This is not a new observation, but it’s something which just popped back onto my radar, sitting as I am in a hotel in Seoul, Korea.

Google really did introduce a horrible flaw when they first internationalised their site; one which hasn’t been corrected to this day.

The flaw is simple: they assume (seemingly by IP detection and nothing else) that the country you’re in is the language you speak, and that you will get a site localised in that language for as long as you’re surfing the web from there.

Whilst only mildly annoying when in, say, France, this is utterly disastrous for most western travelers to places like Korea, because we have no idea what the page is saying. Even worse, there’s no obvious way to navigate back to the English site, barring a small link on the site homepage (which you won’t see if you’re visiting the results page from a browser plugin; and is still bloody useless if you’re, say, German).

It’s always interesting when a company like Google - feted for their flawless execution, makes a schoolboy error like this, because it tends to reveal interesting things about that company’s culture.

In this case, it belies a certain arrogance, a “we know best” sort of attitude in forcing you down a path without really trying to work out what you might want. They don’t look at your site usage history (which they cookie to death, so there must be room somewhere for “is usually happy using the site in English), or your browser’s default language (less unreliable than it used to be), or… well, anything. “We know best, and we can’t be arsed to help you.”

As I say, there are some routes back to a site you can actually read - the link on the homepage or visiting the site via www.google.us (or .co.uk, or whatever - still no good for browser plugins). Somehow, these make it worse, though. Even a half-assed solution is an admission that there’s a problem somewhere; and if you know there’s a problem, why not… just fix it properly?

Yahoo! goes in completely the opposite direction with localisation, providing fairly “siloed” international sites, one for each territory they operate in, each with its own URL. This nicely sidesteps the issue of Google’s “forcing” a localisation on you without really knowing which one is suitable, and it allows Yahoo! to develop versions of a product which are truly geared towards the national community which they’re trying to serve, rather than seeming like an ivory-tower Silicon Valley linguistic concession to the rest of the world.

The major downsides are potential duplication of effort (is uk.answers.yahoo.com really different to answers.yahoo.com?), and the tying of language (or, more properly, localisation) to jurisdiction and other internationalisation concerns, which can marginalise sizable populations of foreign-language speakers who are living abroad. I believe this last point is only becoming more important as global mobility continues to increase.

These issues all interest me because I had to make my own decisions about them when designing the international model for Flickr, and (perhaps predictably), with Flickr we tried to take a more flexible “middle path”.

It’s not the kind of thing which should be readily apparent to visitors of the site, if I’ve got it right because it should “just work”.

On Flickr, we try to take into account your browser settings, where you are, where you came from (if you arrived from an external site) and various other factors to provide what I hope is a pretty accurate “best guess” at the correct language to serve you in (and, seperately, the best “place” to tie internationalisation concerns to).

As a fallback, every page (barring our really complicated javascripty maps and organizr features) has a set of links - in native language - to help you get to the language you want, if needed.

They may all be small markets, but ultimately I want to cater to the German living in Missouri, the Frenchman in Seoul and the Korean in São Paulo.

There may well be flaws in the approach as it stands today, but if so I hope they’re fixable (and I’ve not seen any negative feedback on language weirdness, so I assume we’re doing okay). Certainly, I hope that, unlike Google, our efforts aren’t a definite FAIL.

This article first arrived on the interwebnet on Tuesday, August 21st, 2007 at 4:57 pm .
It's arbitrarily categorised as being about: Geeking, Internationalisation.
Supernerds with RSS readers will probably wet their pants over the RSS 2.0 feed.
You can leave a response, or trackback from your own site.

There are thousands of articles on this very subject all over the internet. 39% of them are cleverly-disguised attempts to sell you little blue pills, 26% of them lead inevitably to a badly-researched, logically inconsistent Wikipedia article and 5% are written in an ancient Navajo dialect which is only spoken by 2 professors. The remaining 30% are, like this one, of very little interest to anyone besides their original author (and even he's a bit bored by it).