4 ms·

(disclaimer, ms employee here and the below is just my opinion). > Just don't harvest the data in the first place. for better or worse, I just don't think th

(disclaimer, ms employee here and the below is just my opinion). > Just don't harvest the data in the first place. for better or worse, I just don't think this is feasible anymore. Telemetry is, by now, a fundamental part of the engineering process. Products that don't incorporate it are going to be clobbered by products that do. Microsoft didn't start this paradigm, but I think they had to incorporate it in order to stay competitive. If you can't avoid telemetry, I think the next best thing is to be transparent about it (with respect to what's being collected, how it's being used) and allow the user some control over it (by letting them 'clear' data that's been collected, and opting in/out as needed). In that respect, I don't think the steps taken today are sufficient, but they are headed in the correct direction. I also think they are being more open/transparent than many other tech companies (who do similar things without receiving nearly as much negative press as MS).

akjainaj 10y ago
Telemetry is important if you've gotten rid of, you know, the QA teams MS had in the past
- bradford 10y ago
  Can't tell if you're being tongue-in-cheek, but that was kind of the problem. The cost it took to do QA for Windows was immense, and even with an army of testers, there were still highly embarrassing (and very expensive to fix) flaws in the product all the time. Windows had to find a cheaper solution or die a slow death in the market. Earlier I said "Products that don't incorporate [telemetry] are going to be clobbered by products that do". High QA costs are the biggest part of this.
  - dingaling 10y ago
    > The cost it took to do QA for Windows was immense As it should be when you're developing an OS for 90% of the market across a huge variety of hardware. There's really no way around that, and jettisoning QA teams as a cost-saving response is incomprehensible to me. Trusting telemetry alone to ensure the robustness of your product will result in a "good enough to ship" focus on mainstream work-flows and leave corner-cases unaddressed since they're rare. And corner-cases are where the bad, system-breaking things happen. Telemetry should help QA people write mundane automated test-cases to cover the 99% and then enable them to go and spend the constructive part of their time focusing on breaking the system. But it's absolutely not a replacement for QA.
    - bradford 10y ago
      I'd say that 'jettisoning QA teams' is a mischaracterization. The role is still filled, it's just evolved a lot since the windows 8.1 (and prior) era. Similarly, I'd say that 'Trusting telemetry alone' is also a mischaracterization. From where I sit, both roles are combined. The output is a far smoother engineering process and a higher quality OS. You mentioned 'corner cases [being] unaddressed' but I've witnessed the exact opposite: most of the criticism I hear about Win10 has nothing to do with quality.
hendersoon 10y ago
I think that's a very reasonable stance. I understand the value of telemetry-- it truly does help improve software. And I'm fine with telemetry being active by default too. If it wasn't active by default its value would plummet as very few people would /choose/ to turn it on. Windows telemetry should be set to "Basic" by default on every Windows install, and the user can choose to turn it off entirely or set it to "Full". As a more sophisticated user who values his privacy, all I want is a supported method to opt-out. That's it. I don't want to be forced to use a third-party program to do it. I don't want Windows updates to frequently reset my clear preferences, so I need to keep running that program on a regular basis. I just want a switch to turn it OFF.
gizmo 10y ago
I appreciate that it's tempting to add telemetry and crash logging functionality. The problem is that Microsoft gets the benefit (in reduced QA costs) but the consumer pays the cost (loss of privacy). Classic externality trap. Microsoft acknowledges that full documents along with stack traces (that can contain sensitive information) can end up on Microsoft servers when error log reporting is enabled. Sending a call stack consisting of function names and pointer offsets and register values to aid debugging is one thing. Sending entire globs of application memory along with the crash log is much less defensible. My objection about linking user data to real world people still stands. There is no technical reason why _personalized_ logging is necessary. Telemetrics can be pretty innocent (e.g. track which percentage of Windows users have multiple monitors), but as they become more personal it's again much harder to justify. The absence of a single switch (during install or first login) to get rid of all logging/monitoring/telemetrics and other privacy-invading nonsense is indefensible.

akjainaj 10y ago

Telemetry is important if you've gotten rid of, you know, the QA teams MS had in the past

bradford 10y ago
Can't tell if you're being tongue-in-cheek, but that was kind of the problem. The cost it took to do QA for Windows was immense, and even with an army of testers, there were still highly embarrassing (and very expensive to fix) flaws in the product all the time. Windows had to find a cheaper solution or die a slow death in the market. Earlier I said "Products that don't incorporate [telemetry] are going to be clobbered by products that do". High QA costs are the biggest part of this.
- dingaling 10y ago
  > The cost it took to do QA for Windows was immense As it should be when you're developing an OS for 90% of the market across a huge variety of hardware. There's really no way around that, and jettisoning QA teams as a cost-saving response is incomprehensible to me. Trusting telemetry alone to ensure the robustness of your product will result in a "good enough to ship" focus on mainstream work-flows and leave corner-cases unaddressed since they're rare. And corner-cases are where the bad, system-breaking things happen. Telemetry should help QA people write mundane automated test-cases to cover the 99% and then enable them to go and spend the constructive part of their time focusing on breaking the system. But it's absolutely not a replacement for QA.
  - bradford 10y ago
    I'd say that 'jettisoning QA teams' is a mischaracterization. The role is still filled, it's just evolved a lot since the windows 8.1 (and prior) era. Similarly, I'd say that 'Trusting telemetry alone' is also a mischaracterization. From where I sit, both roles are combined. The output is a far smoother engineering process and a higher quality OS. You mentioned 'corner cases [being] unaddressed' but I've witnessed the exact opposite: most of the criticism I hear about Win10 has nothing to do with quality.

hendersoon 10y ago

I think that's a very reasonable stance. I understand the value of telemetry-- it truly does help improve software. And I'm fine with telemetry being active by default too. If it wasn't active by default its value would plummet as very few people would /choose/ to turn it on. Windows telemetry should be set to "Basic" by default on every Windows install, and the user can choose to turn it off entirely or set it to "Full". As a more sophisticated user who values his privacy, all I want is a supported method to opt-out. That's it. I don't want to be forced to use a third-party program to do it. I don't want Windows updates to frequently reset my clear preferences, so I need to keep running that program on a regular basis. I just want a switch to turn it OFF.

gizmo 10y ago

I appreciate that it's tempting to add telemetry and crash logging functionality. The problem is that Microsoft gets the benefit (in reduced QA costs) but the consumer pays the cost (loss of privacy). Classic externality trap. Microsoft acknowledges that full documents along with stack traces (that can contain sensitive information) can end up on Microsoft servers when error log reporting is enabled. Sending a call stack consisting of function names and pointer offsets and register values to aid debugging is one thing. Sending entire globs of application memory along with the crash log is much less defensible. My objection about linking user data to real world people still stands. There is no technical reason why _personalized_ logging is necessary. Telemetrics can be pretty innocent (e.g. track which percentage of Windows users have multiple monitors), but as they become more personal it's again much harder to justify. The absence of a single switch (during install or first login) to get rid of all logging/monitoring/telemetrics and other privacy-invading nonsense is indefensible.