Just like people would want the data to be radically usable, one has to find the algorithm that doesn’t strike a compromise between usability and privacy laws. Then it’s peer reviewed, so every data owner gets to run it simultaneously. That makes it scientific, according to Sandy.