Hacker News Reader

Comment by treyd | Hacker News Reader

> don't fool yourself if you think this won't come to the Linux world.

I'm curious what you mean by this. I'm not necessarily rejecting the point, but I also don't see how this could happen without substantial shifts in the industry first.

View on HN

Yeah, curious here too. Torvalds would need to pass first I think, and I just don't see other major players like RedHat, Google, Canonical, or Valve introducing this themselves or agreeing to do it in aggregate. And as end users we could still fork and patch it out. Some shitty company might try but I don't think it would stand.

Lots of bad ideas have come to Linux, like non-consensual telemetry, mobile-first interfaces etc. Don’t believe? Run OpenSnitch.

Traditional CADT means features get lost over time.

It is not immune from these forces, just not a focus by the powers that be. Fewer developers remember the good old days of Y2Kas as well, meaning they don’t resist these forces instinctively, since they grew up in iOS captivity.

Telemetry is anything including a process list. If you're talking about eBPF it's also used for debugging and server fleets and recently in basic task managers. Any data can be used to take a magnifying glass to the system. The kernel has literally thousands of toggles for this from networking to threading. And yes a program can see what your kernel supports and yes it can refuse to run if you're not running a specific kernel with a specific feature. How do you think programs like open snitch even work?

https://github.com/evilsocket/opensnitch/wiki/monitor-method...

How do you think they work?

I’m talking about mostly about consent above, maybe wasn’t clear.

Reply on news.ycombinator.com