Measuring intellectual influence in Chinese. And in China.

By Detlef Gürtler and Yuchen Zhang

One of the biggest obstacles in analysing the influence of intellectuals in China is the relatively strong separation of the domestic Chinese Internet from the rest of the world (some special problems of Chinese rankings and how they were solved here are listed below). This relates both to a technical separation with partially blocked content and also the independence of the providers: people around the world google things, whereas in China, Baidu is used. This means that the Chinese-language areas of the ‘normal’ Internet, in particular, are not representative of the discussion within China – those who e.g. use the Chinese Wikipedia usually do so from outside of China. Sources like Twitter, Google and Wikipedia can thus be analysed as Chinese-language infospheres outside of China. This nevertheless concerns around sixty million people: a good twenty million inhabitants of Taiwan and around forty million Chinese living around the world. Certainly a relevant magnitude based on the absolute figure and also in some points completely comparable with the inhabitants of China, although not in many other points. Only for this part of the Chinese-speaking infosphere network diagrams can be created using the same system as the other thought leader assessments.

Baidu instead of Google

We are moving onto even less familiar terrain when it comes to assessing the infospheres within China. For instance, we have no information regarding the extent to which state censorship distorts possible analysis results. And we have no experience values relating to possible analysis criteria. An example: the number of search results for a person on Baidu, similar to the number of Google hits, is one (of many) possible indicators for their importance. Correspondingly, such a Baidu search ranking was also collected. A Baidu search conducted again several weeks later using exactly the same method and made for exactly the same persons produced a high level of correspondence in many cases: five of the ten best-placed people in the first ranking were also in the top ten in the second round. In some cases, however, there were severe deviations, and these were especially extreme for Amartya Sen, who dropped from 9th to 82nd place. No pattern that could explain the deviations could be identified. For the final assessment, we therefore decided to use a ranking with the arithmetic mean of both results.

The other two criteria included in the assessment also come from Baidu: Baike is a service that – similar to Wikipedia – contains biographies; here, merely the length of the biographical entry was ascertained and a ranking produced using this. Like on Wikipedia, they can also be biased, for instance if the persons create or edit their own article themselves. In view of the high level of fame of the people investigated, such bias, if any, should be rare. However, using this criterion, an older candidate does better on average than a younger participant: those who are older simply have experienced more.
The Baidu Index is most comparable with Google Trends in terms of services known in the West. It is a service that ascertains relatively reliably how intensively a person or a topic is being sought at the moment or recently. The index ranking was weighted as strongly in the final assessment as the two other criteria together. For each of the indicators used for this ranking we have posted the ten best-ranked persons below.

Candidate selection

Unlike in the English- or German-language infosphere, there are no regular top lists of powerful and important personalities in or for China, such as the lists by ‘Prospect’ magazine or the ‘Time’ magazine lists from which many of the candidates for our thought leader assessment were taken. Only a more than ten-year-old list was used for the selection of Chinese thought leaders: the list of fifty influential Chinese intellectuals published in 2004 by the magazine ‘Southern People Weekly’ (‘Nanfang renwu zhoukan’) from Guangzhou. Some of them had already died, while others did not fulfil our criteria (such as TV presenters), yet this, and a small number of Chinese names on international intellectual rankings, produced a foundation of candidates. It was supplemented with suggestions from the Berggruen Institute network. As in the language versions for the German and the Spanish infospheres, the field of candidates was topped up with the forty personalities at the top of the global (English) thought leader ranking.

When it came to selecting candidates, there were particularly intensive discussions regarding the extent to which currently active political leaders could be taken into account. The separation between ‘thinkers’ and ‘doers’ has been an integral part of the study design since the first thought leader assessment: the assessment should not be of those who have an influence based on the power of their position but rather those who influence humanity through their thoughts. However, there is a serious argument why this separation should be waived in the case of China: there is a deep-rooted notion there that it is the role of the best minds to get actively involved in the state for the good of the people. The most well-known expression of this notion is the traditional appreciation of the career as civil servant: over long periods of Chinese history, it has been the dream of all talented adolescents to pass the annual entrance exam to become a civil servant.

Just like the ‘American dream’ is a self-made career as an entrepreneur, the ‘Chinese dream’ is a career in the civil service. And top Chinese politicians, even if they should have risen formally in the Communist Party, are not so much subject to elections as they are active in the civil service. Removing them from the assessment right from the start can mean that some of the best and most influential thinkers in the country are simply not taken into account. The Western idea that thinkers can express themselves better the more independently they are able to act is in contrast with this close connection between mind and state.

In this specific case, the candidate selection for the first thought leader assessment in China was conducted in such a way that active politicians suggested as thinkers in the ‘inside China’ assessment were included in the assessment – and the top-most public official, President Xi Jinping, was included for comparison outside of competition. That would have been sufficient, if very close, for him to achieve first place, ahead of political veteran Henry Kissinger, the Alibaba founder Jack Ma and winner of the Nobel Prize in Literature Mo Yan.

Methodological problems (and solutions) for Chinese rankings

Transcription as a source of problems

There is a clear and consistently-used method of transcribing Chinese names into English: the use of ‘pinyin’, the alphabetical transcription of Chinese characters. The situation when transcribing English names (and names from other alphabets) into Chinese is different. These names are usually transcribed based on their pronunciation. The series of characters created is arbitrary, almost random, as many characters sound very similar. Widely different transcriptions can thus occur for one and the same person. However, for most of the people assessed here, there is a predetermined and accepted transcription according to their level of fame. Where there is more than one transcription version, the more widespread form was used for the analysis.

Names as a source of problems

While there are many different surnames in the West and a comparatively smaller variety of first names, the situation is the exact reverse in China: a great many people have the same surname, with the differentiation being made using first names. However, the differentiation is not necessarily visible in the transcription: names that sound and look the same in pinyin can be completely different when written in characters. As a result, the name in characters is used in each case to identify people. However, there is still the potential for confusion, such as in the case of Wang Shi (王石). It is not only the name of an entrepreneur, the founder of one of the world’s largest real estate companies, but also the name of many thousands of other Chinese people. To ensure that we were analysing the right people, we manually checked the websites or Wiki pages in such cases or used the Baidu Index, as when people look for Wang Shi, it is almost always for the prominent entrepreneur, not just anyone called Wang Shi.

Border as a source of problems

The Internet, Google and Wikipedia are important sources for the assessment in other language areas. However, within China, these sources can only be used with severe restrictions, if at all. This results in the often described separation of the Chinese Internet from the rest of the world (‘Great Firewall of China’), while the thought leader tools can only be used for Chinese-language sources outside of China, such as the Chinese Wikipedia. These sources thus tend to reflect the needs and positions of the Chinese people outside of China rather than those of all Chinese people. An ‘outside China’ analysis was thus created using these sources. Within China, Baidu fulfils a similar role to Google in the rest of the world, while Baidu Baike is most comparable with Wikipedia. These two sources were thus used for the ‘inside China’ analysis.

Software as a source of problems

For the ‘inside China’ analysis, no network depictions could be undertaken as Baidu Baike, unlike Wikipedia, does not include backlinks. Neither the betweenness centrality nor the degree centrality could therefore be calculated in this case.

Censorship as a source of problems

Content regarding certain politically sensitive personalities such as the Dalai Lama is censored within China. As a result, it is possible that more communication and conversation about them takes place within China than shown in the analysis results, and these people should thus be rated more highly than depicted here. However, there is no way of estimating whether there even is a censorship effect and, if so, how great it is.