Email to Candace

Email to Candace Hi Candace, 

I just finished the books you suggested to me—Simon’s The Sciences of the Artificial and Mitchell’s Unsimple Truths. I found both interesting, and in both I found responses to the question I originally asked you about how to approach multiple levels of analysis in complex/dynamic/nonlinear systems. I found Simon’s observation that in studying designed or purposeful systems, we ultimately only see the ways in which the system is hampered or constrained by the environment. The image of the ant walking on the beach–complex behavior may be a result of simple rules in a complex environment—is eerily congruent with the intuitive lessons I am gaining through CS221 this quarter. 

Mitchell’s writing is so clear! As I read, I had the sensation that her prose was a contraction of time, that decades of iterative thought had been condensed into a little over a hundred pages. Her concept of integrative pluralism in epistemology, inquiry, and policy feels like a solid foundation for the work I hope to do in Learning Science—it will be interesting to revisit when I am in the midst of analyzing real projects. I think I originally approached you with an idea of a reductionist approach, locating emergent phenomena such as learner identity in lower-level signals which are easy to collect. I came away from Simon with a richer interpretive frame for thinking about the meaning of the signals, and from Mitchell with a model of how questions at multiple levels of analysis might productively be related to one another. Thanks for suggesting both!

I also wanted to thank you for the opportunity to share my (immature) work in class this week. I left feeling guilty, that I would have benefitted more from peoples’ time if the project had been further along. Meta-lesson learned.  I got a lot of really valuable feedback, and I’m sure my DAPS poster in December will be much better for the experience.  I hope at least that my presentation was a contribution toward a supportive community where people feel safe being open and vulnerable sharing their work. 

I also have a request—whether you have time to meet to help me brainstorm research questions and approach for applying some machine learning techniques to educational data. Katie Cheng and I are taking CS 221 this quarter, and are planning to work together on a final project doing some initial exploration of some data I hope to develop into my QP down the road. Last spring I developed a task for 8th grade computer science students and administered it to a cohort of about 65 girls, my former students at the Girls’ Middle School. My purpose was to observe which students would choose to use Scratch to solve the task, which would use Python, and which would use something else. I was curious whether choosing Python would be an indicator of successful transfer from block-based to text-based, just a different modality, or whether there was some other story. I shared this with my lab last spring, and got some great feedback from Rich, Engin, and Bertrand; there were some interesting results but I came away feeling the study’s design was not as effective as it could have been to answer my question. 

But what’s interesting about it is that I have an extremely rich timeline of their development as computer scientists. I have every code artifact they created over two and a half years (JSON representations of Scratch projects; Python code; websites), my assessments of their work, their reflective writing, the results of the transfer task, and also the results of Michelle Friend’s gender attitudes survey for this cohort, with IRB approval to use this data and consent/assent from over 50 of the students. This seems like it could be interesting to explore via a variety of ML/AI methods: feature extraction from code, textual analysis of writing, and unsupervised clustering into trajectories. It would be cool to be able to characterize a few distinct trajectories of learners in the group. But as I have a newfound goal of thinking more about my approach before diving in, I would be grateful if you had some time in the next week or two to help us think through how we might most effectively proceed. 

Looking forward to class this week, Chris