During the 1% stable experiment period, I don't think those sites adopted
Avail-Language.
For now, we can keep monitoring any significant changes over the
translation metrics and feature related metrics.
2.6% embedded iframes changing means they change content with different
Accept-Language, it includes sites they actually only care about the first
language.

Bests,
Victor

On Wed, Apr 2, 2025 at 12:08 PM Jeffrey Yasskin <jyass...@chromium.org>
wrote:

> On Wed, Apr 2, 2025 at 8:36 AM Victor Tan <victor...@chromium.org> wrote:
>
>> Jeffrey, Thanks for raising this concern. We understand that some popular 
>> multilingual
>> sites <https://en.wikipedia.org/wiki/Category:Multilingual_websites>,
>> based on our study, do indeed use the full language list in the HTTP
>> header. Your concerned use case, in other words, highlights a privacy issue
>> that this feature intends to mitigate. These multilingual sites typically
>> support major popular languages. However, for users whose primary language
>> is less common and therefore might not be supported by those sites, but
>> whose second or third languages are more widely used, exposing the full
>> language list in the HTTP header creates significant passive
>> fingerprinting. This means these individuals could be easily identified.
>>
>
> It's certainly good to reduce how identifiable these people are, but not
> at the cost of making them read content in a worse language. Have any of
> these sites adopted Avail-Language to regain their localization? Is there a
> way to validate the theory that when we break localization, people know to
> file bugs?
>
> Of your tests, I think manual language translation is a useful signal.
> Evidence that very few sites used by multilingual users respond to
> Accept-Language would also help, but I'm not sure the top 10k sites are
> likely to be the sites used by multilingual users. 2.6% of embedded iframes
> changing behavior seems like a negative signal, and Safari behavior,
> crashes, and latency don't seem very relevant. It'd be really nice to know
> what percent of page loads, by people who've configured multiple languages,
> change their HTTP response after this Chromium change.
>
> Jeffrey
>

-- 
You received this message because you are subscribed to the Google Groups 
"blink-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to blink-dev+unsubscr...@chromium.org.
To view this discussion visit 
https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAJh4P7Ezignr8f9b1Pv-6b%2BnWY9j_3tReRuE16B64gzjkFSZ3g%40mail.gmail.com.

Reply via email to