> "What if -every- one of these started doing their own telemetry, their own envvars, etc?"
What bad thing are you suggesting would happen if they did? Your computer and internet connection can't handle four thousand strings or four thousand HTTP POSTS, or four MB more disk space of telemetry libraries? I bet it can. This isn't a technical problem, it's a control and consent problem.
For one thing a classic way of downplaying metrics, "we're only logging X bits of information", turns into 3000*X. And here X is huge already.
If I get access to detailed metrics from go, gcc, make, tar, gzip, bash, python... of course I can tell which programs you have been running (and frankly, I'm disgusted)
I wouldn't be so sure it's not also a technical problem, the limit of execve can be as low as 128kb which for 4,000 strings gives a maximum of 32 characters per NAME=VALUE environment value
Free disk space can be as low as zero, but we don't blame the tool makers for adding an extra 100Kb or 20MB, we blame the computer owner for not having enough disk space to install the thing they chose.
Honestly, I do too, but the world doesn't. If you said "I have 3,500 binaries on my system, imagine if they ALL added 1MB" the reply would be "3.5GB is about twenty cents of NVME storage space" not "oh my, you're right that would be intolerable".
What bad thing are you suggesting would happen if they did? Your computer and internet connection can't handle four thousand strings or four thousand HTTP POSTS, or four MB more disk space of telemetry libraries? I bet it can. This isn't a technical problem, it's a control and consent problem.