Sounds good to me. Solve the end-user problems, since they don't have the ability or care to do it themselves and doing so manually has too much latency and doesn't scale. ----- Mike Hammett Intelligent Computing Solutions http://www.ics-il.com Midwest-IX http://www.midwest-ix.com ----- Original Message ----- From: "Giovane C. M. Moura via NANOG" <nanog@nanog.org> To: "North American Network Operators' Group" <nanog@nanog.org> Sent: Thursday, March 24, 2022 5:43:58 AM Subject: ISP data collection from home routers Hello there, Several years ago, a friend of mine was working for a large telco and his job was to detect which clients had the worst networking experience. To do that, the telco had this hadoop cluster, where it collected _tons_ of data from home users routers, and his job was to use ML to tell the signal from the noise. I remember seeing a sample csv from this data, which contained _thousands_ of data fields (features) from each client. I was _shocked_ by the amount of (meta)data they are able to pull from home routers. These even included your wifi network name _and_ password! (it's been several years since then). And home users are _completely_ unaware of this. So my question to you folks is: - What's the policy regulations on this? I don't remember the features (thousands) but I'm pretty sure you could some profiling with it. - Is anyone aware of any public discussion on this? I have never seen it. Thanks, Giovane Moura