Update on User-Agent Reduction plans

[Mike Taylor]

Hi there blink-dev and Chromium community,

I’d like to give an update on our current thinking and plans around User-Agent Reduction (née “User Agent Freezing”) and provide a space for constructive feedback on the plans. For those looking for concrete timelines or Chrome versions—we’re not quite there yet—but will update this thread when we feel more confident on timing. What we can say at this time is that no UA string changes are coming to the stable channel of Chrome in 2021.

The last major update we had on the topic was approximately a year ago, where we announced a pause for two reasons: 1) the desire to not create an inconvenient migration burden on the web ecosystem during the early days of the COVID pandemic, and 2) to allow us to ship the User-Agent Client Hints API (UA-CH), which is our proposed replacement for UA string-based content negotiation and detection. 

It took slightly longer than anticipated (due to unforeseen web compatibility issues), but UA-CH has been shipping and enabled by default since M89. We’re also starting the roll-out of both Client Hints Reliability mechanisms (Critical-CH & ACCEPT_CH) to address use cases where hints are needed on the first request and hope for those features to be enabled in Chrome by default in M91 (note that some server-side work may need to be done by sites to take advantage of ACCEPT_CH). We also have proposed some UA-CH API updates we would like to prototype and ship.

High-level Plan

We intend to proceed with UA Reduction as follows:

1. Beginning in M92, start sending deprecation/reduction notices for the navigator.userAgent, navigator.appVersion, and navigator.platform getters in the DevTools Issues tab.

2. In the coming weeks, we will send an Intent to Experiment to announce an Origin Trial (OT) for sites to opt into a Reduced UA String so that we can get feedback on the feasibility and compatibility of our desired end state. We expect to run an OT for at least 6 months to provide plenty of time for sites to opt in, test, and provide feedback.

3. Evaluate feedback from OT partners and the community, and based on this feedback proceed to Phases 3 through 7 of our plan (see next section for details), giving the ecosystem plenty of time to adapt in between them. Otherwise, reconsider the best course of action depending on feedback.

4. For sites with complex use cases that require more time for migration, we aim to offer the ability to extend the current User-Agent behavior for at least an additional 6 months (through a reverse Origin Trial).

Proposed changes

We think it's best to roll this out slowly and incrementally in 7 Phases, pending Origin Trial feedback. As mentioned previously, we hope to publish an update soon on proposed timing and milestones for each of these Phases:

Reduction Preparation

Phase 1: Warn about accessing navigator.userAgent, navigator.appVersion, and navigator.platform in DevTools

Phase 2: Launch an Origin Trial for sites to opt into the final reduced UA string for testing and feedback, for at least 6 months.

Reduction Rollout

Phase 3: Launch a reverse Origin Trial, for instances where a site may need more time for migration, for at least 6 months.

Phase 4: Ship reduced Chrome MINOR.BUILD.PATCH version (“0.0.0”). Once rolled-out, the reduced UA string would apply to all page loads on desktop and mobile OSes that do not opt into the reverse OT.

Phase 5: Begin roll-out of reduced Desktop UA string and navigator.platform. Once rolled-out, the reduced UA string would apply to all page loads on desktop OSes that do not opt into the reverse OT.

Phase 6: Begin roll-out of reduced Android Mobile (and Tablet) UA string and navigator.platform.  Once rolled-out, the reduced UA string would apply to all page loads on Android that do not opt into the reverse OT.

Reduction Completion

Phase 7: reverse OT ends and all page loads receive the reduced UA string.

See the reference page on chromium.org which details all the actual User-Agent/navigator.userAgent changes at each phase, as well as values for navigator.platform and navigator.appVersion.

Note: We have no plans to change the User-Agent string on Android WebView or Chrome for iOS at this time, but will make public updates if and when that changes.

We realize that these proposed changes have raised many valid questions and concerns, and we appreciate the feedback we’ve gotten thus far. Please keep the feedback coming! As we continue to discuss the proposal, we ask everyone to be mindful of the Chromium Code of Conduct.

Thanks,

Mike

[Mike Taylor]

Hi Jérémie,

Yes, this will be true (assuming `HTTP_USER_AGENT` is derived from the `User-Agent` request header field in whatever backend system you're working with).

thanks,
Mike

On 9/2/21 1:49 PM, Jérémie Le Pévédic wrote:

While reducing, can you confirm if the HTTP_USER_AGENT variable available on the backend and navigator.userAgent available in JS will remain consistent?

[Mike Taylor]

Hi Vladimir,

Can you help me understand why the JS API (e.g., `navigator.userAgentData.getHighEntropyValues(["uaFullVersion"])`) won't work in a web extension?

thanks,
Mike

On 9/9/21 4:48 PM, Vladimir Pastukhov wrote:

Hi, what about web extensions?

How the extension developers are supposed to implement version-specific behavior after this change? Not every feature can be easily detected by testing objects for properties. Also sometimes we need the browser's full version to work around bugs in particular builds. And obviously, extensions can't use UA-CH thing.

[Vladimir Pastukhov]

Hi, what about web extensions?

How the extension developers are supposed to implement version-specific behavior after this change? Not every feature can be easily detected by testing objects for properties. Also sometimes we need the browser's full version to work around bugs in particular builds. And obviously, extensions can't use UA-CH thing.

On Wednesday, May 19, 2021 at 4:27:48 PM UTC+5 mike...@chromium.org wrote:

[Vladimir Pastukhov]

Hi Mike

Sorry, missed this new method. Is there a possibility that the Promise will be rejected when invoked from the extension? Thanks.

[Mike Taylor]

I suppose there's always a possibility, but I'm not aware of anything that would cause it to reject today. As a fallback, you could parse the major version from `navigator.userAgentData.brands`, or just grab that info from `navigator.userAgent` (the major version will always be there, even after the UA is fully reduced).

[Mike Taylor]

Howdy blink-dev,

We just published an update to these plans over at https://blog.chromium.org/2021/09/user-agent-reduction-origin-trial-and-dates.html which contains estimated milestones for the phases we described in the original message. 

thanks,
Mike