Post AXcTFvsn4ivk0UST2G by w8emv@hachyderm.io
 (DIR) More posts by w8emv@hachyderm.io
 (DIR) Post #AXXOEfJIgRecS4tMbg by w8emv@hachyderm.io
       2023-07-10T02:29:56Z
       
       1 likes, 2 repeats
       
       I don't actually believe that privacy-preserving telemetry is even a thing that can be real. Telemetry is designed to leak information about what's happening on a system, and by doing that you inevitably leak bits of privacy-adjacent material, and if you accumulate enough bits you  get reidentification of previously "anonymous" information.My favorite old-school reading on this is the 33 Bits blog https://33bits.wordpress.com from Arvind Narayanan.
       
 (DIR) Post #AXa4UFmbwhJKIxdZRo by ignaloidas@not.acu.lt
       2023-07-11T09:41:58.766Z
       
       0 likes, 0 repeats
       
       @w8emv@hachyderm.io It definetly can be done, and work is going on to make it possible to do "in-production"https://crypto.stanford.edu/prio/paper.pdf is the main paper that is describing a system that would ensure privacry-preserving telemetry as long as you trust at least one of several server-running parties to not collaborate to everyone else.https://divviup.org/ is the project from the same guys that run Let's Encrypt to make an actual service out of that and another paper from the same authors. They're also working on IETF standards for the protocols so more than one implementation could be had.I'm extremely tired of endless, unconditional yelling of "telemetry bad", when it could be done properly.
       
 (DIR) Post #AXcTFvsn4ivk0UST2G by w8emv@hachyderm.io
       2023-07-12T13:10:08Z
       
       0 likes, 0 repeats
       
       @ignaloidas so noted, but the opt-in telemetry (quite simply, upload your dmesg files) described here https://vielmetti.typepad.com/logbook/2023/07/privacy-preserving-telemetry-isnt-real.htmland https://dmesgd.nycbug.org/index.cgi?do=indexis so much simpler and gives you 99% of the results you are seeking at 1% of the complexity.
       
 (DIR) Post #AXcTFwgm4sueVVqNwe by ignaloidas@not.acu.lt
       2023-07-12T13:28:55.937Z
       
       0 likes, 0 repeats
       
       @w8emv@hachyderm.io Well, that's both very much not privacy-preserving, nor would it be actually useful for anything bug kernel debugging and statistics. What Fedora seems to aim to collect is more akin to popularity contest, which is also not privacy-preserving (though it could be made, there's an algorithm for getting strings with more than N occurrences while preserving privacy), but also some stuff that is more akin to traditional application metrics, like for example how often some settings are accessed, which your suggestions wouldn't cover at all.The fight against data is idiotic. Fight against personally identifiable telemetry, not telemetry that is actually privacy preserving.Though in fedora's case, while it's probably still less private information than just uploading kernel logs, it's still not really privacy preserving.