If Popularity is the Only Tool in Web Search, Every Result Looks Like a Nail!

Popular votes determine the leaders of our societies as a fair practice of equality and human rights. Popular votes among expert physicians can produce better diagnostics of a medical condition. There are many other cases where popularity, as a method, works and functions well.

But, how well does popularity work for the Web search?

Like the saying “if the only tool is a hammer, everything looks like a nail” popularity ranking is the only perspective available out there, thus every result looks like a nail! There is nothing else to compare.

A closer look, however, shows that popularity can sometimes fail miserably without the need for any comparison. Worse than that, it fails in a hidden way at a much higher rate than what we want to believe. I would like to talk about these issues in this post, and explain why we are doing what we are doing at hakia.

Let’s first look at the validity of popular view. Below are some fun examples of how popular view fails. Obviously, this has no impact on our lives other than making us look stupid. There are so many of them that you can find books published in this area, but let’s just list 3 of them.

Contrary to popular belief:
- Nowhere in the Bible is the fruit eaten by Adam and Eve referred to as an apple
- Thomas Edison did not invent the light bulb
- Seasons are not caused by Earth being closer to the sun in summer than in winter.

Fun right? The next category is not fun at all. The history is full of cases in medicine, science, technology, politics, etc., where popular belief produced deadly consequences.

Contrary to popular belief (in history):
- More than 40,000 women executed in Medieval Europe were not witches.
- The Titanic could sink despite its unsinkable reputation.
- To imitate the Marlboro man was not cool at all.
- Lead piping and asbestos were harmful building materials.
- HIV could spread among heterosexuals.
- There were no WMDs in Iraq to pose an immediate danger.

The lesson to be learned from all these examples is that popular view can be wrong. While some of the misconceptions may be innocent due to lack of scientific data, most of them are byproduct of information manipulation for commercial or political benefit. A careful eye will catch millions of such manipulations still at work today.

Now, let’s switch back to Web search. If search results are ranked and organized by a popularity algorithm, which reflects peoples’ choices as to what is right and relevant, how can you trust this view for important queries in health, finance, law, business, etc.? Can some of these results be commercially-biased, politically-biased, or innocently incorrect? Are there better results that you don’t see? We call this the hidden failure.

This is where we separate ourselves from the current wave of popularity algorithms. The semantic algorithms at hakia are not based on collecting statistics on link referrals, click behavior, or on any other similar measure. Our criteria is quality, which is defined as the combination of credibility, freshness, and relevance by meaning match. It is our vision that this new perspective will benefit the Web searcher by minimizing (if not eliminating) the extra burden of quality assurance. There will be no hidden failures by design.

Note that we haven’t even touched the subject of long-tail which is an inherent technical limitation of popularity algorithms. Long-tail limitations directly contribute to the hidden failure.

For the Web searcher, the only tool is no longer just popularity, and we are working hard to bring quality search as an alternative perspective. Until then, keep challenging hakia BETA to benefit from the ongoing progress and to give us feedback.

        

This entry was posted on Thursday, April 24th, 2008 at 9:29 am and is filed under Technology. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.