Influence of Search Engines Christina Pong cs349.

Influence of Search Engines

Christina Pongcs349

Introduction

The web is meant to be the ideal tool of a democratic society- anyone can say anything

We find information on the web through search engines, directories, or by clicking on links on other web pages

Search engines define what we see on the web, but in 2000, the total coverage only added up to roughly 42%. Today the number is even lower.

How do search engines decide what users will see?

Search Engines and Directories

Search Engines index using keywords

Robots search URLsBacklink MetricPageRank MetricLocation Metric

DirectoriesHuman gatekeepersYahoo’s “Criteria of relevancy” Logos taken from their respective websites

Ranking Concerns

Top 10-20 spots are the most coveted People use various techniques to improve

their rankingsWeb spamming- keyword stuffingPaid advertisementsBuying keywords

Idealism vs Cynicism

No one owns the web

Public space Supports democracy Conveys information Narrows gap

between “haves” and “have-nots”?

Web leads to increased isolation, division, and reduced discourse

“Freedom of the press is guaranteed only to those that own one”

-A.J. Liebling, journalist

Introducing the Googlearchy Study (2003)

In 2000 a survey reported that 30% of users looked for political information on the web

Scholars assumed two things: The web will generate lots of new, easily assessable

content, which lowers the cost of information The web will make it easier for people see information

from lesser known sources, thereby increasing equality

Inbound Links

Finding information by web surfingThe amount of traffic web sites through

surfing is still based on inbound links Using Search Engines

Google’s PageRank system is based on inbound links and 85% of users reported using Google as their search engine

HITS algorithm uses “hubs” and “authorities”, but still produces similar ranking results

Concept of Googlearchy

The cost of information and variety of content on the web is determined by link structure

In general, a small handful of sites get a larger ratio of inbound links and therefore a larger amount of traffic

Power-law relationship

Methods

Created lists of top 200 political websites in six different categories from Yahoo and Google

Used web crawlers to download all pages to a depth of four and classified them as “positive” or “negative”

Support Vector Machine (SVM) Classifier

SVM

Results

Other Findings

Significant amount of overlap in the pages found by Yahoo and Google

Generally a small handful of sites at the top get more backlinks than all of the other sites put together

Top sites are from long established interest groups

Also…

Given the small diameter of the web, the shallow depth of most searches, high degree of overlap, and that search engines are designed to copy user behavior, they believe that their study covered almost all sites that matter for their categories

Any site that is more than 3 clicks away from the top 200 on a Google or Yahoo search probably would not have much of an impact on politics

Study Conclusions

People thought that the web’s effect on politics is that it would lower the cost of information thereby reducing inequality

It lowers the cost, but due to Googlearchy, people to focus on only a few, heavily-linked sources

Food for Thought

The web is valuable because there is so much information, but its size makes it hard to study

An even less organized structure would be harder to navigate and it would be harder to find information

While anyone can post to the web, for most people it’s like: “Having the online equivalent of having their own talk

show on public access television at 3:30 in the morning”

-Matthew Hindman et al.

The web as a meritocracy- the idea a site must be good if everyone’s referencing it

Backlinks are the currency of the web

http://www.daily-seo.com/images/link-building-campaign.jpg http://www.usagold.com/images/gold-coins-images.jpeghttp://blog.kir.com/archives/images/money.jpg

As a Result

What do to about the Monopoly?

Market Dynamics Yahoo argues that if users don’t like search

engine results, they can just switch engines, which will encourage competitors to improve their own ranking systems

But so few people understand how engines work, so they wouldn’t know to switch

More Recently

Does Googleopoly mean that Google is Evil? (Nov. 2008 NY Times article)

A study conducted in 2006 (Filippo Mencer et al) found that search engine results are becoming less influential because people are searching using more specific queries, which reveals the more obscure pages.

Sources

Hindman, Matthew et al. Googlearchy: How a Few Heavily-Linked Sites Dominate Politics on the Web. March 31, 2003. All charts and tables were taken from the Googlearchy paper

Introna, Lucas and Helen Nissenbaum. Defining the Web: The Politics of Search Engines. IEEE. 2000.

Hansell, Saul. “Debating the Vices and Virtues of Google”. Nytimes.com. November 20, 2008.

Mencer, Filippo et al. Googlearchy or Googlocracy? IEEE Spectrum. February 2006. http://www.spectrum.ieee.org/feb06/2787 12/4/08.