Despite the influx of foreign upstarts and the fast-changing nature of Korea’s internet landscape, Naver remains the undisputed king of Korea’s portal sites.
As with Google in much of the rest of the world, however, Naver’s market dominance in Korea has sparked occasional fears that it wields too much power. Now, if gripes from an OhmyNews journalist are to be believed, Naver has been using that power to fiddle with search results — for both political and financial reasons.
According to a round-up in ETNews, the controversy began on May 25 when OhmyNews reporter Kim In-song accused the portal giant of pushing illegal copies of original news sources to the top of its search results. Hits on such copies, Kim said, would earn Naver income that hits on the original sources wouldn’t.
Naver struck back. On its official blog on May 27, the portal said:
This misunderstanding arose because there’s a difference in how long it takes for the search engine to gather documents and analyse them. The results are not intentionally compiled according to (the journalist’s) misconstrued standard.
Then on June 1, Naver gave a detailed presentation of how its search engine now includes external blogs (Naver previously included only blogs it hosts in its search results), providing a further opportunity to scotch Kim’s accusations.
But Kim wasn’t done yet.
In another OhmyNews piece on June 10, Kim wrote: “We can’t get past the suspicion that search results related to politics are being manipulated.” Citing examples from Naver’s yearly “Trend Book,” Kim said that the incidence of keywords about candlelight vigils and the campaign to impeach President Lee Myung-bak were very low, as were mentions of a specific candidate for the election of Seoul’s superintendent of education.
Poppycock, replied Naver.
The Trend Book is not a collection of statistics. It’s a kind of magazine documenting trends, and doesn’t strictly enumerate the incidence of keywords in Naver search. The trends can show up through a series of keywords or representative keywords, so there can be a difference.
Regarding the absence of the candidate, Naver said:
In order to avoid abuse of the system, our algorithm automatically excludes words from auto-complete that have been written incorrectly on numerous occasions. We think that’s what may have happened here, so the name of that candidate may not have been automatically included in our “related search” list.
As a quoted expert points out, these cases are probably too isolated to prove anything much. But given how jealously search engines guard their search algorithms, and the Korean public’s often ambivalent attitude to domestic search engines, this story is sure to run and run.