gnome-info-collect is a new tool that collects anonymous data about how GNOME systems are configured. It sends that information back to GNOME servers where it can be analyzed. The goal of this tool is to help improve GNOME, by providing data that can inform design decisions, influence where resources are invested, and generally help GNOME understand its users better.
The more people who provide data, the better! So, if you would like to help us improve GNOME, please consider installing and running gnome-info-collect on your system. It only takes a second.
As of last week, gnome-info-collect is ready to be used, and we are asking all GNOME users to install and run it!
How to run the tool
Simply install the package from Fedora Copr repository by running the following commands:
$ dnf copr enable vstanek/gnome-info-collect $ dnf install gnome-info-collect
The Copr repository also contains instructions on how to install on systems without dnf (useful for Silverblue users).
After installing, simply run
from the Terminal. The tool will show you what information will be shared and won’t upload anything until you give your consent.
There are packages for other distributions as well. See the project’s README for more information.
How it works
gnome-info-collect is a simple client-server application. The client can be run on any GNOME system. There, it collects various system data including:
- Hardware information, including the manufacturer and model
- Various system settings, including workspace configuration, and which sharing features are enabled
- Application information, such as which apps are installed, and which ones are favourited
- Which GNOME shell extensions are installed and enabled
You can find the full list of collected information in the gnome-info-collect README. The tool shows the data that will be collected prior to uploading and, if the user consents to the upload, is then securely sent to GNOME’s servers for processing.
The collected data is completely anonymous and will be used only for the purpose of enhancing usability and user experience of GNOME. No personal information, like usernames or email addresses, is recorded. Any potentially identifying information, such as the IP address of the sender and the precise time of receiving the data, is discarded on the server side. To prevent the same client from sending data multiple times, a salted hash of the machine ID and username is used.
All of this ensures that the collected data is confidential and untraceable.
Spread the word!
The best way to help is to take part by running gnome-info-collect and uploading your anonymous data.
You can also help by sharing this article with other GNOME users, and by encouraging others to run the collection tool themselves. The more users running gnome-info-collect, the better conclusions we can make from the collected data. This will result in an improved GNOME system that is more comfortable for its users.
So, do not hesitate to help improve GNOME. Simply install gnome-info-collect, run it and go tell all your GNOME friends about it! Thank you!
It returned “This tool must be run from a GNOME desktop.” although I am running Gnome.
Probably, you do not use GDM, which sets XDG_CURRENT_DESKTOP. Run
and try to run this tool again.
Nvm, I ran it with sudo because the $ in the article seemed to imply it must be run as root. Without sudo it worked fine.
The $ means to run it without sudo, # is used to indicate the need of root permissions
In that case there should be a # instead of a $ in front of the dnf commands.
I would say the DNF commands should be written as
instead, but point taken.
As a scientist, I know that collecting data hoping to find some “interesting effect”, you will find something. If you have a sufficiently large set of possible “effects”, then it becomes almost certain that you will “see” such an effect, even without actual cause, because well, there’s always a probbility that something just happens at a lot of instances. Sure, the individual probability of anything specific happening is low, but if you have very many things that might happen, you cross the line where it becomes probable that at least one of them “just happens”.
So, experiment design is important, especially in social sciences and behaviourial studies; else you will just do the deepdream thing and find dog snouts in noise, if that analogy still serves a purpose.
Where can I read what specific kinds of things this experiment is hoping to figure out?
How will the generated dataset be accessible, and to whom? How is it ensured that the data cannot be used to identify individual users, or even deanonymized? Are differential privace techniques employed, or some other anonymity-enhancing tools?
I see a very real risk: The fact that a GNOME user has, say, Firefox installed and libreoffice doesn’t identify them, but it’s also utterly unsurprising and wouldn’t need a survey of this kind. The fact that users who have a censorship circumvention tools, input methods for some specific language installed, and, say, Qucs and GNU Radio installed might lead to very targeted state operations at very specific university labs.
As a FOSS community, being utterly careful about the data we collect about users is kind of one of the reasons we advocate for FOSS tools and -methods: no dark channels in which your personal data disappears is one of our “selling points” after all. So, please excuse my eagerness to learn that the survey design is at the state of the art of privacy-preserving data collection, and not just “if you have something to hide, don’t opt-in”: These problems are borderline impossible to communicate to users, and honestly, this article does not even mention the bounds of what the data will be used for.
There is a nice discussion in the open-data issue of the project’s repository about data privacy (https://gitlab.gnome.org/vstanek/gnome-info-collect/-/issues/46)
The list of collected information together with the research questions, which we’re trying to answer, can be found in the projects README (https://gitlab.gnome.org/vstanek/gnome-info-collect/#client-collected-information), as stated in the article.
thanks for the reply! I saw that Readme, but it left more questions open than it answered. Under which conditions, for example, would the data suggest that GNOME needs to focus more on flatpaks? What kind of data distribution would suggest the focus on browser focus (that’s actually where the firefox & libreoffice example came from – two very standard pieces of software, default-shipped by many if not most new-user-distributions, and traditionally underintegrated in GNOME; I’ll wager it doesn’t take a study to find out firefox is there most of the time on a desktop linux system, often chromium, sometimes chrome, and very few people ever use konqueror, qutebrowser, torbrowser…) ?
Note I’m really not doubting the goals here – basically all the questions are good questions to ask, IMHO – but I’m doubting the utility of “gathering data first, defining metrics on which to decide later”.
Regarding the data / privacy / fingerprinting / deanonymization issue: Fully agreeing with Nathan Wills on the gitlab issue you link to – you can’t start a large data collection project, advertise it publicly, and then, while it’s running, say “but we haven’t discussed it or looked into it in detail”. That is not even compliant with European laws, where anyone organizing a collection of data about people (which this unambigously is) is liable to having to answer, upon request, precisely what is done with the data, who will get access to it, how your data is going to be used, and so on. https://gdpr.eu/what-is-gdpr/ , scroll down to “Data Protection Principles”, and especially “When you’re allowed to process data”: That requires “specific, unambiguous consent”; how can my consent be specific if GNOME has not even defined whether the data will be published at all, in aggregate or in individual form?
Interestingly, there’s now a reply in that Gitlab issue that says “Don’t worry, we did our due diligence regarding GDPR.” which is pretty curious considering what you said here.
What defines the machine ID?
Hi! The unique ID that is being sent with the data is a salted hash of machine ID (in /etc/machine-id) salted with your username and a constant string of “gnome-info-collect”.
It is not possible to get any of these information from the hash code.
You can find the code for generating this unique ID in the repository (https://gitlab.gnome.org/vstanek/gnome-info-collect/-/blob/master/client/client.py#L298)
And if you are more interested in the technical definition, here is the official man page: https://www.freedesktop.org/software/systemd/man/machine-id.html
Thank you! I didn’t know the “machine ID” was a thing… TIL 😉
I installed it, and ran it. I notice it set some of the apps as my favourites, but I didn’t select favourites so these would be whatever Gnome sets as default. Other than that, dat was collected and sent.
Hey! By the GNOME perspective, the applications that are in Dash are favourited. In your case, that might be the default ones.
Thank you for your participation!
Geoffrey Gordon Ashbrook
Question: Is this a one-time diagnostic, or an ongoing bug hunt? (e.g. Do I need to (re)run this daily?)
This is a one-time data collection excercise. We’re even trying to prevent users from sending the data multiple times. So once you’ve installed and run it, you can uninstall it with a feeling of a good deed completed.
You Really Don't Want to Know
Gnome does not allow me to shade windows. Every other window manager I’ve tried does. This decision is documented on the developer mailling lists about 7 or 8 years ago.
Ergo I will NEVER use Gnome again except for testing. And neither should you.
How does one run this on Silverblue?
It is pretty simple, however you need to install it as a layered package, due to the absence of a flatpak. Instructions are at the Copr repository:
Navigate to the copr repository using link above.
Download the appropriate repository for your distribution (Fedora)
Copy/move the downloaded repository to /etc/yum.repos.d/
Install using your package manager, for Silverblue
Reboot and run
Mr Leslie Satenstein, Montreal,Que
I have one request…
Separate logout as a separate additional entry from
A wish would be to include the day of the week in the top date presentation. I should not have to use the tweaks to add it. Do the reverse and let the tweaks remove the day-of-the week. (eg Wed is added). There may be some languages where the wish is not possible. There are none for Spanish, French, Portuguese,English
eg Wed Aug 31 12:09
Dive Mask Strap
I wish this gnome-info-collect app were rather an easy way of sending requests and suggestions to Gnome developers. Such as:
(1) Enable wide configuration of the date format in Nautilus. Many users have been complaining about this.
(2) Evince is woefully limited, especially with regards to document annotations. Take a look at KDE’s Okular.
(3) There should be a way of configuring the amount of dimming for non-active windows.
(4) An option for automatically turning off the mouse pointer when the keyboard in being used.
(5) LibreOffice is not affected by the dark theme in Appearance/Style.
gnome-info-collect is a simple data-collection program not intended for sending requests and suggestions.
However feedback is very appreciated! If you want your suggestions and requests to get to the developers, open an issue in the respective app’s repository. They mostly live on GNOME gitlab; here are some links that might be helpful for the things you mentioned:
Are you going to publish a summary of the results?
Definitely yes, once the currently running collection excercise is over.
No.. count me out. I have had enough of this from others etc, its why I am here. Even the mere suggestion of such activity make me suspicious.
How long will it be until gnome-info-collect is an integral part of Gnome, just like parental control, that cannot be deleted, without breaking Gnome.
I don’t want to be joy spoiler, but GNOME is not for me, I prefer (on Fedora 36) to use MATE, cleaner, simpler in my minimalist view.
Still, because GNOME and GTK+ are great libraries, I use (mostly from CLI) useful GNOME applications, so I am not against using Gnome, or KDE libraries for applications…
If one day, GNOME people make a fork, to something more simple, and less wannabe Windows, I may adopt it… again. (Gnome 2 was OK).
It also still surprises me that Gnome still does not feature useful widget available on XFCE, KDE, MATE desktops (Not LxQt)…
Oh, yes, I use a bit GNOME on a Alma Linux Server… and don’t like it, sorry.
Sorry, but past behavior in GNOME’s team leaves a feeling of being a little jaded and cynical on the matter of feedback.
Per personal opinion, it is felt that one could give GNOME exactly all the information the team wants, and they’d still balk from their crystal spires that togas are the only valid form of clothing.
Not to mention that this would involve actually engaging with the GNOME ecosystem when bluntly put, the desire to masticate a RedHat install CD is higher than having user agency removed from the equation. (When getting Window decorators is non-trivial, that’s bad.)
“Sorry, but past behavior in GNOME’s team leaves a feeling of being a little jaded and cynical on the matter of feedback.
Per personal opinion, it is felt that one could give GNOME exactly all the information the team wants, and they’d still balk from their crystal spires that togas are the only valid form of clothing.”
I agree entirely with this. Past behaviour and attitude of GNOME devs and previous one-sided “studies” that were massively biased that then informed design makes me think that this is a token gesture.
“…one could give GNOME exactly all the information the team wants, and they’d still balk ..”
^ exactly this
Renich Bon Ćirić
Done. Happy to contribute.
I somehow lost full functionality of Gnome some time ago, it seems after installing an update for the system. I restarted the computer, was taken to the log in screen to sign in. I entered my password, the computer appeared to be taking me to the desktop but simply took me back to the log in screen. It won’t let me in. I’ve had to revert to Gnome Classic to get to my desktop.
Feedback from current Gnome user (Ubuntu 20.04 -> 22.04.1) Let me help you with some Gnome feedback.
Does anyone use tray icon indicator? Answer: none (b/c Gnome doesn’t have this). Good we will continue to NOT implementing this feature.
Does anyone need the clock in a different position? Answer: No (b/c this setting doesn’t exist). Good, we will continue to make it non-configurable.
Does anyone need colors in theme? Answer: no (b/c libadwaita forces all black and white). Good we will continue with B/W.
Does anyone need to configure anything? No, b/c Gnome is dumbing down. Good we will go dumb for good. To compensate we will add data collection. To improve features nobody needs.
Those who praise Gnome clean flow, UI not getting in the way, etc. forgot they have installed 20 extensions to fill the missing features. Most of those no longer work. A system, claiming to be simple to the point of missing full of basic features (tray icons for example). Is ironically the heaviest in terms of memory fooprint. Are you kidding me?
Sadly I am still using that catastrophic Gnome. Waiting for some free time to migrate my entire dev env to a different Desktop Env.
Sadly I am still using that catastrophic Gnome. Waiting for some free time to migrate my entire dev env to a different Desktop Env.
I don’t exactly know how little free time you have, but it only takes a few seconds to install sway, geany, and…anything else that’s DE agnostic you’d need.
There’s not exactly a shortage of desktops on Fedora: https://fedoramagazine.org/fedoras-gaggle-of-desktops/
And switching though them is dead simple with something like the Terminal Based Session Manager. Today I’m in Windowmaker, yesterday I was in Cinnamon, before that NsCDE, and so on.
If you so dislike Gnome, maybe the best move is indeed to use another DE.
I only use 3 or 4 extensions, and I’m pretty happy with the result, and I do like the simple Gnome workflow. I do understand though that some people prefer KDE, and that’s perfectly fine.
Each UI decision in Gnome is often well explained, if you at least try to search why. Not that I agree with all these decisions (e.g. I indeed think Gnome should bring back a sys tray), but I do understand the rationale behind.
And having to install extensions is not an issue in itself. Such a central piece as a DE can’t satisfy everyone, and that’s what extensions are for. But Gnome could indeed maintain some official ones (such as a sys tray…).
Anyway… Gnome is a superbe piece of software. It has flaws, like any software. But overall, for me, it’s suits almost all my needs.
Done! Thanks for your work and investment in my free desktop environment that I’ve been using more than a decade!
I am wondering…and I am not trying to be controversial….but..why do they care now? they never did..
Indeed, and what will collecting system data do for them?
I’ve favorited BitchX, Xlock, and OpenTTD, what does that tell the Gnome Devs? That I like antique software?
What shell extensions are installed, what good will that do when the compatibility is broken every version regardless? People have told and mentioned their extensions for years on end. The download statistics alone are right there if they would bother to look!
Some of this seems redundant. GOA statistics should be provided by the data from the GOA, right? Shouldn’t multiple users be treated as individual? Do root/superuser accounts get allotted? Wheel groups?
Also, how does this get feedback from people who would go, “I would use Gnome, BUT…” and lapsed users?
And what does the system itself matter?
Do the system settings include modifications, hacks, and kludges, such as dconf tweaks and the Gnome Tweak Tool?
I sent it, 20 years ago Gnome was the best!
No defaults on this current ‘burner’ machine with some dconf editing. Just regular stuff, and maybe not so regular 🙂
On my ‘main’ machine, I don’t even use a DE 🙂
This might be the wrong place to address this, but … I’m a KDE user because my work Station has 4 monitors. I have tried to use Gnome with this setup, but the Gnome settings for multiple monitors are not feature rich like KDE. Am I missing something?
I try kde in last year or so. Switch to gnome because my two monitor setup have problems. Kde not remember last layout or scaling. Two monitor with different resolution on global scaling is nightmare … On gnome all working for me. When I unplug second monitor, first is scaled to scaling parameters that was used before. Reattach and all working as before. Maybe that was wayland/xorg related problem I do not know. Multi monitor setup working for me better in gnome than in kde.
How to improve gnome starts with gnome developers listening to the community, once that bug is fixed we can move forward. Meanwhile I’ll stick with Mate.
Perhaps this is a way to start making the connection with the GNOME developers you and others desire.
How? How is opting-in to send vague feedback supposed to be opening a communication line? If they wanted feedback, they could just ask a few surveys or ask openly somewhere.
Or perhaps they could openly state the feedback they’ve been getting since -checks- over a decade ago and make statements upon them, instead of trying to prognosticate the future of computing and making multiple incorrect assumptions and sticking with them.
I still shake my head dryly in annoying that the current GNOME is based on the wrong bet that the future was tablets & netbooks, when that was but a fleeting fad.
Something I don’t understand is why people hating GNOME so much are so willing to tell it on each and every comments.
I really like GNOME. I find it way better than KDE, to me, as it better fits my needs. Why don’t you just use KDE instead of harrassing the GNOME devs to tell them that GNOME should look more like KDE?
Don’t you think it is better to have very distinct DEs? It simply shows how much Linux can provide software for very different user profiles. When you compare to proprietary OSes, it’s a strength. Let’s embrace it instead of starting flamewars again and again.