Hacker News Reader

Comment by fffrantz | Hacker News Reader

While this is indeed worrisome, I’m not really sure Microsoft is any better than Amazon from a pure end-user point of view.

Microsoft has been a telemetry champion and has been collecting troves of personal data through Windows 10, all the while making it almost impossible for the average user to disable it (and re-enabling it with every major update).

So if Microsoft goes the same route as Amazon (resigning every app), it’s just a matter of choice between a rock and a hard place. On the other hand, even if they allow for the app developer to keep its signature, Microsoft is still probably very capable of gathering data and telemetry through their Android implementation.

Either way, and until it becomes illegal to collect data without explicit consent, I’m pretty sure we’re screwed…

View on HN

I sometimes wonder how profitable it actually is for Microsoft to be so pushy on data. What I mean is: I understand that MS is quite zealous internally about telemetry, but how effective is this data in terms of actually improving their profitability?

From the user side, I am quite sure it is hurting the experience. One example is logins. MS is pretty bad about requiring logins for products which don't really require it in order to complete the user journey - for instance I bought the Halo collection on Steam, and in order to play an offline, single-player game I have to log into an MS account. As a result I haven't played the game in months, because I have to use this MS account that I don't even want to have, and I can't be bothered to reset the password for when there are other games I can play which just start up with no arbitrary hoop to jump through.

And what does MS actually get out of it? I guess they can optimize their dark UI patterns to be better at tricking people into setting Edge as their default browser?

Being data driven is good, but it seems to me it's not working if the end result is compromising the core user experience.

User telemetry replaces what were once focus groups and market research. Without telemetry, Microsoft would be unable to determine what products were used, what features were used, and unable to determine prioritization for backwards compatibility and bugfixes, because it chose to rely on telemetry for basic business operations.

User telemetry replaces what were once focus groups and market research.

These things still exist, and many companies still use focus groups and market research. The company I work for, for example, does this extensively because I'm in the healthcare space and the legal department doesn't want us collecting information about our users without them really really really knowing its happening.

Another example is MailChimp. It not only does video interviews with its users to determine how they use the product, it pays you for your time.

Microsoft has the money and staff to do this basic research. "Telemetry" is just a synonym for lazy and cheap.

Yes and even besides traditional methods like user research, there are lots of indirect ways to measure the success of a product without resorting to invasive data collection practices.

For instance, you can track downloads, page views in documentation, the number of times a certain endpoint is pinged on the server etc. while still respecting the privacy of the user.

The practice to track a user's behavior across a wide array of products imo crosses a line, and is not needed to inform good product development.

A company could probably sell better to you if they hired a PI to follow your movements, and tapped your phone line, but that doesn't give them the right to do so.

I suspect there's also a fear about the accuracy of self-reported data. If you ask a user if he does XYZ, he may say 'no', because of a myriad of reasons (misunderstood the question, embarrassed, not allowed to disclose, etc.), but if you instrument it up and see he clicks it 200 times a day, that's probably the truth.

The problem is that analytics in a vacuum are just as misleading as testimonial data, if not more so. "0.6% of users touvh XYZ and only use it for an average of 17 seconds per year" could mean that it's an unloved feature that can be sunset, or that it solves a single niche task very completely and perfectly, and ripping it out will cost those 0.6% of users five hours each to replace.

I think there's also the risk of talking yourself into bad decisions with data as a justification. For instance, if you have a KPI to increase X, and making your UI harder to understand confuses users into doing X more often, from a data-driven perspective you've done the right thing, but you've made the product worse.

That might be the case, but it seems like mostly what they've been able to do with it is build products like Teams which nobody asked for and nobody wants, but people have to use because MS has some leverage to get it used in certain organizations.

There are exceptions like Office and VSCode, but it seems like for the most part MS products are things which people curse under their breath and don't like to use. Part of that is because of things like how their aggressive approach to telemetry bleeds into the UX, and it just seems a bit off if the tool you have for optimizing your product is actively making it worse.

You may not want Teams, but do not live in the illusion that nobody wants Teams. As proof: I want Teams.

What do you prefer about teams over alternatives? It may be anecdotal, but I have only ever heard people complain about teams.

Teams video-chat is great, easily better than things like Cisco's offering (and zero-security offerings like Zoom).

On the other hand, it's also trying to be a weird combination of Slack and Facebook at the same time (only less searchable than either) and that's pretty terrible.

Really think so? I think the free alternative jitsi is so much better than teams. I use teams for work and its video is pretty slow. Jitsi (I use this for my local makerspace) achieves a much better frame rate. Which makes it more realistic and improves communication.

It's also so much lighter on the computer than teams.

They complain about it in the same way they complain about computers in general. Sometimes it doesn't work right. That doesn't mean that 99.5% of the time it doesn't get the job done.

But for instance people like to be on slack, and Zoom calls were fun before everyone got sick of them. The sentiment around Teams whenever it has come up around me has always been one of pained obligation.

I haven't used teams in 3 years (I use what clients use, mostly it's slack) back then I really liked wiki + channel combo, it was super simple to keep channel specific notes in one place and everyone used it since it was right there, fast to update and search - way better than pinned messages which I find almost useless

From my experience, the only thing teams does mostly better than the alternatives, is the moment you're actually in a call. The audio quality seems to better than something like slack or alternatives in my experience.

But once out of a call you end up in a messy maze of random cluttered functionality and different, confusing styles of chat options, integration with a bunch of stuff that on paper sounds like a good idea, but in reality is annoying as hell and hard to navigate.

> how effective is this data in terms of actually improving their profitability?

I work closely with telemetry data coming back from Windows devices. There's surely a line from collecting this telemetry to an answer to your question, but I've never heard anyone in our group talk about the profitability of this data. In fact, it costs a lot of money to process and store such huge data, which is absolutely a concern -- so there's strong belief (which I share) that this cost of business is worth it because it lets us make what I hope are good decisions about where we should focus our efforts.

The telemetry that comes back encompasses many dimensions of the user experience, and it helps us do things like figure out when an insider feature is ready to GA; what drivers or devices do poorly and how we can improve the experience; how accurately we can predict battery life; etc.

> they can optimize their dark UI patterns to be better at tricking people into setting Edge as their default browser?

I don't think whoever is in charge of 'tricking' people into using Edge actually have the kind of telemetry that most people think about with respect to Windows data collection. A lot of people think there's some sinister plot here -- apparently as evidenced by your phrasing. But if you look at the UI for default programs, it's really just a slightly annoying "are you sure?"-kind of message. (I don't use Edge by default, but I just reset my prefs to Edge then back, and I wasn't harassed again, so this isn't a persistent nagging.)

> how accurately we can predict battery life

You do remember that Windows nowadays only shows "Battery xy %" instead of giving an estimated runtime, right? (Which is, btw., a "holy fuck what braindead idiot thought removing this is a good idea" moment, at first I thought Windows wasn't giving an estimate only in the first few minutes, but after a while I started googling around thinking that the lack of estimated runtime was a bug, but no, someone actually went in and removed it. A very clear and obvious example of actively, explicitly and intentionally user-hostile UX for what sure seems like no other reason than "we could, so we did".)

Apple did the same thing. Big Sur will only show you the percentage. Not the estimated runtime. You can find the estimated runtime by looking at the energy tab in Activity Monitor. There are also 3rd party tools that will show it. But by default, it's not readily visible.

Apple claims they did it because the runtime estimate wasn't accurate enough. Maybe Microsoft's reason was the same. Though it does seem user hostile to remove it, unless the estimate was substantially off and there's no way to fix it.

> Though it does seem user hostile to remove it, unless the estimate was substantially off and there's no way to fix it.

I think the problem is that users expect percent charge to be a direct, linear correlate of time remaining, but battery discharge isn't linear. Even if it were, predicting time left is tough. Giving the user a percentage rather than a (likely incorrect) time estimate seems like the less bad approach.

idk I always found it to be accurate enough. Also it was a helpful canary to see that if the estimated time remaining dropped quickly, it probably meant I was doing something power-intensive, which maybe I would want to do something about.

It seems like it would be trivial to change the default and leave the time estimate as an option.

Battery discharge is quite linear, because the OS is using the information from the BMS chip in the battery pack, which tracks the energy stored in it rather accurately, and also tells you very accurately how much power is drawn or put in.

Personally I don't use Windows on any of my laptops, so I only noticed this one day on my work laptop. On my Linux laptops my status bar is literally just doing a weighted average of "time_remaining = energy_remaining / current_power", which works rather nicely.

That a supposedly top-notch software company like Microsoft is trying to gaslight their users into "predicting battery life is like hard you know! We're gonna remove it because it's so inaccurate, it's basically unsolvable aha" makes me throw up a little, but is also such an accurate and damning reflection of the way these companies conduct themselves.

Maybe a better explanation for why they would remove this is because they don't want to make battery degradation visible to the user. I.e. my battery used to get 4 hours of life, now it gets 2 and a half.

Perhaps, but pretending that this doesn’t happen is gaslighting.

Apple did this concurrent with reports that their laptops had poor battery life. They were likely exaggerated but I don’t think the reason was exclusively that the number was misleading.

In particular, it was because reviewers—particularly Consumer Reports—noted that Apple's new set of laptops had inconsistent battery durations. They'd run their test and get one number of hours, then re-run the test and get something very different. They ultimately decided the lower number was more important, because if you can't rely on a laptop to provide a certain amount of working time, the actual amount doesn't matter.

Apple responding by hiding the estimate—even on their older computers, where it worked very well—struck me as utterly missing the point.

Can you give an example of a type of decision where you feel telemetry was valuable in terms of making the best choice?

Their telemetry pushed me away from buying Windows licences.

I used to pay hundred of dollars for Windows Pro licences for all my machines at home (desktop, self-built HTPC, self-built NAS etc). - After MS went overboard with telemetry, I switched to Linux on everything except my desktop.

Recently I set up two new machines (printer server and gaming server), in the past I would have bought two Windows licenses for those, but I didn't, and installed Linux instead.

That's more than AU $650 in missed sales for MS.[1]

I am aware that I am just one single data point with no statistical significance, but I can imagine I'm not the only person in the world who is bothered by this.

[1] One copy of Win 10 pro is currently sold for AU $339. https://www.microsoft.com/p/windows-10-pro/df77x4d43rkt

I've been using Windows 10 since it's initial release and have never had telemetry automatically re enabled. I check the relevant settings regularly (every few months). I have no idea what all people on HN are doing that causes this to happen so frequently to them but it's never been a real problem. I wish I knew what I was doing differently so I could make a blog post about it haha.

I mean, what the HN and tech in general (edit: not general - sorry) audience wants is a totally zero telemetry, even if said telemetry is literally the list of installed updates. I personally think that's genuinely weird to me, since that Microsoft's "basic" telemetry only sends critical data like how many times Windows crash (and actually has the data reduced from 2015's state).

Yeah, knowing how much telemetry has helped me improve the user experience in the internal apps I work on I really don't think it is that evil. I think people just wish it was easier to enable/disable and that Microsoft was more transparent about it.

I think that the possibility of abuse (which has happened), possibility of government requesting that data (also happened, plus people here tends to be not inclined to give their data to any government), and possibility of leakage (general leakage not happened with Microsoft, but there's state-level hacks) means that even telemetry is tainted with privacy issues.

As someone who has worked on user-facing software, the part which bothers me the most is the reversal of user settings. MS is really bad about this, and even broke a linux install on a separate partition on my machine by changing the UEFI settings after an update without asking me. If I shipped software which did this, I would feel like I had to turn in my developer card.

Frankly this, as a developer. I call them "showstoppers" and if there is a god of programming, they are the ultimate sin. Things like

- breaking a user's multiboot

- deleting user data

Even if these are not bugs and merely the result of bad UX (lack of warnings and security for the user).

Deletion of all user data upon upgrade was a common 'problem' with earlier versions of Windows (I remember having such problems into the XP era, and I hear it happened to people on W10). Breaking other installs by MS is a common theme too in dual boot setups (whereas Linux bootloaders are more likely to pick up the Win install and offer access to it).

I remember vividly the last time it happened to me. I had been dual-booting Linux for some time (on a personal laptop dedicated to work). It was my first Linux OS, I was a newbie (precisely learning bash, programming for servers, etc). After some update, Win10 broke my dual boot. I was livid. It took me the better part of a weekend to fix it and avoid loosing all my work of several months (a lot of it was in the form of system files all over the place, I had no idea how to retrieve all that without ad hoc commands etc, and I had no idea what e.g. chroot was at the time. I learned, the hard way, haha — btw thx to the Arch Wiki for that weekend. It was a life saver).

The very next thing when I got access back to my Linux OS was to backup all my data, format everything and reinstall Linux fresh.

And that was it.

I've never used Windows ever since as my primary driver, only for testing purposes in VMs. That was 2016 iirc.

There are such things as showstoppers that make you instantly drop a product, never to look back. Microsoft's Windows desktop team never learned that, I don't know how they still pretend to be for "professionals". You don't do that to people who make a living with their computer (i.e. the vast majority of businesses nowadays).

> "- breaking a user's multiboot"

I have tried mutiboot with different Linux distros too and had it broken too many times, eventually I got sick of this shit. Fixing the bootloader should be a simple process but requires you to sacrifice a goat.

Never multiboot again, if I need another OS it goes in a VM.

When I absolutely have to, I install another OS on a separate drive with it's own bootloader and select the one I want in BIOS. That is actually reliable and never breaks.

Haha. One goat. Fool! You need at least four.

> When I absolutely have to, I install another OS on a separate drive with it's own bootloader and select the one I want in BIOS. That is actually reliable and never breaks.

Will this effectively encapsulate the UEFI settings? I had a problem where Windows turned off the wifi module on my MoBo to "save power" and it messed up the wifi on Linux.

s/easier/possible/

It has become a meme to be "outraged" about companies collecting any data. People don't analyze if this collection impacts them or not. They react as if the company wants their first-born.

It doesn't matter if it impacts them or not. It's a violation of the social privacy standards that have existed for centuries.

Someone looking in the window and making an inventory of all my furniture doesn't impact me in any way. But it's still an invasion of privacy and wrong from the standpoint of common human decency. A concept that has become lost to a generation that's become acclimated to thinking this is normal. It is not.

> It doesn't matter if it impacts them or not. It's a violation of the social privacy standards that have existed for centuries.

I have no idea where you're getting this from. Governments regularly perform national census, and collect data from all kinds of institutions to inform their policy. Banks constantly analyze your bank accounts and activity. Utilities meter your use for both billing and their own planning.

This is not someone "looking in the window" this is two parties (you and the company providing some service or product to you) needing to exchange data so each can do their work properly.

Those "privacy standards" you cite that have existed for centuries... that's not a thing. It seems to be based on falsifying the history and present of how humans organize and govern systems out of ignorance. Data gathering IS the "normal". Of course it's been very limited and cumbersome before, now it's easier thanks to IT.

So if your only argument is a vague discomfort from this not being "normal"... you have no argument.

> This is not someone "looking in the window" this is two parties (you and the company providing some service or product to you) needing to exchange data so each can do their work properly.

The electric company probably needs to know usage for billing. Microsoft does not need any information about Windows users for the product to work. Even things like fetching updates doesn't need to leak information, although I'll grant that it's a little bit more work to avoid leaking any data.

> The electric company probably needs to know usage for billing. Microsoft does not need any information about Windows users for the product to work.

Honestly, it's amazing you can say this with a straight face? You don't think Microsoft needs to know which parts of its product crash or not, and are used or not, in order to deliver a well-working product?

I don't know if most of you here work with large-scale products, but Microsoft has no the luxury of going to every Windows user personally and asking them how's Windows and what you want more or and less of. That "conversation" happens through telemetry.

They somehow got by for years before telemetry was an option; the answer is mostly testing, focus groups, and user studies, along with intentionally-provided bug reports.

None of what you say replaces telemetry, it's an addition to it. And the question remains why we need to willfully ignore the Internet, and do things like in the 80s. What's the purpose of this idiocy?

> You don't think Microsoft needs to know which parts of its product crash or not, and are used or not, in order to deliver a well-working product? How about asking the user and testing before delivery of the product? MS outsourced the test to the users and enforces a feedback for their faults. Imagine doing the same 40 years ago. A company employees enters your home to gather data about faulty products. You do know how many private data is stored on PCs and smartphones? Do you know what people access this data?

Isn't that what the insider community was meant to be for? Getting cool new stuff in exchange for data and performance stats collection.

Just taking it from everyone without offering the ability to turn it off is very poor practice imo. I understand they want this data to improve stuff but I should have the ability to not contribute to it.

>This is not someone "looking in the window" this is two parties (you and the company providing some service or product to you) needing to exchange data so each can do their work properly. It's even worse. I can see the person looking through my window. I can close the blinds. At OS level I don't have that chance. I can't constantly monitor the outgoing traffic and I can't check the consequences of every update.

>Those "privacy standards" you cite that have existed for centuries... that's not a thing. It seems to be based on falsifying the history and present of how humans organize and govern systems out of ignorance. Data gathering IS the "normal". Ever thought that you and parent live in different countries with a different value for privacy? Even a "simple" thing like a census in Germany led to Constitutional Court ruling and the establishment of a new right, that of informational self-determination. And that was in the eighties and it was about a census in the form of a total survey to be carried out door-to-door by civil servants or agents of the public administration.

Now you private companies fathering more data more often than ever before. So it's worse, calling that "normal" is strange.

Trying to gather data about your product performance and use is "normal". The fact it's been always happening is also "normal". The fact you do more of it because of technology doesn't make it less normal. Is it abnormal you can shop for groceries 5 miles away with your car, because you wouldn't be able to by foot? What kind of a silly thought process is that.

Probably companies haven't been able to serve billions of individuals worldwide, so they had less data. Microsoft has more customers, and has more data to process.

No one here is describing WTF is this "data" harming them personally with. It's not personally identifiable, and it's extremely mundane. Surely Microsoft will destroy your life by having some stats on how often people in a given region play Solitaire.

So tell me how car or TV manufacturers gathered that data 40 years ago? It's neither normal nor was it always happening. It's an effect of the internet and was normally impossible before.

"The Audimeter (audience meter) was used from 1950 during the early days of television broadcasting. It attached to a television and recorded the channels viewed, onto a 16mm film cartridge that was mailed weekly to company headquarters in Evanston, Illinois, and used to generate the Nielsen Television Index. It was based on an earlier Audimeter that had been developed for the 1942 Nielsen Radio Index. Randomly selected "Nielsen families" homes were enticed to accept the Audimeter by including free TV repair service provided by TV Index reps, which was a valuable commodity when vacuum tube televisions predominated." - Wikipedia

In other words, for TV they paid a small portion of the audience to record and report their habits.

So they had to ask and pay. No problem if MS does the same. Currently they do neither.

Microsoft does ask. And it can easily adjust pricing up and then give you a discount for telemetry. But you won’t end up with more money as a result. So what did we achieve?

Consent.

These are two different things that only on surface seems to be similar. The censuses performed by governments happen every few years (should be 10 years for majority of countries, if I'm not mistaken), while telemetry gathered by software happens almost on daily basis if not all the time. Then, depending on place, census may include every adult citizen or just representative part of the population and at the same it might be compulsory for those who were selected or who are required to participate.

Telemetry involves everyone by default and it's hardly optional unless you decide to actively fight it - speaking about Microsoft Windows here. The choices offered by Microsoft doesn't offer users true control over data they gather and process; hell, that actually applies to other big companies as well - you have just the illusion you are in control of what they managed to collect about you and the vague assurances, promises that they aren't up to no good with that stuff.

Government will most likely benefit from planning its politics having the data gathered from census while Microsoft might or might not use the telemetry data and honestly, seeing what they did with Windows 10 over course of all its releases, it doesn't appear they do actually use that telemetry data for anything. They don't even bother with users, power users opinions regarding the features that are causing issues in Windows.

>audience wants is a totally zero telemetry

does that audience lacks of experience with developing software?

How do you check that telemetry is disabled? Do you inspect packets that come out of your network card?

Well kinda yeah... MS is one of the biggest contributors to blocked requests on my pihole.

AFAIK it's tied to feature updates. If you have them disabled or use LTSC you basically never get settings randomly changed behind your back.

Like parent, I run a mix of Pro and Home machines with feature updates enabled, and my privacy/telemetry settings have not been changed back mysteriously. So I too wonder what I'm doing differently.

This is why I treat windows as my gaming console at this point. I game on it, thats it. Everything important I leave to linux/mac.

Wine/Proton is so good that you might find you don't even need it for that.

Games consistently get fewer FPS on my setup. 3750H GTX1650. Part of it is nvidia drivers being poopoo, most of it being CPU cycles overhead.

In my experience the nvidia drivers have been getting much better recently.

Proton's not perfect, but it's been mostly worth it for me not to have to boot windows.

Agreed. For anyone interested in FPS boost, try glorious eggroll proton.

I personally compile it myself off the AUR with the -O3 -march=native flags (look at the PKGBUILD for more details), and I get fewer FPS than Windows, yet more consistently than Wine. No microstuttering.

My experience is that even running Among Us (i.e. a very simple game) via Steam/Proton on Ubuntu LTS (i.e. the most common platform) was quite fiddly. Didn't have chance to try much else, Factorio is luckily native

For most things absolutely- I’ve had great success with proton. There are a few holdouts, especially in the VR space where Windows is still unfortunately very necessary.

Not even close.

> and has been collecting troves of personal data through Windows 10

How much PII is really collected via W10 telemetry? Or is the amount of times you click the start menu, accidentally search for something via Bing, then close that window now considered personal information?

It doesn't matter if it's PII or not. You ask before collecting. That's basic decency.

Not your computer. Not your data.

Not your computer. Not your data.

Unfortunately, Microsoft thinks otherwise. See the huge discussion here for example:

https://news.ycombinator.com/item?id=27629350

(tl;dr: thanks to Secure Boot, Microsoft had to give permission for Linux to boot.)

Also, don't forget the whole "Windows as a service" crap.

Your behaviour is personal to you and it is possible to identify a person based on that data.

For me it's a question if whether it is ever linked or able to be linked to specific app usage or details of that usage (thing Edge browsing behavior).

With industry privacy changes, having a persistent login at the OS level let's you do an end run around privacy protections in a browser, or incognito mode type stuff potentially.

Well, if you're using Windows you're stuck with Microsoft anyway. It's preferable not to add Amazon to the mix.

I'm not sure how the terms of the integration go, this could be avoidable - and it would be awesome if it was possible to port F-Droid to Windows.

The telemetry for Windows is a frequent topic here, could anyone clarify/compare the story for Mac OS?

In the Apple ecosystem you are asked once when setting up a new device if you would like to enable telemetry.

Apple devices still phone home constantly with things that amount to telemetry when you use them, even with analytics disabled.

Opening the App Store app, for example, sends your hardware serial number. Macs and iOS both maintain hardware-serial-linked 24/7 connections to Apple's push servers, too.

There aren't opt out UI settings available for the majority of Apple's phone-home.

The situation is a little better compared to Windows, but it's not perfect.

I always disable telemetry, but iphonesubmissions.apple.com:443, radarsubmissions.apple.com:443 and securemetrics.apple.com:443 are still contacted.

People messing with ocsp.apple.com:80 (https://github.com/StevenBlack/hosts/issues/1460) also need to be aware of ocsp2.apple.com:443.

The iAdSDK setting is somewhat respected, but all the Siri subdomains are pinged even if Siri is turned off.

I’ve never had to try and remove ads from macOS to the extent that I do with windows 10, it’s not devoid of advertising but not quite as bad

Moving the goalposts. The parent was asking about telemetry specifically.

You get ads on Windows 10? I've never seen a third party ad. They have had an ad for Office which was a live tile (which I have disabled because I do not find them useful.

Also not me, but American friends have always complained about this so I presume that they don't want to anger some regulators in countries where they don't put ads.

They were in the start menu. Ads for software and games. There was that candy crush saga game for ages.

Indeed even here in Europe. Cleaning the start menu from all that crap is the first thing I always do. I empty out the entire right hand side, because if you do so it actually disappears! But many updates put things back there.

Ads are telemetry now? Is "telemetry" just a catch all for "anything software does that I don't like"?

I'm pretty sure computer ads have always been telemetry.

This is a unique definition of “telemetry” as it usually requires collection of data. Some ads use telemetry for better targeting, but that isn’t all ads and definitely wasn’t true when things like banner ads were first introduced.

Reply on news.ycombinator.com