Live data from Hacker News

Microsoft: Require user consent before sending any telemetry

github.com

1–10 of 129 posts

Re: Microsoft: Require user consent before sending any telemetry

#2
I'm not qualified to weigh in on the merits of the request, but asking a corporation to change something and then throwing in a bunch of legalese about compliance and GDPR seems like an excellent way to guarantee that the poor reviewer of the requests is not going to deal with it, let alone quickly.

At best, they raise it to their internal legal contact. The inhouse lawyer rapidly advises them to not respond in any written or recorded medium. Issue goes nowhere.

At worst, they realize that this is a hairball with "vaguely legal stuff" and decide to review some other issue instead for a more productive and less stressful day. Issue goes nowhere.

Re: Microsoft: Require user consent before sending any telemetry

#3
No answer is forthcoming from the VS Code team, because they know you won't like the answer.

Microsoft trawls their[1] endpoints mercilessly for every bit of telemetry that they possibly can, and they go out of their way to prevent customers from disabling this.

Windows 10 or 11 with Office requires something like 200+ individual forms of Microsoft telemetry to be disabled!

Notably:

- They keep changing the name of the environment variables[2] that disable telemetry. For unspecified "reasons".

- They've been caught using "typosquatting" domains like microsft.com for telemetry, because security-conscious admins block microsoft.com wholesale.

- Telemetry is implemented by each product group, which means each individual team has to learn the same lessons over and over, such as: GDPR compliance, asynchronous collection, size limiting, do not retry in a tight loop forever on network failure, etc...

- Customers often experience dramatic speedups by disabling telemetry, which ought not be possible, but that's the reality. Turning off telemetry was "the" trick to making PowerShell Core fast in VS Code, because it literally sent telemetry (synchronously!) from all of: Dotnet Core, PowerShell, the Az/AAD modules, and Visual Studio Code! Opening a new tab would take seconds while this was collected, zipped, and sent. Windows Terminal does the same thing, by the way, so opening a shell can result in like half a dozen network requests to god-knows-where.

[1] You thought, wait... that it's your computer!? It's Microsoft's ad-platform now.

[2] Notice the plural? It's one company! Why can't there be a single globally-obeyed policy setting for this? Oh... oh... because they don't want you to have this setting. That's right... I forgot.

Windows: https://learn.microsoft.com/en-us/windows/privacy/configure-...

PowerShell: https://learn.microsoft.com/en-us/powershell/module/microsof...

DotNet Core: https://learn.microsoft.com/en-us/dotnet/core/tools/telemetr...

Windows Terminal: https://github.com/microsoft/terminal/issues/5331

Az module: https://learn.microsoft.com/en-us/dotnet/api/microsoft.azure...

Etc...

Re: Microsoft: Require user consent before sending any telemetry

#5
Have you noticed that MS mostly stopped using EEE, and changed strategy to just ignore rules/laws/licenses, and wait to see what happens? We hear it frequently that "today's MS is not the same as the old MS", but I have my doubts.

This particular one just the latest. But the really big one (IMHO) is the one where they simply started to ignore EFF[0], when they were asking them about the copyright status of co-pilot. If the court decides against EFF, that will have a lot of effect on the legality and enforcement of most of the OSS licenses (though I'm an armchair-lawyer, not even in the US). Fun times ahead.

[0]: if I remember well, it was EFF, who mentioned that MS stopped responding to them. I have found the lawsuit, but filed by not by the EFF. Google is more useless by the day.

Re: Microsoft: Require user consent before sending any telemetry

#7

Have you noticed that MS mostly stopped using EEE, and changed strategy to just ignore rules/laws/licenses, and wait to see what happens? We hear it frequently that "today's MS is not the same as the old MS", but I have my doubts. This particular one just the latest. But the really big one (IMHO) is the one where they simply started to ignore EFF[0], when they were asking them about the copyright status of co-pilot.…

I think this is a tendency of all internationals mega corporations. Law is not homogenous around the world, and since you are consequently anyway in violation, you learn how to use that in your favor and ignore it for quite a while. And then, once its start to be annoying, you can finance an army of lawyers to delay or even change the law.

For one part it is quite reasonable to work like that, on the other side it is really unethically and bad for the society as a whole.

Re: Microsoft: Require user consent before sending any telemetry

#9

To be fair if someone comments to me with things like: > Please give an answer within the next week until the 16th of June. I wouldn't respond to them either out of spite

The issue with society or one of them, is thinking its acceptable for a corporation breaking law to feel spite, the guy was not talking to a person, was talking to a shitty corp breaking law

Re: Microsoft: Require user consent before sending any telemetry

#10

No answer is forthcoming from the VS Code team, because they know you won't like the answer. Microsoft trawls their[1] endpoints mercilessly for every bit of telemetry that they possibly can, and they go out of their way to prevent customers from disabling this. Windows 10 or 11 with Office requires something like 200+ individual forms of Microsoft telemetry to be disabled! Notably: - They keep changing the name of t…

> They've been caught using "typosquatting" domains like microsft.com for telemetry, because security-conscious admins block microsoft.com wholesale.

This seems interesting. Do you have any references for this? I would assume that the main use of such typo-squatting domains is a simple redirect, a la [0][1].

[0]: https://gogle.com [1]: https://gooogle.com

Post reply on HN