Gather statistics about client usage patterns in a privacy-preserving way
Rob and I briefly talked about gathering statistics about client usage patterns. The idea is to configure Tor clients to record data about their uptime, number and kind of directory requests, amount of data read and written, number of circuits built and streams attached, number of bridges used, etc. Tor would aggregate these data points in a privacy-preserving way and write them to a local file. We'd ask volunteers to configure their Tor client and send us their usage data, which may not give us a representative sample, but which is still better than having to guess. We can think of better ways to have clients report usage statistics automatically, if configured accordingly. The goals would be to get better input data for estimating daily users, simulating private Tor networks, and maybe other things. As the first step we should brainstorm about what data we'd like to learn and what we'd do with it.