Can Wikiasari beat Google?
A recent article in businessweek details the development of a new search engine from Wikipedia entitled Wikiasari, which is intended to launch early this year. The concept is to make an engine that delivers a better user experience by opening up various aspects of the system to the community. The two ways in which Wikiasari intends to do this are by utilizing open source code and a user rating system for search value.
Opening the algorithm for developers is a great idea. This allows the code to be improved by an exponentially larger number of contributors. The quality of an open source search algorithm can only improve by granting more users access to it, and because profit driven or self benefiting volunteer code changes would be refused by other community developers, the primary goal of improving user search experience is protected. This should eliminate any concern that a user could succeed in manipulating source code negatively. The open source system of checks and balances has been tested time and time again, and would function no differently here.
People may try to find flaws and exploit the algorithm for its weaknesses to benefit themselves in some way. For Wikiasari, developers may dissect the code to find an unintended means to rank their sites higher. However, exploits usually become publicized, and once they do, they’re corrected by the community. Historically, people have been repeatedly exploiting Apache’s HTTP server since its debut. As a result, the flaws have been continually addressed and corrected; subsequently, it has become the most popular server of its type across the Internet. Opposing the option of opening source, and keeping code closed does not mean its flaws are any more
protected than that of open sourced code. For example, most of Microsoft’s products are closed source, and are famously known for being security liabilities. In short, opening source has produced some of the most secure and widely used programs on the Internet.
That being said, the open source aspect of the engine has its benefits; the user rating system, however, creates a major downside, and may ultimately result in Wikiasari’s demise. In the article, Wales complains about Google results that he says are “very bad” at times. From this statement, it sounds as though Wales is one of the members of the searching public who would sooner blame the engine–which serves up billions of requests daily–rather than his own queries. The opposite of this is the audience that trusts a search engine’s ability to return relevant results, and in the event a search produces unexpected results, they will reword the search. It is easy to make a correlation between these qualities and a searcher’s age or years spent searching.
There is an estimated growth period of 3 years before Wikiasari will return truly meaningful results and take off. While there are, of course, adoption timelines on most Internet tools, there are still concerns that because the results are being changed constantly, people will not be searching on Wikiasari enough to generate and maintain the quality reviewing necessary to make Wikiasari different. The typical user is not likely to withstand the unsatisfactory developmental results while still trying to complete a real search. Incentive for users is to create a better search for everyone, but once typical users have found what they need through search, they’re finished using the tool.
The user rating system is also inherently flawed. People will abuse the system. To begin with, there are challenges in determining who is a trusted resource. Then users will also be willing to believe that the system for determining trusted volunteers is unbiased. Consider the fact that Wikiasari’s trusted resources or volunteers would be able to name a price at any time, and become motivated by profit, or any other factor that is not related to providing a good user experience. These scenarios are easy to imagine. Even today, savvy users are aware of profit motivated Wikipedia entries. Whether the Wikiapedia entries are eventually caught or not, the skeptical group will have a certain distrust for Wikipedia. If any of that negative stigma carries to Wikiasari, there could be a significant demographic that would avoid the engine altogether, and if that bias is not enough, then the profit based results may eventually end up being their own deterrent.
A user rating system can work well for terms like “poison tree frog”, but how can the result for “dress shoes” be qualified by people? In essence, human edited results will become political in nature, which may ultimately lead to the search engine’s downfall. There are similar criticisms with DMOZ, whose structure and idea can be compared to Wikiasari’s. DMOZ has seen many editors that volunteer until their sites are listed. Consequently DMOZ has had some operational issues of their own of late, which may be a useful foresight into the political problems for Wikiasari.
Other major historical evidence against the human editing is the failure of snap.com in 2001. Their results were also aided with human ratings until this system was exploited with programs. The searching public will not feel confident in getting spam-like results or knowing that corporations are buying ratings to circumvent the system.
In summary, I believe Wikiasari is an overly optimistic idea. Pure algorithmic search will probably be a superior method of search well into the distant future, if not always, and even though Wikiasari may fail, it should still be developed. After all, few people predicted the first airplane successes. I’ll even offer to eat my hat in 3 years if I’ve replaced Google with Wikiasari.