- A lovely interview with Cynthia Dwork in the New York Times on bias in computations. In particular, discussing our work (joint with Moritz Hardt, Toni Pitassi and Rich Zemel) on Fairness through Awareness.

- Our Science article, The reusable holdout: Preserving validity in adaptive data analysis (joint with Cynthia Dwork, Vitaly Feldman, Moritz Hardt, Toniann Pitassi and Aaron Roth), is out.

—-

update (8/16/2015): On Privacy and Anonymisation (with Cynthia Dwork and Salil Vadhan) in The Economist

(2) looks interesting. Is the paper available? Is there a concrete problem where we can see what your new method suggests vs. the old scientific method? For example, suppose I want to explain data through a linear model in k variables out of n, k << n or k = O(1). The question is which variables to pick. Does the new method provide insight/methodology on how one should go about this?

Thanks Emanuele! I think that we will be able to share a link to the full paper soon, in the meanwhile http://www.cis.upenn.edu/~aaroth/reusable.html includes links to a companion paper and a subsequent paper with more technical details. There is much more to do, including with respect to more simulations and making it all practical. I hope to write a more detailed blog post at some point.