Hacker News Reader

I am logged on to Google, and they know I want English results and an English interface, because I've gone through and done all the settings. And yet, occasionally something goes wonky and they start showing me the interface in Italian again.

I have no problems if their default is the local language - that's pretty much a no-brainer - but when I've told them in no uncertain terms what my own preferences are, they ought to respect that.

It randomly flicks in and out of German for me all the time (I live in Germany). I have ravaged the settings to change this, but to little effect. At least I am gradually getting better at German!

For me sometimes it switches to Arabic even if I set the settings. And here in north Africa most people are francophone and prefer French or English !

Same here. I travel a lot and use VPN to access the corporate network. So if I am in the Chinese office google goes Chinese (and it takes some random guessing to switch the language back to English) then in the hotel (via VPN) it promptly goes German.

One small improvement would be to write the name of the language in the original and not in whatever country's version. (e.g. English instead of Angol, etc).

For some reason I tend to get Google's authentication pages in Hungarian (at least I think that's what it is), despite being in Sweden with sv_SE/sv/en_US/en configured in the browser and Google's services set to use English.

Fun!

> Google is making the (accurate) calculation that over 99% of people coming from a Japanese IP address whose computer says "Accept-Language: en-gb;q=0.9, en;q=0.7" are actually Japanese speakers of Japanese

Google is making the (completely wrong) assumption that those people cannot type google.co.jp instead of google.com and make their own choices.

But at least in Japan they're probably right about the language. Google pulls the same crap in multilingual countries, and gets it wrong a lot of the time. This isn't Google being smart, this is Google being lazy and insensitive.

It asked her if she wanted Japanese and she clicked yes and it gave her Japanese. What crap do you think anyone is pulling here?

Also, if in Japan, I would be surprised to find google.com not to be in Japanese, and I would expect to have to perhaps make some kind of preparations for the fact that the majority of the world does not speak my language and that my electronic devices and services are not universal metatrons run by omnipresent deitys either.

Yes, but google.co.jp does just fine. The automatic country redirects are the main issue. I wouldn't go to Microsoft.com and expect it to be in Chinese any more than if I went to Baidu.cn and expected it to be in English. The country domain is the ideal way to handle regional preference, I don't need the software doing it for me. I'm not an idiot, stop treating me like one.

Yes, but you're going to run into the problem that patio11 pointed out. You are in a tiny minority and they don't want to inconvenience the vast majority just because your preference is 'better'.

Country domain is the worst way to handle language preference. We already have a perfectly valid way to handle language preference with Accept-Language. There is no need to register domain under every possible country domain you could localize. And countries do not equal languages. Bilingual countries would be screwed if country domains were the only option.

Why is it the worst way to handle language preference?

If you are a local you are probably very used to your country domain, it is probably on par with .com.

Now the Accept-Language is quite subtle and guess how that is set up in an internet café? Can you even change it? Will the next customer be happy about that? I'd be mighty impressed if more than 1% of the people that surfs the web has a clue about Accept-Language.

Also, if one asks for a local country domain that really should be the localized version, sometimes I actually want to use google for a language different other than my own.

If I for instance wanted to search in English on something that has a similar wording as Swedish I'd like to go to google.com in order to quickly get rid of the Swedish-bias (since that doesn't work I sometimes use google.co.uk instead).

>I'm not an idiot, stop treating me like one.

Many people are however, and mass media, which includes Google, however they may try to deny it, lives by the numbers, which means adapting to the lowest common denominator. Just think, as the Internet spreads, Google will evolve (devolve) closer and closer to broadcast TV!

If I was in China and looked at microsoft.com, I would think they were absolute idiots if it didn't default to Chinese. If you don't localise for all your major territories from your .com domain and force them to use a country domain instead, then you should shoot your marketing director immediately.

Google isn't being lazy & insensitive. I18N is hard.

> I18N is hard.

Understatement.

> Google is making the (completely wrong) assumption that those people cannot type google.co.jp instead of google.com and make their own choices.

You're making the completely unsupported assumption that the average internet user understands domain names. Like patio11 suggested, if a user from a Japanese IP is accessing google.com, it's likely that this address was pre-configured in some way: the search bar was set up like that, or the company sysadmin set up the home page like that. But even if the user manually typed "google.com", if they're in Japan it's still (way) more likely that they wanted a Japanese version rather than the English one.

Agreed. Talk to mundane users. You will find out that many of them go to Facebook by typing "www.facebook.com" into Google's search bar.

People don't know the difference between the google search bar and the address bar.

Come on, most users outside the US are perfectly used to a plethora of international TLD's. It's something they are quite familiar with.

Also, they either use the English version of the OS and browser, or the one in their language, and both come with the default correct preferred language settings.

The same goes for company systems. You don't give someone who can't handle English an English language system, and the idea of setting up a system in language X but pre-configuring that in some what that English becomes the preferred browser language is a really odd edge case.

Even if such scenarios are more common, it still doesn't excuse telling people that make explicit choices to go fuck themselves, especially in multi-lingual countries where such explicit choices are common.

Japan may be relatively homogenous, but most countries in Europe are either formally multilingual or have large immigrant populations. Many thousands of people are regularly being confronted with misplaced assumptions about their preferred language.

You can't assume language from geo-IP databases, it is the most unreliable and inappropriate method imaginable, and you are 100% sure to regularly piss people off.

What method is more reliable than guessing the language based on the IP address?

Ask the user, remember preference. Possibly Accept-Language (note - not locale, as locale != language). But, ask the user.

That's kind of orthogonal to how you guess the user's language. It is, in fact, what Google is doing.

Shouldn't US users then be using google.us, and uk users use google.co.uk and should google.com just redirect to the language of the location of the user?

Yes. But google.us acts the same as google.com when it comes to location redirects though (e.g., I have a Canadian IP and en-us;0.9 Accept-Language, it redirects me to google.ca).

>Google is making the (accurate) calculation that over 99% of people coming from a Japanese IP address whose computer says "Accept-Language: en-gb;q=0.9, en;q=0.7" are actually Japanese speakers of Japanese

From the article:

>The problem is compounded by the fact that most of the time, you have no way of doing anything about it. And when you can do something, it’s often buried in layers of menus, which –don’t forget– are all in Japanese, Bhutanese, or Martian…

I think you're missing a key point in the article.

This is only a problem for those who:

1. Are in the tiny minority of people in Japan who don't speak Japanese

AND

2. Chose to switch to Japanese when Google asked them if they'd prefer Google in Japanese or English

Google guessing that she might have been Japanese was not the major problem here. The ultimate problem was her incorrectly stating her language preference. It's hard to see what reasonable measures Google could have taken to inconvenience fewer people here.

Actually, it's a problem for anyone who speaks a language that is different from the official language of the country they are in, and don't have a Google account (a lot of people.) In any case, a persistent language menu on all Google properties that has all languages available would solve either and all other problems associated with language choice.

Most browsers set the default Accept-Language header correctly. If some IT department screwed it up, it's not the reason to ignore the standards. My HTTP request data should be a priority over my IP address location. And Accept-Language is just that mechanism to set a preferred language for the response.

In all my travels I had to deal with language issues from Google. But it goes beyond that, as the vast majority of i18n'ed sites favor IP location over Accept-Language. I find this trend saddening.

Yeah, if an IT department messed up, call the helpdesk and complain. The setting will either be fixed or support should be able to get you through the menus to correct the setting on your own.

I agree with you for logged out users, but these things happen even when you are logged in.

> This is a hack around censorship laws in a few places (most relevantly to the instant case, Germany, I believe). Google has a blogspot.jp which is a copy of .com which is a copy of .de which is a copy of... except that they enforce country-specific censorship laws. This was deemed an acceptable tradeoff versus getting all of blogspot blacklisted in e.g. Germany.

what event/law are you referring to? why blogspot should be censored in germany?

Germany has many laws which are different that e.g. the USA. In particular, in the US certain sentiments which are politically and socially radioactive are protected by the US Constitution and the laws, traditions, and expectations of the American polity. Germany is similar in some regards but there is one particular politically and socially radioactive statement you could make -- support for the Nazi Party -- which, owing to the laws, traditions, and expectations of the Germany polity, will result in the government coming down on you like a ton of bricks.

I know Germany and many other countries in Europe are more sensitive about promoting extreme viewpoints. My question was more about something specific related to censoring blogspot, and how the domain redirects are used to bypass that. I mean, how does it matter what domain is used to display the same content? OP says he's redirected to blogspot.jp - what problems does it solve for Google?

Presumably articles that need to be censored in Germany would be censored on blogspot.de and not on blogspot.com, so Germans are redirected.

> * how does it matter what domain is used to display the same content?*

It only displays the same content if that content is compliant for the local region.

Thus, in the Nazi example, all off blogspot get the posts except blogspot.de, who filter it out somehow.

That's something that Youtube doesn't do with adverts very well - I'm in the UK, I've not left the UK, my IP geolocates to the UK, but I often get US ads in Youtube. Advertising law is different and often the ads are not compliant for the UK.

All countries have different laws in this department, but moves to segment or restrict Internet content all kinda goes back to a French case against Yahoo a while back:

http://en.wikipedia.org/wiki/LICRA_v._Yahoo!

I have noticed this in referral logs. They do it for tons of countries (I would guess all based on how many I see). Does Google automatically remerge all that to treat it as one domain for pagerank (so links to those separate domains are just treated as if they linked to the .com page), duplicate content, etc.?

This is not entirely so. When I visited Japan, my browser prompted "We've detected you've moved to Japan, would you like to change..."

That to me seems wrong. No one learns a language in 7 hours.

But many people learn a language before going to a country, and some portion of those will want to use it because switching back and forth between languages can be hard. I don't think it's a bad question to ask, at all.

You could speak both and prefer to use the Japanese version when in Japan (to get better local results, for instance). The site has no way of knowing your particular situation, so they ask rather than defaulting to one or the other.

»This is a hack around censorship laws in a few places«

Wow, I had no idea. I thought it was some obnoxius kind of load-balancing or something like that :)