Students at Sal Khan’s new school combine a progressive sense of care with a Silicon Valley focus on data-mining consumer behavior.
The students of Khan Lab School are back from lunch, standing in a circle, trading public accolades. “I have a shout-out for Mary, because when no one would take me to the bathroom, Mary did,” one student announces. “It showed conscientiousness and social intelligence.” Another student adds, “I have a shout-out for Mishal for being a really good sport about going inside and about not eating with everyone else. It showed social intelligence, self-regulation, self-awareness, and conscientiousness.” After each compliment, the entire student body waves their fingers and chants “faaaantastic!”
It’s the kind of Kumbaya moment that could easily occur in squishy-minded, confidence-boosting schoolrooms across the country, with one difference: Orly Friedman, the school’s director, asks the students to add every remark to a Google form that tracks who delivered the praise, who received it, and which specific traits they called out. Over time, she says, she will have a detailed analysis of her students’ character development. (html)
There are a couple routes into this, but the most interesting one is to ask what this record adds to the students education. And perhaps the most honest answer is we don’t know yet — it could be a very profitable activity if certain actionable patterns are discovered. And of course, this is a lab school, and labs must keep good records.
Another way to look at it, however, is that they are trying to find a formula of sorts for care, one that could be applied in classrooms with less skilled teachers. Whatever the first results of this experiment will be, Goodhart’s Law will prevail: the minute that metrics of care are made a quantifiable target students, teachers, and administrators will manipulate the record to goose the scores.
A third way might be to see the data collection as a religious rite of Silicon Valley, a way to signal the importance of a thing. Viewed this way, we might be excited: it’s a school that signals the importance of caring in the best way it knows how: by logging it. This is not a bad thing at all.
A final concern might be the trust in a spreadsheet of care over the intuitions of a teacher. There are many places metrics give us a clearer picture, but do we honestly believe that care is one of them?
Bibliometrics provides a good example of Goodhart’s Law. It also shows how a biased environment uses metrics to reinforce privilege. See Impressions of Objectivity
Bad Adsorption Activity shows how rigidity in delivery method can be counterproductive.
The model of the lab school is a good one, and we should celebrate et tech’s embrace of it. See The End of Open-Endedness