The default option is "I do not wish to participate", so if someone isn't paying attention during the install process and just hits 'Enter', their privacy remains intact.
This, IMO, is the perfect way to do telemetry.
This is not privacy-hostile in any meaningful way.
Because they have the data proving that opt-in telemetry means no telemetry. Every time this topic comes up, HN sticks its fingers in its ears and refuses to accept this, but it's the truth. 99% of users, even for a developer-focused application like VSCode, just click through and accept the defaults. So if you want telemetry that isn't useless for any practical purposes, it has to be opt-out.
If you want, you can have a separate argument about whether having the telemetry is worth it, but if you've had that discussion and decided the answer is yes, then opt-out follows immediately from that.
Like, instead of having a prechecked box that people would have to notice and uncheck before proceeding or a greyed out "maybe later" button, what about two different continue buttons labeled "yes, help us improve with anonymous statistics" or "no, I'd rather not contribute"? (Either way with a link available to more details in the privacy policy before deciding.)
My guess is that it would be less useful telemetry than opt-in, more useful than opt-out, and a slight obstacle to sign-ups for consumer webapps but not much of an obstacle to enterprise webapps or anything pre-downloaded.
Maybe it's still useful and smooth enough to be the right balance vs privacy?
If they care about their users and it's not possible to have worthwhile telemetry with opt-in, they simply can't have it.
[0]: https://insights.stackoverflow.com/survey/2019#development-e...
And this is not even something that can't be replaced : if you want to know about your users habits, ask them. I would have no problem with seeing once in a while a broadcast message after syncing packages showing me an url of a poll. But automatic data collection is never ok.
The reasoning is simple.
The reality is complex.
Always, always think about second order effects.
Change your frame of mind from "my software" to "your computer". If your users don't explicitly opt in, you have no right to take what they do not freely offer.
Come again? Is there telemetry in the F# compiler I didn't know about? Or you mean that OOTB VSCode has telemetry, so you won't use it for your F# code, opting instead for a different text editor / IDE?
How do people building these things become convinced this is ok?