Field Trials in Chrome, how to randomize or force them

Google is well known for the experiments that it likes to conduct in regards to the company's products. The company usually tests new features and changes to its products by enabling them for a fraction of the product's user base.

You may remember that Google likes to make changes to search, for instance how many results are displayed or where certain elements are placed on the page, or on YouTube where new layouts are tested regularly. These A-B tests provide Google with information that it uses to determine whether a change should be made available for all of the product's users, or discarded instead.

The Chromium and Chrome developers call these experiments Field Trials. These Field Trials are the reason why your Chrome browser may behave in a different way than the one your friend has installed, even if both versions, languages, operating systems, and other parameters match.

One of the latest Field Trials is a new bookmark prompt that is displayed to some users of the browser. It appears automatically underneath the bookmark star as a notification message and reads: "Like this site? Click here to bookmark it!". The idea behind the feature is to raise awareness for the browser's bookmarking functionality, at least that's what I assume it is been designed for.

While that explains the differences between your version of Chrome and the Chrome version of other users of the browser, it does not really provide you with the means to understand what is really going on behind the scenes.

While you can use about:version to display all variations that are currently active in your browser - many are listed there believe me - it does not help you understand what each variation tests in the browser.

A search for the variation strings does not reveal any information that you can use to identify the experiments.  What you can do though is reset the variation flag so that your browser is assigned to other field trials or groups of the same field trial.

This can also be helpful to avoid the fingerprinting of your browser based on the variations and other data, as they are submitted to Google regularly when you run Chrome.

• --disable-field-trial-config -- Disables all field trial tests in fieldtrial_testing_config.json
• --fake-variations-channel -- Fakes the release channel of the browser for testing.
• --force-fieldtrials -- The option can be used to enforce certain field trials in Chrome. Requires knowledge of values.
• --force-fieldtrial-params -- The option can be used to force certain parameters but it is necessary to know valid values to use it.
• --force-variation-ids -- Enforces additional variation IDs.
• --reset-variation-state -- Forces a reset of all Chrome variation states.
• --variations-override-country -- Overrides the country used for evaluating variations.
• --variations-server-url -- Specify a custom variation server.

You can find information about experiments that are conducted right now here on this page. Note that not all commits detail experiments but many do.

Advertisement