[ChuckMcM]

First, let me say that I think what they did was wrong and it should only be opt-in and clearly stated.

That said, having managed fleets of machines that were nominally running the "same" software, getting updates from all of them is a really powerful debugging tool. Once you get above about 1,000 machines logs comparison of all the machines immediately surfaces software issues (happens on all machines), connectivity issues (machines in a certain area), bad machines (unique problem signature), and environmental issues (time of day correlation with other things like power/temp/humidity/etc).

And that gives you a bit more courage to release things early because you'll see problems faster and can fix them.

So with a typical roll-out of 10% of the population followed by an additional 15% followed by the rest, you can catch a lot of errors and 75% of your population sees a really good experience (and in web services where 66% of the populations the minimum requirement for delivering rated service you can often get close to 100% uptime).

Does that justify their action? No. But since you really don't need everybody to participate to get the benefit I could see a path where you opt in for early access to drivers which requires the telmetry, and people who are ok waiting for the driver to be clean in the 'canary' population get a driver without telemetry.

[white-flame]

This isn't about just technical issue monitoring. Their privacy policies and such include sharing your personally identifying information (email, full name, etc) with 3rd parties & advertisers, and associating your random web browsing with your video card's account.

Remember that it's not just games, but your web browser, media players and anything that might use hardware acceleration runs through these drivers, which then reports back to nVidia. Plus, all telemetry data is unencrypted during transmission.

On the plus side, it looks like this is only transmitted back to the mothership if you're running GeForce Experience, which I don't think is even available on Linux.

[ChuckMcM]

In my experience it is more nuanced than that, engineering will add a feature or capability and someone will come along and find a use for it that was unintended. At the search engine where I worked we came up with ways to inspect and adjust host rank and page rank such that we could knock out bad sites from the search results that were doing a good job gaming the system. But that biz dev folks also noted the same pieces could be combined to favor someone who advertised with us over someone who didn't.

The engineering feature is designed with pure intentions, keep crap out of the results stream, but the mechanism is amoral. If you're executive team wants to exploit it for more money so that the company does better financially, your options are limited as an engineer.

[white-flame]

At a data design level, they could (and should, by default) at the very least keep all grossly personally identifying information out of the data collected. That's just simply responsible engineering in the modern era.

They could have implemented telemetry without requiring you to link your graphics card to your email or Facebook account, as GeForce Experience never required that in the past and still worked fine.

This isn't execs using innocuous data that the good engineers simply happened to have. This is new and specific personal data collection and association of information that they have no business collecting, except for the sole purpose of invading their users' privacy for side profits.

[ChuckMcM]

I'm not disagreeing, I'm just saying it is hard to build tools and features into your project that your own business development team can't abuse.

[arkitaip]

Isn't the easy option to allow customers to participate in the program? Their software is already so buggy and bloated that a simple checkbox that ANONYMOUSLY sends telemetry would be trivial to implement.

[Esau]

This is something that, I hate to say, will probably take a law to fix. Companies in general have shown a resistance to understanding why this is a problem (even when well intentioned).

[ocdtrekkie]

This. As long as consumers have no way of knowing whether or not their information is being transmitted, what information it is, and how it will be used, corporations will continue to do it at will. We need strong user data protection laws here in the United States that make it extremely clear what rights you are signing away any time you allow a service to collect personal data about you.

[doubt_me]

Its bigger than that. We are at a point where I would pay a monthly premium in order to keep my privacy and would rather have a bunch of free tiered alternatives that are all explicitly non-private. Because lets be honest with ourselves nobody but us really gives a shit about their privacy. And it is a billion dollar industry. Let people who do not care keep the free shit at an up front cost rather than all this running around corporations do now. Or I can even see a future where the government gives you free internet in return for data mining 100% of it.

I would actually probably use that.

[hydragit]

Actually, even if you're paying, the service operator may be mining your data. This is the case for example with cell phone operators: you pay a monthly fee, and yet they use your geolocation to send you custom advertising, sell market studies, etc.

[ocdtrekkie]

Paying for email sounds weird to most people these days, but now that I do, I'd never have it any other way again.

[grzm]

I'm days away from moving to fastmail. May I ask what service you're currently using? Are you happy with it?

[ionised]

Same here. Three separate, paid services. I'd never go back to something like GMail again.

[michaelt]

  And that gives you a bit more courage to release things
  early because you'll see problems faster and can fix them.

I gotta say, if the major benefit of telemetry is that vendors can test less before they release, that sounds like a bad thing not a good thing for users?

[Namrog84]

Bad for few users. But great for most.

Testing less means they can release sooner.

Here is a bad example. New game comes out. Graphical driver glitch crash. Typically would take 2 weeks to get to public as whole. Now with this they can release in 2 days to 10% fix up and release to whole by end of first week. So 90% of population got a fix sooner(1 week) And some percentage of the 10% may have had some annoyances. Sucks for them. But if they roll out in a way that minimizes this. It'd be beneficial to all overall.

[michaelt]

Here's a different example: A new driver improves game performance for 99.9% of customers, and disables graphics output for 0.1% of customers. The bug could have been caught with a few days of automated testing in a test lab with lots of different cards and system configurations, but that's slow and costs a lot of money to maintain - chucking it over the wall to consumers is quicker and cheaper.

Seems to me getting the poorly tested drivers earlier isn't much of a deal for users. After all, just because I'm in the 99.9% this week, doesn't mean I will be next week.

[Klathmon]

But in my experience there are a few more 9's on your "works" number for a specific issue.

Hardware-related problems are so fucking hard to debug in a lab. Many of the issues i've dealt with in the past only show up with pretty esoteric hardware mixtures, or are so specific that i'd be surprised if more than one person has that exact config on the planet.

A test lab might be able to get maybe a hundred different configurations. But in the wild, something like 60% of our users are on unique hardware configurations (as in nobody else has that exact set of hardware among our users).

Telemetry is pretty much the only real way you are going to be able to support a wide range of hardware when you start interacting with it on the lower levels, because buying and assembling all the different configurations is all but impossible, even if you prioritised it.

[SixSigma]

My AMD drivers & Steam Client already report crashes