September 5, 2003    Business

The Politics of Search Engines - Re: Google’s PageRank

“PageRank: Google’s Original Sin” by Daniel Brandt raises interesting questions about the power structure of the Internet which at this juncture is measured by Google’s massive popularity. Though his points are interesting, there are some confusing aspects that I would like to clarify.

Brandt argues that Google is “undemocratic” because “rich people get more votes than poor people.” He states, “the rich get richer, and the poor hardly count at all.” He is referring to the algorithm Google uses to calculate page rank, a standard by which Google measures the significance of a page. Google ranks pages largely on two factors: (1) The number of pages that link to your page and (2) the page rank of the linking pages. (e.g., If nytimes.com links to your page, it counts much more than if I link to your page). The former is what Google promotes as their “democratic” component while the latter is what Brandt objects to as “undemocratic” or “tyrannical” and upon which the essence of his confusion lay.

If by “democracy” Brandt means ‘equality of power’ then the US capitalist economy is a wholly inappropriate metric of fairness. So let’s help him out a little and distinguish between different kinds of power, namely political and economical. Here we find definitions of freedom (fairness) ranging from ‘equality of power’ versus ‘inequality of power’ which, in the political sense, defines democracy vs. totalitarianism (respectfully) and in economics, communism vs. capitalism. Now let’s extrapolate that “economic democracy” would mean “communism” where everyone has equal monetary power and “economic totalitarianism” would mean “capitalism” where some are allowed to have more monetary power than others. Distilling further, the US has political democracy and economic totalitarianism (capitalism); the former Soviet Union arguably had political totalitarianism and economic democracy (communism). Counter to the effect Brandt is trying to achieve by associating Google with the word “undemocratic”, being undemocratic in the sense of inequality is not a bad thing by default. Part of the reason why the Soviet Union failed was because it gave equal power to the people in the wrong place.

The United States guarantees political equality of its citizens by its political structure. What this means is that those who maintain it are independent of and impartial to the power granted to elected officials. By weighing links (which are votes) Google is quite democratic. In this sense, they maintain the structure, and outside of what the structure itself grants, they themselves do not subjectively grant any power to any particular website. Of course Google managers have the power to design the structure and in this they carry the mantle as “the founding fathers of the Constitution of the web.” If their structure does not have an effective mechanism of checks and balances, it is sooner or later doomed to failure, or even worse, disaster. Considering the fact that Google is a private institution and has no legal obligation to guarantee fairness they are being rather decent about the whole affair. It could be a lot worse. (Imagine if Bill Gates and Steven Ballmer owned Google; what they would do to the search results.)

Brandt admits in his essay that Google is capitalistic (economic totalitarianism). This should not be automatically deemed bad unless you believe in communism. After all, capitalism is working fine in many civilized nations. The aspect of Google’s page ranking Brandt calls “undemocratic” is the fact that higher-ranking sites have more power in ranking other sites. If they eliminated this element then it would be more democratic, but the quality of the search results would most likely suffer (as they have in the past — remember old AltaVista?). An economic equivalent of this would be a company where all employees have an equal power to decide your salary. Although this may sound idealistic, in reality it won’t work. The salaries of executives and mangers would be slashed and there would be little incentives to move up in the world; a situation very much like the one the Soviet Union faced economically. In short: If you gave users a choice between a capitalistic system and a communistic system, most would elect the capitalistic model. This is true in the real world of politics as well as on the Internet. A capitalistic system has ways of producing more desirable effects. You cannot force people to take something less desirable based on ideals.

You can also look at Google’s page ranking as a system that measures real fame. It has all the characteristics of real-world fame: you not only need a lot of people who are less than or as famous as you are to recognize you, but you also need someone more famous to endorse you. An endorsement from someone who endorses a lot of other people is not as potent as an endorsement from someone who hardly endorses anyone. This aspect of fame is reflected in Google’s ranking formula. Say, a page on nytimes.com that links to your page has 100 other links, and a different page on nytimes.com links only to my page. I would receive a higher ranking than you would. This makes intuitive sense because that is how fame or recognition works in real life. There is a difference between when “special thanks” on liner notes of a CD lists 100 different names and when it lists only a few.

These aspects of fame create a turbo-charging effect. The more famous you become, the more famous you become. You can be famous for the sole reason that you are famous. That is, quality becomes secondary. This effect is especially evident in the world of acting. Many actors are famous not because they are good actors, but because their relatives were famous.

Take the artist, Maya Lin, for instance. No offense to her or to her work, but there is a good chance that she would have been just another unknown artist if she hadn’t won the anonymous competition for the Vietnam Veterans Memorial. There is a good chance also that the artist who was in second place would be as famous as she is now had he/she been chosen as a finalist. Her luck is undeniable by the fact that the others who made it to the final round are virtually unknown right now. If the choice was solely based on some sort of definitive standard of quality, the others who made it to the final round should be just as famous as she is solely by their own merits, independent of the memorial competition, but this is not the case. Whether you like her work or not, she is famous. Just arguing about how she should not be famous makes her more famous. Fame breeds fame. It inherently has a turbo-charging effect as with any other forms of power. Trying to prevent this natural phenomenon in page ranking would only be artificial.

This phenomenon is partly due to the fact that fame or recognition has linguistically utilitarian value. Fame has value as a tool of communication because of the way it asserts psychological pressure on members of society. There are certain facts and names we must know not because they are important in of themselves, but because everyone else knows it. The word “Internet”, for instance, may not mean anything to you personally, but you have to know it in order to communicate comfortably with others in our society. In the same way, there are certain websites you want to know not because they are important, but because everyone else knows about them. Many websites are popular because of this reason, not necessarily because of the quality of their content or the quality of their service. There is value in reading something we know to be popular. It is like looking up a definition of a word in a dictionary; we look up a website if enough people mention its name. Fame, or recognition, is part of what makes our language tick. Words like “baseball” are more famous than words like “grammatology”. There is a good reason why you want to know what “baseball” is even if you hate it.

Interestingly, if you want to be famous in real life, you can employ the same strategies you use for increasing your page rank; they are effective in both sides of the computer monitor. You want to be kind and supportive of those who are less famous than you are because you need their votes. You also need recognitions from those above you. It is also important to get an endorsement from someone who truly cares about you and appreciates you, rather than getting someone who raves about everyone. You also learn from page ranking that success is a collaborative effort. You can’t do it alone. And, it takes time. Trying to get attentions from those above you, while dismissing friends at your own level or below, is a bad idea. If you are lucky, you may be able to get that attention from above, but otherwise, you are missing out on a more reliable, long-term solution to success. Find and support your own community. In time, even if you are unlucky, some others in your community will be and they will try to help you move up. This is the reason why many famous artists and writers were all friends even before they became famous. It is no coincidence. They helped each other move up in the world. A few among the bunch get lucky, they become famous, they mention your name everywhere, and you too become famous.

If you don’t link to anyone, and if no one else links to you, and if you are just hoping that someday someone famous would suddenly link to you, there is a good chance that your site would forever remain in the lowest rankings. The quality of your content helps, but success is always a package deal; you need everything else. You cannot just focus on quality, or just on promotion. This is a fact of capitalistic life. Google simply mimics real life in the USA.

Brandt also mentions the fact that it is now more difficult to get traffic to new sites than it was several years ago before Google became dominant. He attributes this phenomenon to the favoritism built into their page ranking algorithm. As frustrating as this may be, this is also how the real world of capitalism works. It is always easier to get ahead in new markets. This is why everyone rushed to dominate the Internet market in the late 90s. As a market becomes more established and more saturated, the competition becomes tougher. Those who became famous and powerful in the early stage will not only try to hold on to their power, but also try to leverage that fame and power to get even more famous and powerful. Latecomers always have harder time in getting ahead. Again, Google should not be criticized for being true to capitalistic life. If you have issues with capitalism, you should be criticizing capitalism.

Page ranking is a useful concept. There is no problem in itself. What we need to do is not to change Google as Brandt suggests, but to build a search engine with a different structure. We simply need alternatives and leave Google to do what it does well. The alternative that I would like is a search engine that offers a more pluralistic view of the world. Google believes in monism: a formula that calculates what the world thinks is important, which is useful, and what is a good thing to know. I’d like to see an alternative that calculates what I think is important; a thoroughly customizable search algorithm (and associated spiders) that knows a thing or two about me, and based upon my preference settings the more I use it, the more it learns about what interests me. It is somewhat like Amazon’s recommendation page. When I search for “chair” using this search engine the results I get are completely different from anybody else’s. My dream engine should even allow me to specify my own measurement of page ranking and create my own standard for keyword/phrase values. For instance, if I search for “John” using this search engine the first thing that it should return is a page about the composer John Cage, not John Wiley the publisher. And, it would also be interesting to be able to search using other people’s settings. I would, for instance, like to be able to search for “fine cigars” under the settings of Bill Clinton to see what comes up first, which would be what the search engine considers to be important for him.

I’ve suggested this to Google a while ago, but I’m sure there are other companies already working on this as we speak. The idea itself is nothing new; the difficulty lies in its implementation. But, sooner or later, we’ll probably see more pluralistic search engines come out on the market. The know-how we gained so far with monistic search engines would not be entirely wasted on this pluralistic system. We share much of what we personally consider important with what the world thinks is important. Even if pluralistic search engines become available, I would still want to use monistic search engines like Google — if only to maintain my knowledge of hip stuff.