← Back to context

Comment by teraflop

4 years ago

"20X more telemetry", in terms of data usage, is a pretty meaningless statistic on its own (unless it's large enough to affect your mobile data cap or something).

For instance, I would consider it a much bigger privacy violation for my phone to transmit my exact location every hour than my current CPU usage every 10 seconds.

Which Apple is apparently doing - they send location, local IP, and nearby wifi mac addresses even when you're not logged in. Similarly Apple is collecting more data types than Google according to the research paper.

  • Please provide evidence of this because Apple's official documentation says otherwise:

    https://support.apple.com/en-us/HT203033

    They do send nearly WiFi hotspots for crowd sourcing purposes but it is never in conjunction with your local IP address (which is an identifying piece of information).

  • Allowing wifi mac addresses, ssids, bssids, etc. of leased equipment in combination with subscriber address/goelocation to be shared or otherwise disclosed to third party affiliates, partners, agencies is a requirement included in the fine print of some residential ISP's agreement terms I've read, fwiw.

    I assumed that this probably is implemented as a 'non-public' goelocation service api as well raw data sharing agreements in some cases, but I'd doubt the data 'processors' and 'controllers' are known to anyone outside those parties.

I was also surprised by the emphasis on "20x more data" aspect. The table on kinds-of-data sent was showing Apple in a much more negative light.

  • The article is surprisingly short on details. E.g., Google does know your location and the Wifi APs, it's just not the OS that sends it.

Sending telemetry can get expensive: in situations where bandwidth/throughput is restricted people often get picky about giving PCs with Windows installed internet access because of this. It can be bad even in normal situations: My girlfriend's laptop has so much broken telemetry crap between Microsoft and HP that her applications actually get pushed into swap (or whatever it's called on Windows.)

  • Sending telemetry _poorly_ can get expensive. A good client can aggregate, even compress locally, and publish telemetry in batches. Let's not rule out telemetry entirely because of bad implementations.

    I'd say the moral of the story here is that Microsoft and HP just write shitty software.