can anyone provide insight into the business reasoning behind this? i really can't conceive of why you would want to supersede a user's exact, known language with a guess. sites are pretty difficult to use when you can't read anything. maybe there are technical issues for some sites, but Google search is the worse i've ever dealt with, and they def have some resources behind that.
Mostly it's because people don't know how to change that setting. Imagine walking up to a computer in a shared space (hotel lobby, library, etc.) and it's been configured to send out accept-language: <something you don't understand>.
Many of the people reading Hacker News will be able to find and change that setting. My mom never will. She'll just know she went to google.com, and saw Chinese.
If you're using a computer in a country, and websites seem to be showing you things in the language of that country, that's something you can probably understand. If you're using a computer and some websites insist on showing you some other language, you'll be confused.
It's true that most people leave the defaults. However, there may an easy solution for a subset of cases.
Do the browser defaults in any country include multiple language settings? It seems likely that in most countries, the default would be only one language. And if this is the case, then if multiple alternatives are present in the request headers, it's very likely the user or computer admin has deliberately changed it, and that in turn would mean that sites should respect the choices.
This might still be wrong in when the settings were made by someone other than the current user, or there are multiple languages default-configured, but it might be a step in the right direction.
> Imagine walking up to a computer in a shared space (hotel lobby, library, etc.) and it's been configured to send out accept-language: <something you don't understand>.
If web browsers could somehow figure out they were running either under a guest/public-use account, or in a kiosk mode, they could avoid sending an Accept-Language header at all. Then,
1. in cases where the header is sent, it would mean a lot more (and hopefully override both online-profile-stored and IP-detection-based answers);
and 2. in cases where the header isn't sent, using an answer from an online profile setting or IP-detection would no longer be against-standard.
While that excuses ignoring the Accept-Language header, it doesn't make sense for overriding the user's explicit configuration in their profile; you wouldn't expect that to be shared.
That, and I'd expect public computers to disallow that sort of configuration anyway, so it would be stuck at the default value, which should be sensible for the location it's set up in; it's not like it would move around...
