Like many people I’ve talked to, I tend to imagine using Croquet for automation. We envision physics and molecular chemistry simulations running on their own, while the people in the collaboration walk around among the ball and stick model forest and observe. Maybe we reach up and grab an atom or two and pull on it to see how that changes the path of the simulation. That’s my nature. I’m an engineer and I want to automate stuff so I don’t have to work so hard, even in visualization. I’m so lazy I even want to automate my imagination.
I worked for more than a dozen years creating some kind of automation or another. The biggest misconception I had to clear up with my clients was that you can’t automate what you don’t understand. You have to tell the computer exactly what to do. I learned this lesson in high school when we had a model bridge-building contest in physics class. Everyone assumed that I would design my bridge on a computer, and I sat down to try it.
Nothing.
I could program formulae and iterate to an optimum, but I had no idea what the formulae were. It would be another two years before I would take introductory mechanics in college. Instead of automated analysis, or going another level to automate the selection of designs to analyze, I had to think about what I did know. General ideas of levers and moment. I played with it, and came up with an odd asymmetric design that was first judged by the art class to be most beautiful and then lasted longest when tested to destruction with increasing weight. It’s possible that a fully automated design process could have produced a still stronger bridge, and I sometimes wonder if it would have looked as good. But I’m sure that I wouldn’t have internalized the basic concepts as well as when placing individual sticks together in my head and on several packets of graph paper.
There’s an instructional tradition in fine arts and crafts based on performance. Rather than analysis, or maybe in addition, students internalize concepts by learning to physically act them out. Musicians play in addition to composing. Artists paint. Potters throw. Actors perform before an audience. I’m on a bit of a Japanese kick after my recent visit, and I recognize performance-based learning in their culture. Poets, for example, paint out works on thin rice paper that will tear and ruin if the stroke is wrong. Generals traditionally learn individual martial arts in which they perform classical forms to get them into muscle memory. (I just started Karate lessons this week at age 40. I think my muscles have Alzheimers.)
In addition to being a professor of biology, my boss is an experienced performer of musical theater. Last week he postulated a brilliant idea: don’t use Croquet to automate, use it to coordinate performance-based teaching and learning. Instead of automating a simulation that no one directly participates in very much, what if an instructor staged a production of the physical or molecular chemical process being taught. Teaching assistants would be in the Croquet space but not visible to the students. However, the assistants could manually manipulate the objects in the demonstration. The students see the objects perform as though they were automated. It’s puppetry. Croquet has powerful abilities for coordinating activity and controlling what individual participants can see or do, but the same principle could be applied in other learning environments. For example, instead of passively watching a computer-generated graph of a sine wave, younger students could practice drawing one (perhaps translucently over a template), until they internalize the shape.
The current Wired quotes a blogger as noting that some social software Website is limiting to 1000 the number of friends he can have. The user must get rid of some of his friends before adding more. The user felt this was natural because he can only remember so many friends in the real world. This drives me nuts. It’s a computer! It can count past 1000! (Maybe the blogger was being sarcastic?) I want to use computers to enhance human performance, not limit performance even as it automates some aspect. In the pupeteering model, the learning is done through performance, and the computer is used to enhance or facilitate the performance.
What about research & development? Automation of things we know how to do is fine. Automation that enhances our performance. But by definition, at the cutting edge of technology, we don’t know how to automate the advanced development. Think about, for example, cars and all the tools that Mercedes and Honda use to automate their design and construction. But in Formula 1 racing, these companies turn to McLaren, which employes on the order of 1000 people to perform the design and hand construction of just a few cars a year. I think performing advanced design, art, and scientific research, are all heavily “about” internalizing and externalizing concepts. This performance learning should be explored for that purpose, and Croquet should be used to facilitate that. The next steps involve presentation and other means of reproducing results, and Croquet can be used in automating that as well. But I don’t want to skip over the manual work.
