For the sake of argument, the only interesting discussion about telemetry is about whether it’s ok when done right.

I don’t think anyone thinks it’s ok when it’s not done right (not anonymous, dark patterns for opt in/out, etc).

So it’s not a very interesting discussion to have since there is no one arguing for it.

Instead my argument is: when done right, anonymous telemetry isn’t “evil”. To be fair I don’t know if many argue it is either. There are a few absolutists that think not even opt-in telemetry is acceptable and that developers should do more expensive studies to find how their software is used. It’s really only those I disagree with.

'Done right' quickly becomes one of those no true scotsman arguments. Nothing in this world is fully done right.

Just from the top of my head: Telemetry means extra code, hence extra bugs and maintenance overhead. It costs you in extra ram/cpu/storage/network. The networking means NSA and friends have a beacon declaring a windows computer exists, and they probably can derive other facts from the message statistics. After Snowden, we should assume they have a backdoor and get the unencrypted data if they want.

All this assumes Microsoft has only the good of you as end user in mind, are not hackable, and can't be coerced by governements. All of this now and in the future.

'Done right' is not a good yardstick. There are tradeoffs needed, Microsoft decides which ones, and they decided the user has almost no voice in these tradeoffs, and doesn't even get to see te choices made. These tradeoffs are the interesting discussion.

Not having this info means you pay for unnecessary features, get bugs fixed slower, pay for expensive surveys or user studies etc. Sending it in and wasting some cpu/network may be the cheapest option. Because users are paying for it one way or another.

Windows was much leaner and more stable before it was infested with telemetry. Go figure.