Bing accused of censoring Chinese language results outside China

[realmayo]

Thought this was an interesting twist on the whole search-engine censoring thing: "English and Chinese language queries for terms such as ‘Dalai Lama’ return radically different results" depending on if your input is in English or in Chinese. This is for people using Bing outside China.

The Guardian story is here: http://www.theguardian.com/technology/2014/feb/11/bing-censors-chinese-language-search-results

The Guardian being the Guardian, it doesn't bother explore the possibility that the differences could be down to search engine algorithms -- admittedly I have no idea how they work. But I'd have thought if a search engine is ordering its results by relevance and it uses popularity of websites as a proxy for relevance, the fact that the vast majority of people who use Chinese characters to search for the Dalai Lama will be inside mainland China and directed to baidu rather than wikipedia might have something to do with it? Then again the article says that Google's results don't differ much whatever language you use. 

[roddy]

That's an interesting point. An algorithm might quite easily conclude that Chinese people just don't like Facebook and Youtube, as they never visit. 

 

I suspect the internal reaction at Bing was "Huh? We have Chinese-language search results in the US? Oh yeah, I guess we do. Hmmm, going to need to fix that someday."

[Lu]

From the pictures the Guardian included, one would think that searching for Dalai Lama in Chinese brought English-language results. That's not the case, fortunately.

 

Anyway. I had to try this out of course and sure enough, lots of Baidu Baike, some CCTV, but no DL homepage. Interestingly, a 'related search' at the bottom of the page was [a certain square at a certain date] and that did yield Wikipedia, Youtube (!) and Boxun (!!).on the first page. Searching for [a certain religious organisation] brings the Epoch Times, Minghui (which is also FLG IIRC) and the FLG homepage.

 

Unless someone at Bing is scrambling to fix stuff, I think those censorship allegations can be laid to rest. Perhaps the Dalai Lama just doesn't have very good search engine optimisation (which I'm sure the FLG does have, and TAM is just something that more activists are interested in so it would optimise itself). And if anyone was monitoring my computer they have now added another page to my file.

 

Self-censored this post to avoid censorship, I hope the words left in it don't trigger anything.

[gato]

The DL's Twitter page showed up on the second page of results when I searched for DL in Chinese, so I don't think Bing is doing any censoring here.

Search result rankings are primarily affected by how many other pages that discusses a certain topic point to that page. If most websites that discuss the DL in Chinese are controlled by the Chinese government and refer to other websites controlled by the Chinese government, then it's no wonder the DL's websites rank pretty low among the results. But if you search for June 4, it should be just the opposite. Very few Chinese government websites talk about June 4, so most of the top search results should be dissident websites.

The Guardian's editorial standards seem a tad low these days. I thought it was just its "Comment is Free" section, but maybe not just there.

[Silent]

 

English and Chinese language queries for terms such as ‘Dalai Lama’ return radically different results" depending on if your input is in English or in Chinese. This is for people using Bing outside China.

I wouldn't expect anything different. If I search in English I expect websites in English to be returned. If I search in Chinese I expect websites in Chinese to be returned. These are off course different results. Actually I tend to get annoyed if I search in English and get Dutch results back which happens far too often lately with google. As such I'm very dissatisfied with 'personalised' search results. My location, previous searches, content of my gmail box etc have nothing to do with what I want to find. 

[m000gle]

There's certainly an air of hysteria in The Guardian reports, which is unnecessary; but I don't think I can be quite so naive, or so apologetic, pinning these radically different results on page ranks, algorithms and/or input language alone.  This is especially true given that Bing's search results clearly state, when searching for topics considered sensitive in the P.R.C., that:

 

"Some results have been removed"

 

This disclaimer only appears when searching for sensitive topics like “达赖喇嘛” or “天安门 1989”, and corresponds strongly with search results which appear to be censored; but does not appear when searching for "美丽中国", for example, nor does the disclaimer appear when these same searches are made in English.  This is no mistake.  These test searches were all made from overseas, using www.bing.com (not .cn), and subject to neither the Great Firewall nor Chinese internet regulations.

 

In a nutshell:

1. If Bing maintained either two Chinese language databases, or if it ran one which compensated for the selection bias of a huge market share being legally forbidden from accessing a significant subset of websites, then this wouldn't be happening overseas;
   
2. If Bing filtered results based on geographic locale, as its legally required to do and indicates as such on its disclaimer page, then this wouldn't be happening overseas;
   
3. If Bing maintains a single database, without any compensation for the selection bias introduced by P.R.C. regulations, then Bing is broken by design and is very much censoring the results overseas, even if inadvertently;
   
4. If Bing is intentionally applying the P.R.C. restrictions to its entire Chinese language search, then, well, maybe the click-bait Guardian headline is justified.
   

This post is meant neither as a critique of China's internet regulations, nor in support of any particular view on them.  I'm no fan, but have no problem living with them when required to; with China, it literally goes with the territory.

 

Objectively, however, it seems clear that Chinese language Bing search results are, indeed, being censored overseas; to what extent, and the reason(s) why, we can only speculate.

[gato]

I would go with your 3rd possibility, that Bing did not adjust for the bias among web pages caused by Chinese government censorship.  The fact that DL's Wikipedia is listed on the first page of search results and his Twitter page on the second page of results suggests that it's unlikely that intentional censorship is taking place.  

 

If you looked at the search results on Google for 达赖喇嘛, you'll see that the major difference between Google and Bing is that Google has given higher ranking to overseas Chinese language news sites like Voice of America Chinese, FT Chinese, and Radio Free Asia. Otherwise, there is no reason for overseas sites to rank ahead of mainland China news sites when you are only searching for the name 达赖喇嘛. Google made some manual adjustments.

 

Further evidence that Bing is not applying censorship: search for the Chinese phrase for "Tiananmen Massacre".  See attached screenshot.  All of the top-ranked pages are non-government sites showing the massacre.  

 

[realmayo]

It's a good spot by m00gle though: what explains the "Some results have been removed" at the bottom of the page?

 

Like m000gle, if I search 天安门 + sensitive_year I get the "Some results have been removed" message, but if I just search for 天安门 and no year, I don't.

 

Would be interesting if within China Gato's "Tiananmen massacre" in Chinese returns the "Some results have been removed" message. 

[Lu]

I also get 'some search results have been removed' when searching for Tian'anmen [certain date], but given that seven of the ten results on the first page refer to the bloodbath that happened there (and one or two other others perhaps too, this is just going by title), I have to wonder what results they removed. It can hardly have been on instigation of the Chinese government.

 

Searching Liu Xiaobo in jianti doesn't get a 'some results...' and only brings up Chinese sites on the first page. Searching in fanti gives me facebook, the Liberty Times (pro-independence Taiwanese newspaper), and a website called saveliuxiaobo, among others. Similar for Chen Guangcheng in jianti/fanti.

 

Searching for FLG doesn't get a 'some results...'. Either they're not censoring or they're censoring really, really badly. I'm inclined to go with the former. Half-assed coding, yes. Malice, no. Incidentally after this whole exercise I see why google is better and am appreciating it all the more. The information hoarding isn't nice, but the search results are great.

 

What I am curious about: is it possible to search for leaders' names in jianti in China on Bing?

[roddy]

Looks like Google is trying to steer me away from even investigating censorship....

[imron]

It's a conspiracy!

[realmayo]

They're saying it's a glitch:

http://www.bing.com/blogs/site_blogs/b/search/archive/2014/02/12/update.aspx

 

Would be interesting to compare resutls from HK/Taiwan with those in the US ... perhaps most people searching for this stuff in Chinese in the US would be mainlanders, and typically prefer mainstream mainland websites -- Bing learns this choice over time and therefore bumps results that lead to those sites up the list. People in HK/TW would be less likely to default to those websites so might appear to get a 'less-PRC' version of the web than Bing users in the US. But then if the algorithm distinguishes between simplified and traditional input then that difference might make any comparison between mainland and HK/TW messy.

[Lu]

Would be interesting to compare resutls from HK/Taiwan with those in the US ...

From what I've seen so far, I think they'll likely be the same, except if you search in fanti instead of jianti. I don't know what their algorithm looks like, but it appears to me that it doesn't differentiate by country and goes purely by search term. If searching in the US and the Netherlands takes one to mostly mainland sites, there doesn't seem to be much geographical sophistication. Searching things in fanti in the Netherlands brings me the Cantonese wikipedia alongside a Taiwanese newspaper, I don't imagine it'll be much different elsewhere.

[gato]

I just did a search for 共产党, it seems that Google gives much higher ranking to anti-Communist Chinese sites than Bing does. This is probably due a manual tweak by Google rather than the result of a politically neutral algorithm.  

If you search for "Communist" in English, you don't get anti-Communist sites among the top choices.

[MeiMay]

gato: Google is known to be extremely political and selective. It's THE search engine known for censorship, not just political but also financial. Certain companies can go from page 1 to page 234234 in a Google search with the same search terms overnight because Google added them to their "google censored" list.

Several other search engines censor results too, including Duckduckgo and Startpage. Wikipedia is censored a lot more than search engines. On the other hand baidu often finds me exactly what I'm looking for. After trying out all the different search engines I find it's not a matter of 'if' at all. It's a matter of choosing the search engine or engines that censor the results you're not interested in. It's probably an easy choice to make if you spend a little time seeing what results different search engines give you about different things.

[realmayo]

Google is known to be extremely political and selective. It's THE search engine known for censorship

What you write is untrue. To write that and then go on to write about Baidu! I don't know what to say. Too simple. Sometimes naive.

[Lu]

MeiMay, it's simply not true that Google is very censored. Do you have an example, some source?

Wikpedia is a polical free-for-all, it's pretty much the opposite of censored. Unless you mean that some countries don't allow access to Wikipedia, well that's true but that's hardly Wikipedia's fault.

Depending on what you're looking for, Baidu can be very useful. But it's heavily censored by the Chinese government, unlike Google which is not censored by any government. I have searched for Maoist things on Google and found it without a problem.

[MeiMay]

https://en.wikipedia.org/wiki/Censorship_by_Google

http://www.2600.com/googleblacklist/

http://www.webpronews.com/google-preventing-u-s-users-from-disabling-safesearch-2012-12

http://www.telegraph.co.uk/news/uknews/3130883/YouTube-censors-comedians-anti-Sharia-video-called-Welcome-to-Saudi-Britain.html

http://www.pcworld.com/article/id,131519-page,1/article.html

http://www.telegraph.co.uk/technology/google/9954990/Sweden-rows-with-Google-over-term-ungoogleable.html

http://www.chillingeffects.org/keyword.cgi?KeywordID=2

http://www.techspot.com/news/42163-google-starts-censoring-torrent-related-searches.html

That's about google's censorship BEYOND censoring everything the Chinese government asked Google to censor which they happily did, until very recently. If you object censoring what the Chinese government wants censored (I don't) then you should object a lot more to what Google censored for its existence, which could be described as "Chinese censorship and then some" with a heavy stress on the latter.

Wikipedia is censored too:

http://www.douban.com/group/topic/4821659/

http://www.bbc.co.uk/blogs/legacy/technology/2008/12/wikipedia_is_censored.html

Look up Wikipedia articles on Dalai Lama, Li Hongzhi, Falun Gong. They are mostly censored and the articles are written from a very narrow political viewpoint. Do you have any proof of wikipedia being neutral, can you show examples of this supposed neutrality?

realmayo: I hope you know that both Yahoo and Google offered censorship ("filter" as they call it) services to the Chinese government, not the other way around? in other words the Chinese government wasn't the one who made the first step. (I don't know how this was portrayed in the media but I'm sure you can look it up.)

[MeiMay]

Censorship isn't the only issue though. MSN, Bing and Google are logged by the NSA, all logs go to the NSA (and to the UK and other countries with agreements with the USA). Not subpoenaed logs but all logs (it's been in the news). baidu logs probably go to the Chinese government though not quite to the extent the American government takes it. baidu works more on the principle of something being reported/spotted. (it may be hard to swallow but USA surveillance is more extensive than Chinese surveillance)

if it matters to you who has your information then it's probably best to choose the government who finds your opinions least objectionable. based on that alone I don't think I want to use search engines controlled by the USA. my political opinions would probably make me a "terrorist" or something according to the American government.

[ouyangjun]

 

 

my political opinions would probably make me a "terrorist" or something according to the American government. 

 

Not at all.... just a 五毛党