unable to disable field trials
Reported by michaelx...@gmail.com, Nov 13 2013
UserAgent: Mozilla/5.0 (X11; Linux i686) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/29.0.1547.57 Safari/537.36 Steps to reproduce the problem: 1. open chromium 2. look for how to opt-out of google market research, despite using an open source browser What is the expected behavior? this browser should allow me to choose whether to participate in research What went wrong? there is no option to guarantee my browser against communicating valuable usage data back to google. Provide a field-trials-disabled version, and/or deobfuscate your build tools. Did this work before? No Chrome version: 29.0.1547.57 Channel: n/a OS Version: Debian 7.1 Flash Version: Shockwave Flash 11.2 r202 A browser advertised as having its google-affiliations removed, shouldn't still then communicate with that gestalt corporate entity.
Nov 13 2013,
Jan 6 2016,
jschuh.. you're an idiot
Jul 4 2016,
Michael has made a valid point and it is disconcerting that it was brushed aside by chromium devs without any explanation or discussion. I understand that Google funds Chromium but it should be easy for chromium devs to understand why it is a problem that a purportedly open source software sends usage report to a single company. Why not publish this data openly? Any way, to answer Michael, I think there is a way to workaround these "involuntary" field trials. At least on Mac OSX, with Chrome.app version 44.0.2403.157 there is a separate sub-application called "Google Chrome Helper.app" which received "--force-fieldtrials" flags from the main Chrome application. I replaced the binary in this app with a shell script that strips the "force-fieldtrials" flags and calls the original sub-application that has been renamed. /Applications/Chrome.app/Contents/Versions/44.0.2403.157/Google\ Chrome\ Helper.app/Contents/MacOS/Google\ Chrome\ Helper.bin $( echo $@ | sed 's#--force-fieldtrials=[^ ]* ##g' ) I am not sure if this actually removes the field-trials in the main application but it seems like a step in the right direction.
Sep 13 2017,
Been about a year since last post but here is information on "Google's Forced Market Research" or maybe after reading this "issue" sounds like we are in "Chem-trail" territory here. FYI, Google does Information Systems and Network Architecture R&D. Market Research would indicate that they need to collect statistical data and then spend the time and energy to make statistical observations on it. Makes no sense when they can just ask us over at Amazon for the Market Trends. :D The Google Chrome Helper App is used as a wrapper (of sorts) to expose chrome apps/extensions to the OS environment settings/API and visa-versa. This gives chrome apps and extensions the ability to do things like desktop notification and matching system-wide settings like fonts and colors. It is done this way to expose the extension while putting it in a controlled-little-box to keep would-be malicious code from executing outside the IMO-robust sandbox chrome has. I'm unsure of the nomenclature of --force-fieldtrials, my guess would be back to when they were exposing this feature for extensions to go beyond the browser there was a flag #force-fieldtrials that would enable the then "experimental" feature. And when it became part of the functionality of the software it was never changed to something less knee-jerk-foil-hat sounding. Similar to the current: #enable-spelling-feedback-field-trial. If memory serves me the extensions and their functionality were all wrapped into one flag, not spread out like the current GPU stuff. A little look-see into any "how to make a chrome app" tutorial and every app and extension project starts out with a "background" page that is used as a resource for the OS to assign a Handle/ID and will keep the app "live" until the exit function is called so that the app or extension is exposed to events and properties. Of course if someone was that concerned about the processes why not just END or KILL them with chrome running and see how it responds. I for example had 8 listed. When I ENDED them using a system monitor I got notifications that 7 (all) of my extensions had crashed. The 8th one was for enabling auto offline reload as all of my open pages in chrome unloaded as well and required a manual reset of one page to re-enable the auto-reload functionality. Lastly (quoted from original post) |look for how to opt-out of google market research, despite using an open |source browser Open Source doesn't mean GNU-Free. I code and re-code for a S-ton of servers that run the Linux Kernel and no GNU but lots of $$$ (my paychecks) in http/smtp/sms/ect... servers. Several hundred-thousand js libraries that you won't see included with the next SSL update for data security or SQL updates for optimized. In fact the amount of Open source software that include "required" packages that are named in such a way as to look "official" that report on your usage are staggering. Heck even DuckDuck... uh wait they send all my traffic off US networks to an "overseas" hop and then back circumventing all wire-tap (criminal/government use) and wire-fraud (commercial use) laws of all my traffic, and "obfuscate" ping requests and packet headers so you can't see that your route is a proxy-proxy-proxy screen. Google "underhandedly obtaining" Data on what sites you visit isn't even worth worry. They know 80% of where you go on the internet since that's about how much of the infrastructure they provide. Sorry about the long post, but the "Google is doing this..." not followed by "... and advancing data technology so fast they are catching up to futurists' dreams. Oh and they don't charge me a cent to use it." is sounding like the old M$ = bad ($1000 pc) Apple = godly ($6000 pc)
Sep 13 2017,
@charlies : Nobody is denying that google provides very important services to users. But that doesn't mean that people should not call out a infraction (perceived or real). In any case, it's not like all the data collection that google does is necessary to have a functioning web browser. Check out https://github.com/Eloston/ungoogled-chromium Also before you dismiss people who report such things as conspiracy theorists you could look at e.g. https://www.theguardian.com/technology/2015/jun/23/google-eavesdropping-tool-installed-computers-without-permission
Nov 2 2017,
It is 2017 and Google is still conducting these field trials. Right now, I can only open a few tabs on my system because of memory usage I have observed by both Google Chrome and Chromium browsers. On my work desktop, I uninstalled Chromium because of these silly field trials consuming memory to the point I can't *work*. I still use both Google Chrome and Chromium on laptops but this is getting out of hand. I practically can't run any other program when either of those two are running because their field trials consume too much memory. So whatever the use of this market research for Google maybe or how useful they are to the wider community, please give us the option to opt-out of this market research so we can do something else beside reading our mails. I installed Ubuntu 16.04 LTS for a friend recently and then told him not to run Chromium because it will make his system unusable because of this. But since we all prefer Chrom(e|ium) to other browsers, here we are still clinging to this little monster just to read our mails. I hope you folks look into this and give us the option to turn this off from the UI not through the command line.
Nov 8 2017,
The problem I see here is that field trials make systems, especially my development system, highly unstable. Whether they "report data back" or not, there should be a ways to disable and limit this feature and its negative impacts, which include CPU and memory hogging (literally today, trials made it take 20 minutes to *open Chrome*, while killing processes from htop). This would also seem to be a liability for Google; something needs to be done here.
Nov 8 2017,
I want to share my PoV as a Chromium committer. I'm (actually) working on a UI change which will be guarded by this system. It has enough new code that I want a "kill switch" to disable it without rolling out an update. Having a kill switch also comes with some risk (maybe I broke the old path too). I could do the launch over time: 1. Feature complete but turned off to test the kill switch. (Wait for bug reports.) 2. Feature turned on. (Wait for bug reports.) 2.5(?). Feature turned off/on multiple times for all users as bugs are fixed. 3. Old code and switch removed. Field trials are a way to put the world in a superposition of "I launched the feature" and "I didn't launch the feature", then migrate everyone to one version of the world or the other by moving a slider. This isn't a way to experiment on *some* users, and I don't know what opting out would even do because there's a period of time when the feature is on for all users via the field trial system and the old code hasn't been removed yet. I also have an idea for a change that disables a visual effect in Chrome's UI to save power. I don't know how much power it will save in the real world, but it's a reasonable change to make for everyone. Without the field trial system, I would: 1. Add power monitoring to the next version of Chrome, wait for data. 2. Turn the feature off for everybody in the following version, wait for data. 3. Either turn the feature back on, or leave it off. This would take a long time and every user would see at least one switch. Field trials let me create a superposition where users have monitoring and the current behavior, and also have have monitoring and the new behavior. Everyone gets put into one world or the other (I don't know which one it'll be in advance) as soon as it's clear if the power impact is significant. It's not well-defined what opting out should do in this case, either. This is a way to make probabilistic config changes for all of Chrome, not a way to experiment on some users and not others. (It's worth reiterating that sending data back to Google is separate and can be switched off in Settings. I think everyone is asked to choose during installation, too.)
Nov 8 2017,
Re. c#7: Could you file a separate bug with more information about your issue? It's likely that some of the field trials you see are already turned on for everyone, and if you're seeing performance or stability issues in any Chrome release, it's a bug (and not necessarily related to field trials).
Sign in to add a comment