Schneier on Security
A blog covering security and security technology.
« Phone Tapping in Greece |
| Friday Squid Blogging: Giant Octopus »
February 3, 2006
Here are the side-by-side search results for "tiananmen" on google.com and google.cn.
Posted on February 3, 2006 at 2:01 PM
• 45 Comments
To receive these entries once a month by e-mail, sign up for the Crypto-Gram Newsletter.
if you misspell it as "tianamen" on the google.cn site, you get lots of pictures of tanks.
I'm not sure this proves anything one way or another. Both web page's results appear to be sent directly to my browser; they could be tailoring the results to my IP address.
You forgot to link to the search people from China can use without Google.
The mispelling hole appears to be "fixed", I just tried tienanmen and .cn gives no results...
try "tiananmen massacre" in the link....google.com shows lots of pages while google.cn goes "????"
Interesting. After clicking around a bit, I learned that "According to the local laws, Google.cn always uses SafeSearch." (Very loose translation.)
Yes, the images.google.cn results for "tiananmen" are definitely censored. However, you really ought to work harder than that to confirm that conclusion.
In Chinese, "Tiananmen" only refers to a square in Beijing. The demonstration crackdown is always referred to as the "six-four incident" (because it occurred on June 4). Google could claim that ther .cn search results simply reflect the fact that the Chinese-language media don't associate "Tiananmen" with massacre. To draw a conclusion about the effects of censorship, you should compare a few more search results:
1) Search for "Tiananmen" in English on images.google.com (tanks everywhere):
2) Search for "Tiananmen" in English on .cn (looks rosy; censorship notice appears):
3) Search for "Tiananmen" in Simplified Chinese (the script used in Mainland China) on .com (mix of rosy and protest photos):
4) "Tiananmen" in Traditional Chinese (the script used in Hong Kong & Taiwan) on .com (similar mix of rosy and protest photos):
5) "Tiananmen" in Simplified Chinese on .cn (looks rosy again; censorship notice appears):
6) "six-four" in Chinese* on .com (protest photos):
7) "six-four" in Chinese on .com.hk (identical protest photos):
8) "six-four" in Chinese on .cn (random photos, with censorship notice):
The article mentions pair 1-2. The triplet 3-4-5 shows that repeating the experiment in Chinese makes the difference less dramatic, thus showing that at least some of the 1-2 might be plausibly attributable to language issues. However, when we use the proper Chinese-language search term in 6-7-8, we confirm the censorship effect marvelously. In fact, #8 basically says "I don't know what you're talking about!" -- a wonderful demonstration of censorship.
* The characters for "six-four" are identical in Simplified and Traditional Chinese.
For the Chinese government to meet its goal, google.cn must be the only Google that their citizens can access. It is straightforward for them to block access to google.com, google.co.uk, etc. But what is to stop people outside China setting up trivial Google(.com) proxies on their own servers? They could be restricted to users with Chinese IP addresses to keep the load down. And, like the Jihadi sites, they could change their own IP addresses and URLs frequently and these could be publicised by email and word of mouth within China. There are vast numbers of ethnic Chinese outside China who have degrees in Computer Science and access to servers.
Did you notice the disclaimer on the bottom of the .cn page?
I sent it through google translation and it said:
"According to the local law laws and regulations and the policy, partially searches the result does not demonstrate. You are not must look for: Tiananmen"
Original Chinses text:
So they are letting people know that the search result is a restricted set.
x, this is one situation where the Google translation leads to a misunderstanding. The last sentence actually means:
Did you mean to search for: Tiananmen
I changed the query to be:
that resulted in a slightly different set of pictures from google.com but returned no pictures (just some notice in chinese that I cannot read) from the google.cn site
It is unlikely that a Chinese person would use romanized search terms. If you search on the Simplified Chinese "天安门" (used in Mainland China), the results are well sanitized, but searching on the Traditional Chinese "天安門" (used in Hong Kong and Taiwan), google.cn will still show some very graphic links to www.boxun.com on the very first page, though many of the other images are missing.
Hitting google.cn from the States proves....what? I went to boxen in Shanghai and Hong Kong to which I have access (they use Chinese ISPs). Browsing to images.google.com and searching for "Tiananmen" and "tiananmen" both returned the tank/protest pics. Most of the links, when followed, returned 404s, but some did not.
Here's another good link:
"The authors are studying exclusions from search engine search results, and have found some 113 sites excluded, in whole or in part, from the French google.fr and German google.de compared with google.com."
and this one:
"There is one player that has been curiously absent from the climbdowns, shutdowns and other nefarious activities that the likes of Microsoft, Google and Yahoo have been engaging in in China. It is the US government.
China protests loudly, and rightly so, when minor textile products are excluded from entry into the US. Yet when the fastest growing product of all - information - is excluded from China by its government there is hardly a squeak from the White House."
Have you tried the same comparison restricting your search to site:.US and seeing how censored it is ? Or searching for images of the Vietnam war in Swahili ?
Google censors shit for China. However using stupid methodology in checking this does nothing to improve your understanding of it.
FYI "Tiananmen" is the entrance to the imperial palace as you will find out with a search on the very uncesored Wikipedia. It also features a distinct lack of tanks.
So it is actually a great example of a biased POV of western media. In China the site has thousands of years of history and national pride. In the west it has existed for only 20 and is perceived to be a source of shame.
Another telling find is if you compare search results for Falun Gong in english and chinese text. In English the chinese are doing the censoring, in Chinese the rest of the world is. Or in both cases it's just a stupid way to test something.
Search for 法輪功 = "Your search - %u6CD5%u8F2A%u529F - did not match any documents."
Same problem as the original. It is designed by someone who doesnt understand that Chinese have their own language.
In the side-by-side, SafeSearch is off for google.com but SafeSearch is on for google.cn. There doesn't seem to be a way to turn SafeSearch off. The explanation is (loosely translated): according to local laws, regulations and policy, basic SafeSearch is set.
The fact that google is facilitating any form of censorship, especially blatant censorship of any truth and Tiananmen is just an example, is a definite problem for me.
Then again, I live in Canada, and I can't really SAY that since it's not happening here, so I guess it's just business as usual?
Not sure if anyone already linked this, but a good side by side done at computerbytesman.com:
Tank, it's not exactly easy to convert English to kanji.
Interesting attempt at relativity, but I think you are way off-base with regard to human rights abuses and who should be allowed to contribute ideas to the collective memory.
You can say we all live in a different time zone, and thus should have different sense of what time it is, but that doesn't mean relativity makes it impossible to understand time universally.
Moreover, the nature of oppressive regimes is to manipulate or deny common consciousness. That's very different from a I/thou context. The Chinese authority is essentially saying Google is forbidden to use their normal algorithm, which so many have come to know and love as a useful reflection of consciousness, because it undermines a very particular political objective.
It's irritating that a discussion about how to probe Google's censorship is even necessary. Google justifies censorship within China by pointing to Chinese law, but how do they justify keeping us (outside of China) in the dark about which sites are censored?
google.com is still available now in China, But I believe that somedays later it will be cut off in China, just like VOA and BBC.
Also, the China's government is doing too much to rape it's people that we(i am a chinese citizen) have little freedom of speech.
based on local law, some of the results of the search are not allowed to display.
are u searching for "tiananmen"
@ Davi Ottenheimer at February 4, 2006 12:48 PM
> Interesting attempt at relativity, but I think you are way off-base with regard
> to human rights abuses and who should be allowed to contribute ideas to
> the collective memory.
How can I be off base if I haven't referred to it ?
> You can say we all live in a different time zone, and thus should have
> different sense of what time it is, but that doesn't mean relativity makes it
> impossible to understand time universally.
Was this your way of saying you'll be happy to provide accurate translations of all .ch search results ?
Or do you understand that the universal adoption of time and language standard differ ?
"Your search - site:.us "mai lai" - did not match any documents."
You think if we search for mai lai in Swahili that'll get better or worse ?
"It's irritating that a discussion about how to probe Google's censorship is even necessary. Google justifies censorship within China by pointing to Chinese law, but how do they justify keeping us (outside of China) in the dark about which sites are censored?"
This is a really good point.
I think you thoroughly misunderstood Tank (unless I have!)
Tank is not talking about cultural relativism, i.e. that it's OK for the Chinese government to censor because it's part of their culture or some such rubbish; he's simply saying that most of the proposed tests have given misleading results because of cultural misunderstandings. In particular:
* searching in latin text is likely to give very different results to searching in hanzi, and the great majority of Chinese users will be searching in hanzi, so the proposed tests in latin script are not especially useful; and
* quite apart from any censorship, searching for "Tiananmen" in hanzi or from *.cn will preferentially return Chinese language sites on Tiananmen. Due to Tiananmen having a much broader significance to Chinese than it does in the West, this will give more results unrelated to the massacre (ergo proportionally fewer references to the massacre) compared to searching for "Tiananmen" in latin script. This is expected even if there is no censorship at all so the proposed test does not give meaningful results.
Try the words "Fallon Gong".
The Chinese government hates them.
Exactly. Control of the search algorithm and results means control of the flow of information, whether it be insiders looking out or outsiders looking in...the Chinese have perfected the opposite of a google-bomb.
A related announcement:
"Microsoft says blogs or journals blocked inside one nation would remain readable outside that country. [...] The change in policy applies to weblogs or journals written on Microsoft's MSN Spaces service.
In June 2005 it was revealed that Chinese bloggers using MSN Spaces could see their entries being blocked if they mentioned banned words such as 'freedom', 'democracy' and 'demonstration'.
Brad Smith, Microsoft's senior lawyer, said it would now remove blog entries only if it gets a 'legally binding notice' from the government of that nation."
There was no mention of whether the words must be in any particular language to be considered relevant/meaningful to the nation in question.
Here's an odd twist to the Google ethic:
or maybe it should be
Nations that want to decieve citizens: 1
Companies that want to decieve users: 0
"Google confirmed that BMW.de had been removed from search results, adding that it would not tolerate any attempts to manipulate searches."
oops, make that "deceive". irony not intended.
Is this caused by censorship? or is it pagerank?
it's in google's interest to provide what the user wants when they search for tien an men, and to westerners who are the people who mostly influence global page rank, they have only ever heard of tien an men in connection with the chinese democracy movement.
But tien an men is one of the icons of chinese culture. To many chinese viewers, the images presented are what they associate with the place, I think.
This is like expecting all pictures of New York to feature kamakazis.
American culture has pressed itself onto the whole world, so there is no place where New York is only associated with suicide flights. But there are many places where tien an men is only associated with tanks.
So google.cn is targetting its market, more than applying censorship, in this case.
I don't mean to argue that there is no censorship in china, I have no idea how much there is. But there is censorship outside china too. Cultures differ vastly, so the censored pages differ vastly, and the methods of censorship differ vastly.
I am not an apologist for totalitarian censorship, don't get me wrong. I just think the western system does a very good job of directing media, and it does use totalitarian censorship just as heavily as China does, but most westerners agree wholeheartedly, or at least half-heartedly with its aims, so it is hardly even considered censorship.
The issue with China is, I think, over exaggerated for political reasons. Do not fall into the trap and become an anti-china flag-waver. China has improved its human rights consistently from the Manchu era onwards, while the west has gone backwards since the post-vietnam era.
Chinese government is a real feudality ...
& Chinese pepole all snooty Chinese government ...
if you misspell it as "tianamen" on the google.cn site, you get lots of pictures of tanks.
I can't find "tanks" ... ...
In China, all Web site have to apply certificate ... or else that will be disabled ...
I think many people have misunderstood chinese governent and chinese. it is not as terrible as what you imagine.
Schneier.com is a personal website. Opinions expressed are not necessarily those of Co3 Systems, Inc.