Fw: Telling Google what to think
Alkistis Fleischer
fleischa at georgetown.edu
Wed Nov 12 00:36:51 UTC 2003
An intriguing article that will be of interest to all of us who use Internet
search engines.
AF
----- Original Message -----
From: "Le Monde diplomatique" <english at monde-diplomatique.fr>
To: "Le Monde diplomatique" <english at monde-diplomatique.fr>
Sent: Tuesday, November 11, 2003 7:21 AM
Subject: Telling Google what to think
Le Monde diplomatique
-----------------------------------------------------
November 2003
HOW AN ONLINE SEARCH ENGINE INFLUENCES ACCESS TO INFORMATION
Telling Google what to think
___________________________________________________________
Google seemed a brilliant idea - an internet search engine based
on an algorithm that quickly chose and displayed the most
relevant and popular websites. But the web-savvy can easily
manipulate it for the purposes of propaganda or advertising.
By PIERRE LAZULY *
___________________________________________________________
THE 3bn pages of the internet are the ultimate encyclopaedia:
an incompar able quantity of information made available
through tools that can answer almost any request in seconds.
The web's search engines are so powerful that, with a just a
few words to guide them, they can retrieve information on
unknown topics.
Surprisingly, ever fewer of these vital tools survive. Just
four companies in the United States now provide the entire
world's quality net search services. But the amount of data
that has to be assimilated to point the inquirer in the right
direction is increasing: thousands of computers are used to
scour the web and sort information. The most important aspect
is knowing how to recognise the most relevant pages, which
constitutes a search engine's intelligence and determine its
success or failure.
Google proved this when it became the most popular search
engine on the web in less than three years. Its fresh
approach meant it could usually display the required
information on its first page of results. The word quickly
spread. Those in the know encouraged their friends to use
Google and it went from 10,000 searches a day in 1999 to more
than 200m in spring 2003. Google now runs 53% of the world's
internet searches and its ubiquity means that many of its 70
million users identify it with the internet itself.
"Google has discreetly established itself as an essential
tool whose role goes way beyond the usual idea of a search
engine," says journalist François Pisani. "I don't just use
it to find sites containing information. The nature of the
results that come up is answer enough for many of my
questions and the sites referred to are now just a means of
verification."
But Google's supremacy leads to serious questions: how can an
algorithm, however ingenious, choose the 10 most relevant
responses to the request "Iraq" from the 3m pages containing
this word? Like any search engine, Google has a fundamental
limitation: it can only bring up publicly available
information. If no one has published an article on the habits
of the bearded vulture, any search for information on the
bird will be fruitless. When you search the net you are not
examining all available knowledge, but only what contributors
- universities, institutions, the media, individuals - have
chosen to make freely available, at least temporarily. The
quality of it is essential to the relevance of the results.
And although the total number of web pages grows unstoppably,
some institutional information sources have deliberately
reduced their content. Immediately after 11 September 2001, a
number of official sites in the US deleted all sensitive
data: one US army site had until then proudly shown the world
its eight chemical weapons warehouses (1). Much civilian
information was also removed. Geographical Information
Services denied access to road maps (2), while the State of
Pennsylvania shut down a web service that gave people access
to charts of its telecommunications network including
locations of school districts, hospitals and microwave sites
(3). Some companies used the struggle against terrorism as an
excuse to get rid of information that environmental groups
had fought tough battles to release. California electricity
gener ators withdrew data on their power stations' pollution
emissions (4).
The collapse of the new economy in 2001 also contributed to a
reduction in information. More web publishers now reserve
their articles for subscribers. This strategy, aimed at
increasing revenue, makes sites seemingly disappear. Pages
only accessible to subscribers (even free ones) are ignored
by search engines. Even if the New York Times had published a
remarkable investigation into the habits of the bearded
vulture only last month, it wouldn't come up on a web search.
Most of the press is now almost invisible.
New players are taking over the net. Com panies wanting to
increase their visibility make it their premier communication
tool; militant organisations find it a great way of promoting
their causes. And it is becoming more common for web-surfers
to create their own personal sites. Self-publishing, once the
preserve of geeks, has been democratised thanks to
easier-to-use software.
Faced with the influx of information in the mid-1990s, two IT
students at Stanford University, Sergey Brin and Larry Page,
had an idea: a search engine based on a mathematical analysis
of the relationships between websites would produce far
better results than the basic techniques then in use.
Convinced that the pages with the most links to them on other
sites must be the most relevant ones, they decided to test
this thesis as part of their studies, and did the basic work
for a mathematical search engine. They founded their company
in September 1998 - Google.
To assess the relevance of web pages, Brin and Page invented
PageRank, Google's special page evaluation system. The rank
of a web page is constantly re-adjusted according to how
many, and which other, sites carry links to it. The profile
of isolated sites to which no one has provided a hyperlink
remains low and so they are thought to be without legitimacy.
Google considers sites that are frequently linked to as key
pages. This novel algorithm produces impressive results.
But a problem presents itself. Newly created sites are at a
disadvantage: they will only become visible if they manage to
attract the attention of well-established sites. Google's
founders say: "PageRank is based on the purely democratic
nature of the web." But they concede "votes from pages that
are important themselves count for more, and boost the
importance of other pages". It's a strange democracy where
the voting rights of those in a position of influence are so
much greater than those of new arrivals.
Andrew Orlowski (5) explained how on 17 February 2003, an
article on the front page of The New York Times described the
protest movement: "The huge anti-war demonstrations around
the world this weekend are reminders that there may still be
two superpowers on the planet: the United States and world
public opinion." The idea spread rapidly and the phrase was
soon used by the United Nations secretary general, Kofi
Annan. Over the following weeks, if you searched for "second
superpower" on Google, this was the definition you got.
On 31 March James F Moore of Harvard University put on his
personal website the title "The second superpower rears its
beautiful head" heading an anodyne text that used the term
"second superpower" to describe net-heads (6). (This
application could seduce even a Repub lican party member.)
Techno-utopians seized on Moore's obscure article and made it
a reference point. A month later, 27 of the first 30 results
Google presented for the term "second superpower" pointed to
Moore's piece. Moore, an expert on economic strategy,
technology and leadership, knew what he was doing. Orlowski
says: "Although it took millions of people around the world
to compel the Gray Lady [the New York Times] to describe the
anti-war movement as a 'second superpower', it took only a
handful of webloggers to spin the alternative meaning to
manufacture sufficient PageRank to flood Google with Moore's
alternative, neutered definition. If you were wearing your
Google-goggles, and it was your primary view of the world,
you would have a hard time believing that the phrase 'second
superpower' ever meant anything else."
For Orlowski, this shows that "Google is not authentic but
synthetic" since a search does not lead you to the principal
reference on a subject, but to its most commonly referred-to
application. Copyright legislation has exacerbated the
problem: it forbids the publishing of copyrighted material
online. A search for Raoul Vaneigem will lead you to his
writings, since many of his books are available online, but
type in the names of most writers and you will only get an
invitation to buy their books, or at best, a review by
another surfer. It is as though a library had to give up free
loan of its books and replace them with summaries written by
readers.
This absence of reference points leads to much careful
manoeuvring on the ideological playing field. Curiously, the
symbolic power - the possibility of making your own
perception of a subject the dominant one - is one of the few
outside conventional power structures. The dominant ideology
is not over-represented. Searching for the French home office
minister in charge of charters will throw up charities that
support asylum seekers. Asked about a businessman, Google
could easily ignore press releases and unearth financial
scandals.
The players' influence depends mainly on their relationship
with the net. Creating a website is not enough: you must also
be able to foster links with other sites and get recognition
from those that matter.
While many benefit in good faith from recognition of the
content of their sites, others cunningly exploit weaknesses.
Some companies specialise in producing, for the benefit of
lobbies, information sites whose contents resemble business
memos. Their apparent objectivity is often enough to seduce
surfers into linking to them, bestowing on them valuable
symbolic authority.
Sensitive topics, such as genetically modified organisms or
the Israeli-Palestinian conflict, are the object of bitter
struggles between different parties trying to make their
ideology seem the more legitimate to Google. The manager of a
key website - key to the search engines, anyway - was
surprised to receive a business proposition: "I am interested
in buying hyperlinks on your site to promote our clients'
websites. There is no need for these links to be displayed
prominently on your site: we are not expecting a direct
increase in hits from this. Since your site is popular with
the search engines, these links will make our sites more
visible on them as well." The expert who sent this counted
financial sites, travel agents and pharmaceutical companies
among his clients.
Google's limits are most evident when confronted with
political questions, where there are radically different
points of view among web users. Its mathematical criteria
give de facto priority to certain opinions and can confer
undue importance on texts that represent the opinion only of
a few. Those who got there first in net use are now so
well-established that they enjoy a level of representation
out of proportion to their real importance. The quantity of
links they maintain (especially through the mainly US
phenomenon of webloggers) (7) mathematically give them
control of what Google thinks.
Google has proved brilliantly effective, technically and
practically. But algorithms can't always determine what is
important.
________________________________________________________
* Pierre Lazuly manages and writes for the French-language
websites menteur.com and rezo.net
See also : For sale - to whom?
(1) "Security Concerns Prompt Army to Review Web Sites
Access", Defense Information and Electronics Report, 26
October 2001,
(2) San Francisco Chronicle, 5 October 2001.
(3) The News Media and the Law, Review of the Reporters
Committee for Freedom of the Press, Autumn 2002.
(4) Jean-Pierre Cloutier, "Crise: Sites Web censurés,
modifiés, amendés", Les Chroniques de Cybérie, 30 October
2001 (in French).
(5) Andrew Orlowski, "Anti-war slogan coined, repurposed and
Googlewashed in 42 days", The Register, 4 March 2003,
(6) James F Moore, "The Second Superpower Rears its Beautiful
Head",
(7) Those who publish their own writings on a personal
website.
Translated by Gulliver Cragg
________________________________________________________
ALL RIGHTS RESERVED © 1997-2003 Le Monde diplomatique
<http://MondeDiplo.com/2003/11/15google>
More information about the Linganth
mailing list