Tencent (QQ.com) to Build a 3000-person Search Army to Power its Search Engine
Zhang Liming (张黎明) from Beijing Morning Post (北京晨报) has a report titled as “Tencent learned from Korean Model to Hire a 3000-people Human Flesh Search Army” (腾讯参照韩国模式招三千人肉搜索军团) on April. 10, 2008 on Sina Tech. The author learned the news from industry insiders and quoted quite some comments from CEO Huateng Ma (马化腾) of Tencent Inc.- the largest and most used Internet service portal in China with annual revenue of $520 MM in 2007, about this big bet action. Here is the summarized translation of the report.
Tencent Inc. (QQ.com) is building a 3,000-person search result editor team. The employees will be/are hired as engineers but in fact, their job nature is to edit search results of its search engine called SOSO (搜搜, means “search search” in Chinese) soso.com (which was launched in Dec. 2005).
CEO Huateng Ma (马化腾) didn’t comment on the size of the editor team directly, but compared with the practice in Korean search engine industry: “a 700-person search result editor team in Korea is very common.”
Ma continued to explain why “it is common”: “for example, 20 users might search one same key term, and what they need might be the same information in two paragraphs. But nobody locks the two paragraphs (on the top of search result thus enables a more efficient search for majority of users). So actually people want editing of search results.”
“Tencent is experimenting with ‘human+search’ model. In domestic market, Baidu Zhidao (百度知道) is a similar model, but its editor team is not strong enough.”
When continuing to compare SOSO with other human-powered search engines in Korea, Ma admitted that “one key reason that Korean local search engines beat Google and Yahoo to win the local market is that there are relatively less pages in Korean Internet (for Google and Yahoo to crawl). So I don’t know if human-powered search engine will be successful or not in China. I have a question mark for this model. But Tencent has a portal (qq.com), the edited search results are valuable to the portal anyway (so it worth a try).”
Other posts about Tencent (QQ.com) and SoSo here:
- Techcrunch: Soso.com partnered with Google
- Kaiser Kuo from Ogilvyy Digital Watch: Coin of the Realm, revisited
- Paul M. Denlinger from China Vortex: Understanding China’s Youth Through Tencent’s QQ: A New Must-Read Report
UPDATE Elliott: 4/14 made minor edits
2 Responses to “Tencent (QQ.com) to Build a 3000-person Search Army to Power its Search Engine”
While the whole idea of hiring an “army of search experts” to power Tencent’s search makes some sense, there are some things about human-powered search which have not made it outside South Korea, where Naver.com is by far the predominant human-powered search engine, and last I heard, commands more than 40% of the search market. Naver’s human-search model formed the core of Yahoo! Answers, which has achieved some degree of success in the US and other markets.
In South Korea, for the most part, people want to be known as the experts in their respective fields. If you want to recognized as an expert, you need to post under your real name, right? And it has worked tremendously well for Naver in South Korea, where some have achieved “god” status in their respective fields. So far, that has not happened in China and the US, and with Baidu’s 貼 吧 which has been popular in China.
I can’t speak for others, but I’d really like it if the person who claimed to be an expert in a field let me know his/her real name. A little bit of credibility goes a long way.
Fascinating news, Min. It reminds me of the discussion of human-powered search engines Wikia and Mahalo at the DLD in Munich (TechCrunch coverage here) when suddenly Google jumps in and defends its algorithms. With all this talk about human results editors, Google suggests its algorithms already use humans to determine which results are most relevant, with PageRank factoring who links to who and how much.
Google has a point there, even as we all know what the point of “human search” and “human-edited search results” usually is. Services like Digg or Mixx also fundamentally subscribe to the notion that information with people linking to or promoting it inherently implies it has authority or relevance for others. In a way, it is actually quite, well…democratic.
The problem with such algorhithms is that it requires user-participation that isn’t accessible to everyone or isn’t desired by everyone, which is, well…also quite democratic. Many people need relevant search results (just like they need good government) but not everyone wants to help the process (learn and vote). So, there’s SEO gaming (special interests and lobbying) of results.
Introducing a human, even 3000 humans, to edit search results is very much a technocratic concept, with the ideal that a few selfless, enlightened humans should exist to look out for the interests of everyone else because they’re not suited to manage themselves. With regards to search engines, these editors would be entrusted with disregarding white noise to unilaterally decide the value of any given search result for all net users.
Both strategies have their promoted pros and criticized cons, both in search services…and in politics.