leap in knowledge or performance. One of the PLTL undergrad leaders working with us gets it. He says that one of the greatest benefits of being a peer leader is interacting with the students in his groups. He has learned different way to approach many specific problem and different high-level approaches to solving problems more generally. And he is a group leader. Later we had fun with a problem on text compression, using Huffman coding as our ultimate solution. I came up with an encoding targeted to a particular string, which used 53 bits instead of the 128 bits of a standard ASCII encoding. No way a Huffman code can beat that. Part way through my work on my Huffman tree, I was even surer. The end result? 52 bits. It seems my problem-solving ego can be bigger than warranted, too. Open mind, open mind. -----