I mostly leave telemetry on in products I care about specifically so that my advanced use cases get logged.
But if a whole subset of users exclude themselves from that tool, they're going to get the UX that's only as good as the other tools in the toolbox are capable of building.
But telemetry gives web developers an extremely simple and convenient tool to know what users are actually doing without even inconveniencing the users with explicit questions. I've done web development with a good telemetry set built into a page, and it is extremely informative regarding how users actually use the tool, as opposed to how the UX designers have predicted flow through the tool will be.
To give a concrete example, a user might tell you that configuring permissions is "hard," and sitting with them during over the shoulder testing (which is expensive) might tell you a little bit about why. But without even asking the user, page telemetry can tell you that they are making a transition jump from the permissions configuration page to the page listing all of the resource names, because that's what's slowing them down---the UI didn't give them enough information to configure the resources because we assumed they knew what the resources were named.
For a browser, anonymized usage stats can tell you whether most users keep all their bookmarks flat at the top of the bookmark bar or deeply nested in multiple subfolders, and that's usually valuable for deciding whether you want to emphasize a flat bar or folder management in the design.
If most power users disable automatic anonymous telemetry and also use deeply-listed folders, no one should be surprised if deeply nested folders doesn't get better.
People who don't show up to vote also "deserve" a good government by virtue of being people who have to live in a governed society. It's harder to make one for them if the system for selecting leaders is missing their input, regardless of what they deserve.
Popping out of the government analogy and back to software, power users are also in a position where they are more capable of adjusting their experience to suit their needs. All things being equal, a company with finite resources to develop software should dedicate those resources to assisting the non-power users more often than power users.
b) for being "not a good idea", it's pretty much industry standard now for everything from business software to video games.
So you're claiming that it is typical for software with telemetry support to ignore your choice and still send telemetry about you turning off telemetry? That sounds wrong, but I cannot say I investigated this deeply.
> b) for being "not a good idea", it's pretty much industry standard now for everything from business software to video games.
As I understood the discussion, we were in fact discussing whether this is a good idea and whether it makes sense, so I think it's fair game to comment on it. As for it being an industry standard, that sounds like an overgeneralization. It is certainly not typical of software I use.
No; I'm saying missing data leaves holes that can be measured. They know, for example, how many people have downloaded Chrome and how many daily Chrome users they get at google.com (because Chrome will still send a valid UA string if it has telemetry turned off). They can estimate how many users have telemetry turned off from those signals to a pretty decent degree of accuracy; certainly enough to know whether telemetry is telling them about 90% of users of 30%.
For (b), I'm curious what software you use. It's pretty standard in games, online apps, and business software. It's absent in a lot of open-source (mostly because a lot of open-source lacks a centralized vendor who would be willing to pay the cost to collect and interpret that data to improve the software).
I avoid online apps, I don't play a lot of games (and if I do, they're not big titles which are likely to have telemetry) and yes, I primarily use FOSS.
> (mostly because a lot of open-source lacks a centralized vendor who would be willing to pay the cost to collect and interpret that data to improve the software).
This is almost surely an element of it, but I think a respect for privacy and a general distaste for telemetry among FOSS users are more important.
* Apple Watch battery cycle count (not viewable in any UI but is viewable in telemetry logs)
* Clues about why a particular app recently crashed