> it seems that most of them have never worked on an user-facing product

If you have worked on user-facing products, then I think you do understand. I think you just don't like the answer.

Here it is in a nutshell: In the general case, how do you distinguish 'telemetry' from 'surveillance'?

And the answer is you can't, because the difference between 'telemetry' and 'surveillance' lies in the intent of the consumer of that data. Technically useful information collected in good faith today can become intrusive surveillance tomorrow with a court order, or business pivot, or just a bored insider & lack controls.

Now consider the fact that the author of every stupid new piece of code out there now seems to think it is so special it deserves to be allowed to send... who knows what back to some opaque collection endpoint as the cost of executing.

The only reactions in response to this mess that don't cost a huge amount of overhead are "block everything" and "have your way with me". Guess which a technically inclined user picks.

Granularity. Telemetry data can be privacy-respecting (e.g. anonymously aggregated) while still being useful.

E.g. "143 people flew from NYC to LA this month" vs. "John Smith flew from NYC to LA at 2:00 pm on Tuesday."

You can do it with cameras and ID plate detection tools, which allows tracking where anyone is at any time.

You can also put magnetic loops into the roads, which just tells you how many drive over a certain area of the road, without any individualized tracking.

Good telemetry is like the magnetic loops. But nowadays very few implementations are that privacy respecting

We promise it's anonymized.

For starters, consent. Telemetry generally supposes that the user opts into, either directly or by accepting the terms of using a service, and is free to walk away.

I find the increasing usage of the word surveillance completely silly and ideological. If the NSA vacuums data from some underwater cable that's surveillance, when I exchange information, aware of what's being handed over in exchange for some service, that's a trade.

Who has the time to read (let alone "become aware") of the full scope of that?

[1] https://docs.microsoft.com/en-us/windows/privacy/windows-11-... [2] https://docs.microsoft.com/en-us/windows/privacy/required-wi...

I'm all for informed consent, create laws to give users transparency and control over their data. But if users have that and they still decide that exchanging info for better services is worth it, calling it surveillance is I think not appropriate.

Case in point, Vscode. I'll gladly give them telemetry access for getting a better editor.

That's nice. I find downplaying the reality of surveillance to acclimate people to even more of it very dangerous and ideological.

By never sending PII to your telemetry system. Or if you must use PII (you don't need to) then you lock it away with proper access controls.

Leaving your veiled insult aside, I think it's pretty easy to see but you don't want to listen to it. Namely: telemetry and surveillance are literally the same thing and the only difference lies in the goodwill of the steward of the data.

I believe we've been witnesses, dozens of times, here on HN, about yet another "private data leak" because some company can't be bothered to hire an intern for $2000 a month to do proper data and security hygiene of some meager 3-4 disks that don't even total 10TB.

So why should I trust companies by default to take good care of my data AND to use them for good only, when all the historical evidence points at them being unable to do either? Shouldn't I take the matter in my own hands? I answered yes and so I did. My PiHole black-holes anything such I can think of and until that gets criminalized I'll be freely admitting it anytime when asked.

I think it would be far less hostility if we as users got access to the information ourselves. E.g. I would think it would be very useful to see stats on which features I'm using in firefox, how often, which features I'm not using, how many tabs I have (compared to the average user) etc., and I would be much more happy to share that data with firefox if I can see it first.

Go to about:telemetry

There are several pages detailing all of that information, and you can also get all of that data as the JSON payload if you want to munge and display it yourself.

[0] about:about

You could probably find more with searchfox, using a better query than https://searchfox.org/mozilla-central/search?q=about%3A&path...

> This is a list of “about” pages for your convenience.

> Some of them might be confusing. Some are for diagnostic purposes only.

> And some are omitted because they require query strings.

Some are omitted. Like about:blank.

https://docs.microsoft.com/en-us/windows/privacy/diagnostic-...

Once upon a time, Firefox probably had a normal ratio of privacy-conscious users to privacy-indifferent users.

But if the privacy-indifferent users are increasingly lost to Chrome, FF is probably left with an unusually privacy-conscious userbase.

I'd bet a graph of their privacy-focused:indifferent users over time would be U-shaped. Very high before, oh, I dunno, FF2.0 when the only reason non-nerds had it was because their nerdy friend/relative installed it for them. Lower when they were advertising all over the place and marketshare was growing ('08-'12? My memory's hazy). Higher now, as their userbase continues a years-long contraction and shifts toward being mostly nerds, again.

I'll leave it to others to argue why it's a privacy problem, but I'd like to push back on the "useful" and "feedback" arguments. It's appealing to say "X% of our users use this; we should prioritize it", but Mozilla has a charming history of saying "only Y% of users use this feature; we'll remove it", and for bonus points, "after we pushed this option into a really obscure sub-menu, nobody uses it, so now we can remove it while claiming that nobody cares". That is, telemetry isn't even useful for the thing that people claim it's good at, because it only provides raw data that gets used to justify things regardless of reality.

Telemetry would fix this, but it's not our right to collect data from other people's machines. So our best bet would be to detect the condition we expect to cause problems, notify them in advance, and allow time for feedback before implementing a change.

Is it less pretty and a worse user experience than telemetry? Absolutely. But when you realize you have no right to data collection on someone else's machine, telemetry isn't a viable option, and we would betray people's trust by doing it.

We've resorted to the "scream test". When there are features that are expensive to maintain that we suspect aren't being used (based on surveys, interviews, and usability tests) then we insert a notification when the feature is used announcing it's pending removal in some subsequent version. The notification includes contact information so if we get a higher-than-expected number of complaints then we may keep the feature.

Edit: typo.

You seem to be assuming that your users are always online and not sitting somewhere isolated, the ideal place for platforms old enough that you want to stop supporting them. Telemetry only tells you if something is there, it doesn't tell you if something isn't.

Telemetry showed that users overwhelmingly approve of shitty option B!

The point of the comment is to show that telemetry is used to conclude things that aren't true, and it absolutely is, and only says the subset of things that someone wants to hear, or that even occurs to them to ask.

It's not even really a seperate issue, because even if you think you are only concerned with say how long a certain operation takes or how frequently it fails & retries, the more important issue may be why or whether that operation was even used at all in the first place.

This aspect of telemetry is simply a fact to be aware of and stands on it's own, and does not necessarily mean to say "therefor telemetry is an invalid tool that should not exist and should never be used" It just means it's incomplete at best and very susceptible to both deception and self deception.

People don't want to be spied on. Yes, you can find positive uses for it, but you can find positive uses for many negative technologies. If anything some of the most pervasive negative technologies also have positive uses to sweeten the pill (or even give the impression that this is a sweet and not a pill you are swallowing).

[0] Session-recording spyware "telemetry" products practically are this, and are widely used. Watching their session playback, or ability to monitor live users on "your" site (but actually, on their computer) is disturbing.

In case of mozilla specifically telemetry didn't help them to not destroy firefox's UI.

Also go read the docs at https://docs.telemetry.mozilla.org/. Those will give you insights into every way they use data.

I've never seen a company that's more open about their data usage.

> …Firefox has demonstrated a strong ideological bias towards those groups that love to engage in cancel culture.