20 ms·
Sure. But I have a question: why? Why should we opt out of the telemetry? To me, this idea seems to not just be admitting defeat, it's ensuring defeat right fro
by TeMPOraL 5y ago
Sure. But I have a question: why? Why should we opt out of the telemetry? To me, this idea seems to not just be admitting defeat, it's ensuring defeat right from the start.
Telemetry should always be opt-in. Yes, that means vendors will get much less data. It's on them to deal with it.
On a related note, I wonder how long it takes until one of the vendors of popular CLI tools or desktop apps get fined for GDPR violation. I wonder how much of existing telemetry already crosses the "informed consent" requirement threshold. I'll definitely be filing a complaint if I find a tool that doesn't ask for it when, by law, it should.
- 542458 5y agoBecause massive chunks of the population will never turn it on not due to ideological commitment, but simply due to no knowledge that it exists. Furthermore, if we define “tracking” as broadly as “any crash reports and any checking for updates” that effectively means these features will not work, and the open-source maintainers will have a much harder time tracking down bugs and encouraging people to update to less buggy or more secure versions of their software. Why not simply fork or choose not to install code you don’t like, rather than forcing your beliefs about what does or does not constitute acceptable code on the developers?
- account42 5y agoYou are not entitled to crash reports - that applies to open source developers just as to anyone else. If you wan't crash reports, have some kind of wizard or command to submit them and point to that when a crash happens, but you must always gather informed consent before submitting that data.
- beaconstudios 5y agoI like how games typically handle this; if the application crashes a dialog appears asking if you want to send a crash report, and often tells you what's included in said report.
- genezeta 5y agoMany Windows applications used to do that. Just search for "crash report dialog" images and you'll find many of them. Firefox does it too. And it keeps a log of past crashes on about:crashes where you can decide which ones to submit.
- jeroenhd 5y agoMany versions of Android do the same (although the details may differ, that's the beauty and the curse of open source). After crashing a (on some devices, only after a few times), you get a popup that says something like "<app name> appears to stop unexpectedly, would you like to report this to the developer?" If you click on the details button, you can see almost everything outside of pure hexdumps of RAM. There's no need for automatic sample submission if you respect your users' privacy.
- marcosdumay 5y agoOn a more CLI oriented design, take a look at Debian's report-bug program. It's completely transparent, and still gathers enough information that most times a one-line description of the bug is enough for anybody to understand everything that happened on your system.
- elliekelly 5y ago> but simply due to no knowledge that it exists. If it’s so important to the project the devs can ask users, educate them, and receive informed consent. There are plenty of ways a dev can force a user’s attention for a few minutes to hear their “pitch” and if the users still don’t want to opt in after hearing the reasons perhaps the reasons aren’t nearly as compelling as the devs believe them to be.
- baliex 5y agoI agree wholeheartedly with the idea that it should be opt-in but both approaches are equally unenforceable. The inverse of what's suggested in the article would be: export DO_TRACK=0 Project owners that want to track you simply won't take any notice of these flags anyway.
- TeMPOraL 5y agoI agree. The approach I'd like to see is, standardizing on some kind of DO_TRACK for convenience [0], and then doubling down on legal enforcement of opt-in telemetry. Project owners should be incentivized to seek consent, by threat of legal action from data protection authorities - and then, standardizing on some sort of DO_TRACK flag would be a no-brainer for them. As it is, letting the industry standardize on a DNT opt-out is just making telemetry more established as a standard practice, making it harder to argue that it should, in fact, be opt-in. The problems we have with tracking on the web are in big part because it was an established practice before appropriate legislation against it was drafted. In the CLI space, we have an opportunity to nip it in the bud, because it's not - as of yet - standard practice for console tools to silently spy on you. -- [0] - And while we're at it, standardizing on a browser-provided consent UI, instead of each site providing its own, with its own dark patterns. It's the same idea.
- MereInterest 5y ago> Project owners should be incentivized to seek consent, by threat of legal action from data protection authorities Definitely agreed, and I'd want to have some form of strict liability for data breaches, based on what kind of information has been leaked. Currently, a company holding data about me (e.g. name, email address, phone number, credit history) causes a large amount of risk to me, but themselves carry no risk in case of a data breach. They are the ones who can decide to collect less information, keep shorter retention policies, or restrict access to prevent a breach, but they have no incentives to do so. Yes, this would be a complete up-ending of many business models, but if your business model relies on collecting data without collecting the associated risk, it's a business model that society shouldn't allow to exist.
- 5y ago
- yoz-y 5y agoI'll play the devil's advocate. Most people will shoot support e-mails at you which are more or less "app crashes". If you have not already encountered the problem you have to walk them through tedious debugging process. If you collect crash reports, you have probably already fixed the problem. For usage data, it allows developers to focus on features that matter and know which ones you can remove. For example I don't collect any data in my app, but it also means that I fear removing any features that are slowing me down, because I have no idea about how people use it. As for why sometimes this would be better as opt-out, well, on iOS crash reports are opt-in, and only about 20-30% of users have them enabled. That is fine for huge programs or ones with little surface.
- finnthehuman 5y ago>I'll play the devil's advocate. Your point is basically "surveillance data is useful". And, well, yes. There would be zero debate over surveillance if there were literally no desirable reasons to have it.
- PragmaticPulp 5y agoCalling it surveillance is misleading in most cases. The most recent crash reporting system I worked with was little more than a stack trace without any user data recorded at all. Not even IP addresses of the report. We didn’t care who was crashing and we didn’t collect any PII at all. It was a simple report about where in our code the crash occurred. It was very useful for fixing bugs. No surveillance or PII involved.
- tjoff 5y agoIf that was the norm then people would opt in. But the trust is gone now, bad actors have ruined it for everyone. And the solution to that is not to enable it by default.
- nlarew 5y agoMost people have no idea what "the norm" or "telemetry" even is so I highly doubt this would affect opt-in rates at all.
- BurningFrog 5y ago> Telemetry should always be opt-in. Yes, that means vendors will get much less data. It's on them to deal with it. This approach leads to fewer vendors.
- amelius 5y agoOr users getting paid for their data.
- jefftk 5y agoAre you imagining a world in which volunteer developers of open source CLI apps pay users for the privilege of receiving crash reports?
- TeMPOraL 5y agoHow? And even if, would that be a bad thing?
- BurningFrog 5y agoTelemetry data is valuable. Vendors will get less of it, and so there will be less value in being a vendor. Not sure why this is not obvious? > would that be a bad thing? Not saying it would be.
- GekkePrutser 5y agoNot if users get something in return. Like with the windows insider program. Opt in to get betas and provide automated feedback. If MS turned off telemetry for regular users they'd be doing a great job.
- nxpnsv 5y agoOPT-IN is so much better for users. But if analytics is helpful, then probably opt in means you get no data, unless you keep bugging people asking people to opt in - which would be horrible too
- TeMPOraL 5y ago> But if analytics is helpful, (...) [opt-in telemetry] would be horrible too I don't question that analytics can be helpful. I do question the degree to which it is, relative to other methods of gaining the same insights (such as better QA, user panels, surveying people, etc.). I also don't think it would be horrible too. Inconvenient, yes. But horrible? People used to ship working software before opt-out analytics became a thing.
- cbm-vic-20 5y agoHow can users be incentivized to opt-in, particularly with a free-to-use application? I can see a case for ad-supported software, a developer could reduce or eliminate ads for that user in exchange for telemetry data...
- TeMPOraL 5y agoAsk nicely. After all, you want them to do you a favor. If you're less into respect and more into manipulation, offer them a meaningless trinket. A sticker on the app's home screen saying "I helped", or something.
- zerkten 5y ago>> I wonder how long it takes until one of the vendors of popular CLI tools or desktop apps get fined for GDPR violation. How would GDPR help with anonymous data? Say you have a CLI that sends back the frequency of usage for all top level commands daily. If the user doesn't log into the tool, or that information isn't sent then the developer would have IP address. If they discard that, how would it land under the remit of GDPR? I'm curious because I think it's easy for small developers to try and jump on this bandwagon. The big companies will all have vetted their telemetry strategy with their legal teams and have compliance reviews in place, as well as people who will handle cleanup from data spills. Bob is less likely to have this for his popular CLI tool.
- TeMPOraL 5y ago> Say you have a CLI that sends back the frequency of usage for all top level commands daily. If the user doesn't log into the tool, or that information isn't sent then the developer would have IP address. If they discard that, how would it land under the remit of GDPR? I think it wouldn't, given proper handling of the IP address. Where I'd expect your Bob to land in trouble is in mishandling crash reporting, in particular wrt. logging. It's very common for log files to accidentally acquire passwords or PII, or potentially other secrets protected by different laws. To be safe here, you'd have to ensure no user-provided data, or data derived from user input, ever enters the log files - which may include things like IP addresses and hostnames, names of files stored on the machine, etc.
- X6S1x6Okd1st 5y agoIt totally depends on your personal views on what victory or defeat would mean in this space. Not everyone shares your views
- 015a 5y agoI don't believe its unreasonable for it to be opt-out. Building software is very hard, and even something as inane as "automatically report crashes to the developers so we can fix them quicker" or "tell us how many users are on each version so we can estimate the blast radius of some backward-incompatible change" would be categorized as tracking. Here's the problem: people are idiots. You can manage or visit any Github issues page for a major project for ten minutes and recognize that even our industry is not immune from this. People also, overwhelmingly, use the defaults. When presented with the option to turn on tracking, most people won't, despite the fact that for most developers, its a legitimate good which benefits the user over the long-term. You can say "well, if people want to be idiots, that's their right". Idiocy never remains in isolation. If they refuse to update the app, then update Windows and it stops working, Users don't throw their hands up and say "oh well that's my bad". They don't complain to Microsoft. They complain to AppDevs. That becomes a ticket, which is written far-too-often from the perspective of anger and hate. Its triaged by, usually, overworked volunteers. Telemetry is not all bad. There is no "ensuring defeat" right from the start, as if its some war. Most developers just want to deliver a working project; telemetry enables that. Giving users the ability to opt-out, maybe even fine-grained control over what kinds of telemetry is sent, is fantastic.
- TeMPOraL 5y ago> even something as inane as "automatically report crashes to the developers so we can fix them quicker" or "tell us how many users are on each version so we can estimate the blast radius of some backward-incompatible change" would be categorized as tracking. Devil is in the details. Unless you are very careful, even a basic crash report may leak PII (commonly, through careless logging). > Here's the problem: people are idiots. I know what you're referring to, but I have a similarly broad and fractally detailed counter-generalization for you: companies are abusive. Their consider individual customers as cattle, to be exploited at scale. They will lie, cheat and steal at every opportunity, skirting close to the boundaries of what's legal, and considering occasional breaches into outright fraud as costs of doing business. Yes, I know not all companies are like that - just like not all users are technically illiterate. But the general trend is obvious in both cases. What this means is, I don't trust software companies. If a company asks me to opt into telemetry, with only a generic "help improve experience" blurb, I'm obviously going to say no. It would be stupid to agree; "help us improve the experience" is the single most cliché line of bullshit in the software industry. There's hardly a week without a story of "some well-known company selling data to advertisers". Introduction of GDPR revealed the true colors of the industry - behind each consent popup with more than two switches there is an abusive, user-hostile company feeding data to a network of their abusive partners. So sorry, you have to do better than tell me it's in the long-term benefit of the users - because every scoundrel says that too, and I have no way of telling you and them apart. And now for the fractal details part: > Idiocy never remains in isolation. If they refuse to update the app, then update Windows and it stops working, Users don't throw their hands up and say "oh well that's my bad". They don't complain to Microsoft. They complain to AppDevs. Yes, idiocy on both ends. There's a reason why users refuse to update the app. It's because developers mix security patches, bugfixes, performance improvements, and "feature" updates in the same update stream - with the latter often being a downgrade from the POV of the user. I'm one of those people who keep auto-update disabled, because I've been burned too many times. I update on my own schedule now, because I can't trust the developers not to permanently replace my application with a more bloated, less functional version. (Curiously, if usage telemetry is so useful, why software so often gets worse from version to version?) Secondly, if the user updates Windows and your app stops working, it's most likely your fault. Windows cares deeply about not breaking end-user software, historically it bent over backwards to maintain compatibility even with badly written software. It's entirely reasonable to expect software on Windows to remain working after Windows updates, or even after switching major Windows version. > Most developers just want to deliver a working project; telemetry enables that. Telemetry does not enable that. Plenty of developers delivered working projects before telemetry was a thing. What enables delivery of working projects is care and effort. Telemetry is just a small component of that, a feature that gives the team some data that would otherwise require more effort to collect. Data that's just as easy to lead you astray as it is to improve your product. > Giving users the ability to opt-out, maybe even fine-grained control over what kinds of telemetry is sent, is fantastic. Yes, and all that except making the telemetry opt-in is even more fantastic. You want the data? Ask for it, justify your reasons for it, and give people reasons to trust you - because the average software company is absolutely not trustworthy.
- s17n 5y agoOr maybe software "should" do whatever the creators of the software feel like making it do, and you should fork it if you don't like their decisions.
- pavon 5y agoMaybe contractors should take whatever they want from your house while they are doing their job. If you don't like it you should monitor them more closely, and opt-out, or do business with others. The software you write is yours. My data is not. You have every right to include or not include features, but you have no right to take my data without my permission. Your rights end where mine begin.
- yeahforsureman 5y agoIf you think GDPR always requires consent, you would be wrong. Consent is just one of many possible legal bases, and usually the one you use only when you can't use any of the rest. In this case, and more widely, it's not at all clear which types of personal data processing do or do not require consent.
- TeMPOraL 5y agoI know consent is only one of the possible bases. My comment was not about GDPR per se. I mentioned this law because of the spirit behind it - the law itself sets the bar very low (arguably below the point of what I'd consider ethical behavior). The other reason I mentioned it is because GDPR is currently the only stick we (at least, some of us) have to push back on the invasive telemetry. It's not nearly enough. GDPR notwithstanding, I'm of firm belief that any kind of telemetry in software should be strictly opt-in and require informed consent. I say should, it's an ethical view, not legal.
- xboxnolifes 5y agoI really do believe the best solution is a checkbox at install, but the checkbox starts filled. Yes, that's technically OPT-OUT, but it's extremely obvious and in front of you to anyone who actually cares about opting out.
- Teever 5y agoWhy do you believe that the best solution is opt out?
- dheera 5y agoLet's not forget that GDPR really only works for European companies serving European users. For pretty much anywhere else, GDPR is just a buzzword.
- zomglings 5y agoThe problem is a simple one. Most telemetry tools (Mixpanel, Sentry, etc.) don't give the developers who are adding them into their products the ability to quickly add respectful consent flows. This really needs to be a feature of the telemetry tools in the first place. Because, ultimately, most telemetry is being implemented by startup engineers who are burning the midnight oil to complete the telemetry JIRA ticket before going back to the long list of other stuff they have to implement. I have experienced this from all three sides - as a software engineer implementing telemetry, as a product manager consuming telemetry, and now as a founder who is building a tool to collect telemetry in the most respectful manner possible.
- TeMPOraL 5y ago> now as a founder who is building a tool to collect telemetry in the most respectful manner possible. Thank you for taking being respectful to users seriously. I'd be very interested in learning how your consent flows look, and what other aspects of your product are driven by the goal to "collect telemetry in the most respectful manner possible". I couldn't see much on it on the landing page, so if you have a moment, could you provide additional information, either here or in private?
- zomglings 5y agoHappy to show you some code. You cannot set up Bugout telemetry in your codebase without first defining your consent flow. We have a library of consent mechanisms that you can chain together like lego blocks to build these flows. For example, our Python consent library is here: https://github.com/bugout-dev/humbug/blob/main/python/humbug/consent.py https://github.com/bugout-dev/humbug/blob/main/python/humbug... Consent is calculated at the time that each reports are sent back. This means that your users can grant and revoke their consent on a per-report basis, which is the only respectful way to do things. We are also building programs which will deidentify reports on the client side, before any data is even sent back to our servers. This work is still in the early stages, but here's v0.0.1 of the Python stack trace deidentifier: https://www.kaggle.com/simiotic/python-tracebacks-redactor/edit https://www.kaggle.com/simiotic/python-tracebacks-redactor/e... Besides Python, we also support Javascript, Go, and we added Java support last week. I would really love to hear any feedback you have.
- kazinator 5y agoIf the variable is widely implemented, then that provides the nice, single point of control that the users need. In the FOSS world, we typically have distros between the applications and the users. If the applications honor the variable, then that's all the control that is required. A distro can implement an opt-in model by defining the variable with a value of 1 in the base system, so that it's present right from boot.
- TeMPOraL 5y agoIt's not a big enough of a change for people to actually choose Linux distributions over it - those who know about the variable will set it themselves, those who don't will be stuck with a bad default. My issue isn't with the point of control - it's with the default. Telemetry of all kinds should be opt-in. People shouldn't have to worry that they're constantly being watched. They shouldn't have to hope that every single telemetry stream is operated by competent and careful software engineers, guided by honest and law-abiding managers. You know how this industry works; it's a rare case where a data collection scheme doesn't overreach, accidentally ingest too much, leak data, or turn malicious and pass it to bad actors.
- kstrauser 5y agoYeah, this makes opting out my responsibility. If a company collects the names of all the files in my home directory, it's my fault for not setting some random variable correctly. Oh, and you did remember to also set it in your crontabs, right? If not, oopsie! You're gettin' spied on! This proposal is terrible and comes at the problem from the exact wrong direction. If someone wants to come up with a "export GO_AHEAD_AND_SPY=yes" envvar that enables telemetry, fine.
- 1vuio0pswjnm7 5y ago"If you want me to test your app(lication), pay me." - so-called "power user" What many of today's software authors want/expect is free testing. "To me, this seems to not just be admitting defeat, it's ensuring defeat right from the start." While I do not use any of the example programs the mentioned, it seems like these environmental variables would be appropriate if the user wants to toggle between tracking and no tracking. However, for users who would never want to enable tracking, "no tracking" should be a compile-time option. It would not suprise me if that is not even possible with these programs. How is the user supposed to verify that "Do Not Track" is being honoured.
- mradzikowski 5y ago> What many of today's software authors want/expect is free testing. Many of apps and tools are open-source and free. While I assume everyone wants to provide best experience, it's hard for me to justify being angry for bugs and problems in tools that I got for free, not bought them. Secondly, the industry realized that going fast, releasing often, measuring results, and improving over time is a winning strategy. No matter how often we as users will complain that "they changed something again", we still want to get things fast. Deploying new version once per year is not something we would really like in most cases. And fast development cycle inevitable comes with bugs, but they can be fixed quickly, not in the next year. Because even if you spend 2 months on testing your app, it will still contain bugs that will surface after the first real user touches the app.
- indymike 5y ago>Why should we opt out of the telemetry? Let's imagine the ls command with telemetry. What happens when you make an error like this? $ ls all-the-pr0n ls: cannot access 'all-the-pr0n': No such file or directory Um, what did I just tell the ls vendor? Who are they sharing that data with? > Telemetry should always be opt-in Opt in needs to be very precise and spell out exactly what is being shipped. For a lot of command line tools, telemetry is going to create more problems than it is worth. > I wonder how much of existing telemetry already crosses the "informed consent" requirement threshold. This is the question we all have to wonder about.
- trackno53324 5y agoRight on point. Also its at the mercy of the application and even then one there must be a trusty third party who can certify that an application follows the spec. This is not practical. The reliable and practical way is to have a ad blocker at the kernel level similar to browser ad blockers.
- lazysheepherd 5y agoI am installing Windows to a new machine right now. And here's the "Privacy Settings" setup step; Title: "Diagnostic Data" Explanation: "Send all Basic diagnostic data, along with info about the websites you browse and how you use apps and features, plus additional info about device health, device activity, and enhanced error reporting." And this is ON by default. "websites I browse"... ...Diagnostic Data?