2/2 Steve: diSessa Round 2

1)      The Narcessian paper addresses ‘asking questions and defining problems’ when it discusses how the bursts were initially considered noise but turned out to be meaningful.  This was a case of an improperly defined problem which modeling helped resolve.  The scientists thought that bursts were the problem, but they were actually part of the solution.  The practice most stressed by the Narcessian paper is ‘developing and using models’.  This is the whole focus of the paper.  The scientists used several different models and combined them in the end after making numerous iterations that improved each time.  The models not only communicated certain aspects of the network to others, but they helped the scientists understand what was actually going on in the network.  The paper addressed ‘analyzing and interpreting data’ by showing how the models continuously evolved based on new data coming from the neural network.  As information flowed from the network, it was used to improve the models.  ‘Computational thinking’ was stressed when Narcessian discussed how the computer model was made by one student who preferred to think in a more mathematical way about the network.  This model ended up being extremely useful in benefiting from the study.  ‘Constructing explanations’ was a skill used by the scientists throughout.  They were continuously bombarded with new information both from the models and the network itself, and they had to keep trying to find explanations that were consistent with all of the available data.  ‘Communicating information’ was discussed in the context of sharing models.  When the scientists came together to combine the benefits of each individual model, they had to carefully communicate to each other all of the aspects of their model.  Working as a team collaborating together was an important aspect of the study’s success. 

2)      diSessa thinks that computational literacy is useful because it could allow students to see motion and other concepts in a new way that would make it much easier for them to understand kinematics. He shows how computer programming is well suited to describe motion, particularly with the help of software like Boxer, and the reader can well imagine there are other concepts than motion that would be similarly aided by computer software.  DiSessa speculates the same thing that Narcessian anecdotally chronicles: computers allow us to model scientific phenomena in ways that pen and pencil cannot.  Traditional static models or even video animations cannot encapsulate the variation present in complex systems like kinematics or neural networks.  To properly understand complex systems, we need to engage with models that we can not only see move, but that we can change and modify and poke and watch the results.  In diSessa’s piece, he references the intractability of computer models when he mentions children changing the slider for the variable a which affects the direction of the motion vector v.  In Narcessian’s piece, the same thing is demonstrated in the way that the computer model continues to be changed based on new pieces of information gathered from the real neural network.  Both of these practices involve complex systems, modeling, and using the power of computers to understand how the system responds to change.  The students in Narcessian’s paper would not have been able to understand the neural network had they not possessed the computational literacy to be able to make a working model of the network.  Because they possessed that literacy, they were able to make a breakthrough discovery, and that is the kind of thing that diSessa claims would happen much more often in a world where computer science was taught to all children.