I really hope that they're using the Prio system (https://crypto.stanford.edu/prio/ https://crypto.stanford.edu/prio/), perhaps via Divvi Up (https://divviup.or

I really hope that they're using the Prio system (https://crypto.stanford.edu/prio/ https://crypto.stanford.edu/prio/), perhaps via Divvi Up (https://divviup.org/ https://divviup.org/), which is offered by the fine folks at ISRG. But, the fact that they don't mention anything other than "All event data will be de-identified and processed in aggregate before it’s used for analysis" leads me to believe it's an in-house solution for a very complicated problem, if one chooses to do it correctly.