I agree that his evidence is sound. But like any other constructed data set, Piketty makes choices about measurement indicators and categories that are open to

I agree that his evidence is sound. But like any other constructed data set, Piketty makes choices about measurement indicators and categories that are open to challenge. The way that he composes his measures of wealth and capital could be more nuanced, if those nuances make a significant difference in the results produced by the models built on top of them. This MIT student seems to be doing just that—arguing that the composition of land and capital in measures of wealth need to be treated with more nuance, and that measures of capital need to take account of depreciation in a different way than Piketty does. I make similar arguments in my dissertation about one of the longest-running, most widely used data sets in my field. Sometimes the categorization of data is misleading in certain ways that elides phenomena that are actually very significant. Piketty's construction of the data set is fantastic work. But it doesn't mean that all the choices he made in constructing and categorizing the data are the best choices that could have been made.