As eager as I am to hop on the Windows 10 hate bandwagon, I've gotta agree with you. The author should disclose the original source if nothing else.
At this time, the closest thing to a source is "Some Czech guy did a traffic analysis of data<link to an unrelated disclose.tv page> produced by Windows 10, and released his findings the other day."
This is just appalling journalism if nothing else, and hurts both sides of the case.
Thanks for providing a source. The translated link you provided doesn't inspire a lot of confidence in their reliability. I'll believe what they're saying once they provide the exact methods they used to get these results in order to corroborate them.
At least they plainly state at the beginning of the article:
"Note: Some readers have commented that the original source for the article is of questionably validity. If anyone can confirm or refute the original author's finding with actual data, please let me know in the comments, and I'll update this post accordingly."
It would be good to see something confirming any of this. Even if all of this article's claims are false, it would be good to see a deep dive into exactly what is actually being sent and when.
Saw this claim the other week, while I've already disabled many of these things, the largest point of concern to me here is the microphone transmission and it's one I've been unable to independently verify or deny. Anyone done any deeper digging on this?
What, really? Your source is "some Czech guy?" I mean maybe it's true and maybe it's not, but at the same time an article that presumably doesn't link the original article because they don't want people just going directly there and which might be quoting the most lunatic fringe conspiracy theory is not worth taking particularly seriously. A five-minute search tends to indicate that this guy:
Note the comment of "Stephen Smoogen" to the apparent systemoverlord's "check" of the claims (which was on VM, and using pre-release Windows):
"There is also a difference if your network link is detected as being 'metered' (in that if the system thinks the network is metered it will not send as much data). I have noticed that for other windows what a system sends if it is on 'real' hardware or virtual machine are also different. My guess is that Microsoft assumes a 'guest' machine isn't real and worth checking.
To test the article, I would expect that I would need to set up a hub, put the box and the capture system on it and run it on a live hardware with a 'real' account. Without doing that then 'results will vary' is going to be the norm."
I am not sure that I buy this story. A second source would be good.
I was really enjoying Windows 10 and got beat up on HN after defending it last week. That said a few days ago I switched from having 3 linux laptops and 1 windows 10 laptop to having 4 linux laptops.
I think that Microsoft, after fixing some privacy issues, has a winner in Windows 10 but I decided to just make my life simpler.
This is straight up blogspam. Notice how all the links in this article link to the same website?
Here's the original source (In fairness to OP, it's not in English.) [1]
Here's a summary of his findings in English by a guy that isn't trying to drive in ad-traffic. [2]
Now that that's out of the way, here are my thoughts:
Some of this is certainly true and they don't try to hide it-- just last weekend I showed my father the above version of this article. He's a pretty pro-microsoft/anti-FOSS/internet-privacy-is-a-ridiculous-idea guy, but even he was a little shaken to see that we could open up control panel and disable something like the reporting of all of your keystrokes-- assuming it actually disables it.
With regards to data collection and report that users have no control over, I think it's probably true that Windows 10 is capable of this and has code in place to do it, but the idea that it's actually doing this at a large frequency on every user is pretty ridiculous. If it were truly as bad as he's painting it it would be as easy as opening Wireshark, right? We can look at blobs in Windows 10 and find hardcoded strings, and while we probably can't figure out what it's doing with them, we can at least confirm that this guy didn't make it all up. So why hasn't anyone?
I dunno. But IMO the implications of this being true are moot. We already live in a world where Chome is telling Google what you're saying to your microphone[3], where OSX is telling apple where you are and what you're typing into spotlight[4], and where the most popular Linux + GNU distribution is putting out releases that tell Amazon what you're searching for in a default install[5]. Is Microsoft jumping on this same wagon really such a monumental shift when other operating systems and browsers are already doing it?
I'm glad that people are upset at the idea though, and if you're with me so far I'd like to make the case that if you're upset about this you ought to be upset about a possible future that is much worse. I'm willing to eat these words 10 years from now if I'm wrong, but I truly believe that two imminent events are going to be catastrophic for software freedom.
1. From a software-engineering standpoint, LLVM is superior to GCC. It is well-engineered and will inevitably outperform GCC to a great extent. It's modular and that means that adding new languages and targets are far easier. The future of compiler development is decidedly LLVM-like compilers. But this modular nature (and the license) allows ARM/Intel/NVidia/Ati the ability to simply put out blobs that extend LLVM. Nvidia/CUDA already do this. [6]
2. Moore's law is dying, and it's going to result in a transformation of hardware. (Cannot recommend this article enough, though my conclusion from his reasoning is the opposite.) [7]
I know it sounds ridiculous at first thought-- "LLVM is under a FSF-approved license", you say. "Even if it wasn't, nonfree tools have never meant that free tools cannot exist besides them." "The author of that article made a pretty compelling case that we're headed towards more open hardware, not less."
And that's all true.
But consider Apple's iOS app-store lockdown and how anti-developer it is. A $100 fee to even begin developing apps? Approval processes? It's completely appalling that we develop backends for services on the shoulders of free software and yet we're at the whims of Apple when we wish to actually get it in the hands of users. But that's the small price we pay to be able to get our software in the hands of the common man's iPhone.
The fatal assumption in the IEEE article is that developers decide what languages or architectures we wish to target. I don't think we do. The market does; rest of society does. We make software for the products they choose. Open hardware is something that we...
22 comments
[ 2.9 ms ] story [ 57.0 ms ] threadAt this time, the closest thing to a source is "Some Czech guy did a traffic analysis of data<link to an unrelated disclose.tv page> produced by Windows 10, and released his findings the other day."
This is just appalling journalism if nothing else, and hurts both sides of the case.
EDIT: I retract some parts of that. The link to the source (in Czech) is tucked away discreetly at the bottom of the page: http://aeronet.cz/news/analyza-windows-10-ve-svem-principu-j...
http://www.respekt.cz/z-noveho-cisla/putinuv-hlas-v-cesku
It was discussed in the top comment and the replies to it: https://news.ycombinator.com/item?id=10053420
At least they plainly state at the beginning of the article: "Note: Some readers have commented that the original source for the article is of questionably validity. If anyone can confirm or refute the original author's finding with actual data, please let me know in the comments, and I'll update this post accordingly."
It would be good to see something confirming any of this. Even if all of this article's claims are false, it would be good to see a deep dive into exactly what is actually being sent and when.
There's already lots of commentary there on the results.
https://systemoverlord.com/blog/2015/08/16/so-is-windows-10-...
hasn't been able to reproduce the traffic analysis on casual inspection.
Edit: Here's what I believe to be the original source of the claim:
http://aeronet.cz/news/analyza-windows-10-ve-svem-principu-j...
And the translation:
http://localghost.org/posts/a-traffic-analysis-of-windows-10
"There is also a difference if your network link is detected as being 'metered' (in that if the system thinks the network is metered it will not send as much data). I have noticed that for other windows what a system sends if it is on 'real' hardware or virtual machine are also different. My guess is that Microsoft assumes a 'guest' machine isn't real and worth checking.
To test the article, I would expect that I would need to set up a hub, put the box and the capture system on it and run it on a live hardware with a 'real' account. Without doing that then 'results will vary' is going to be the norm."
[0]: https://cs.wikipedia.org/wiki/Aeronet.cz
I was really enjoying Windows 10 and got beat up on HN after defending it last week. That said a few days ago I switched from having 3 linux laptops and 1 windows 10 laptop to having 4 linux laptops.
I think that Microsoft, after fixing some privacy issues, has a winner in Windows 10 but I decided to just make my life simpler.
http://www.disclose.tv/news/The_Russian_Sleep_Experiment_dur...
(creepypasta if you don't recognize it)
Here's the original source (In fairness to OP, it's not in English.) [1]
Here's a summary of his findings in English by a guy that isn't trying to drive in ad-traffic. [2]
Now that that's out of the way, here are my thoughts:
Some of this is certainly true and they don't try to hide it-- just last weekend I showed my father the above version of this article. He's a pretty pro-microsoft/anti-FOSS/internet-privacy-is-a-ridiculous-idea guy, but even he was a little shaken to see that we could open up control panel and disable something like the reporting of all of your keystrokes-- assuming it actually disables it.
With regards to data collection and report that users have no control over, I think it's probably true that Windows 10 is capable of this and has code in place to do it, but the idea that it's actually doing this at a large frequency on every user is pretty ridiculous. If it were truly as bad as he's painting it it would be as easy as opening Wireshark, right? We can look at blobs in Windows 10 and find hardcoded strings, and while we probably can't figure out what it's doing with them, we can at least confirm that this guy didn't make it all up. So why hasn't anyone?
I dunno. But IMO the implications of this being true are moot. We already live in a world where Chome is telling Google what you're saying to your microphone[3], where OSX is telling apple where you are and what you're typing into spotlight[4], and where the most popular Linux + GNU distribution is putting out releases that tell Amazon what you're searching for in a default install[5]. Is Microsoft jumping on this same wagon really such a monumental shift when other operating systems and browsers are already doing it?
I'm glad that people are upset at the idea though, and if you're with me so far I'd like to make the case that if you're upset about this you ought to be upset about a possible future that is much worse. I'm willing to eat these words 10 years from now if I'm wrong, but I truly believe that two imminent events are going to be catastrophic for software freedom.
1. From a software-engineering standpoint, LLVM is superior to GCC. It is well-engineered and will inevitably outperform GCC to a great extent. It's modular and that means that adding new languages and targets are far easier. The future of compiler development is decidedly LLVM-like compilers. But this modular nature (and the license) allows ARM/Intel/NVidia/Ati the ability to simply put out blobs that extend LLVM. Nvidia/CUDA already do this. [6]
2. Moore's law is dying, and it's going to result in a transformation of hardware. (Cannot recommend this article enough, though my conclusion from his reasoning is the opposite.) [7]
I know it sounds ridiculous at first thought-- "LLVM is under a FSF-approved license", you say. "Even if it wasn't, nonfree tools have never meant that free tools cannot exist besides them." "The author of that article made a pretty compelling case that we're headed towards more open hardware, not less."
And that's all true.
But consider Apple's iOS app-store lockdown and how anti-developer it is. A $100 fee to even begin developing apps? Approval processes? It's completely appalling that we develop backends for services on the shoulders of free software and yet we're at the whims of Apple when we wish to actually get it in the hands of users. But that's the small price we pay to be able to get our software in the hands of the common man's iPhone.
The fatal assumption in the IEEE article is that developers decide what languages or architectures we wish to target. I don't think we do. The market does; rest of society does. We make software for the products they choose. Open hardware is something that we...