Mandelbrot set image very small
Serious writing for
serious readers
Follow @Mohammed_Amin
Join my
email list

Search this site

Custom Search
Tap here for MENU

Review of "The Myth of Digital Democracy" by


29 April 2011

Forty years ago I could never have imagined that I, alongside billions of other computer users, would be able to sit down at my keyboard and type something that could then be read by every other computer user on the planet. The Internet and blogging technology makes that possible for everyone - it would have been impossible before the computer age.

Many commentators believe this makes it possible for every citizen in a democracy to take part in the public debate. The argument is that the Internet is redistributing political influence, increasing political participation and challenging the monopoly of traditional elites. The key belief is that the technology will amplify the political voice of ordinary citizens.

Matthew Hindman is an assistant professor in the School of Media and Public Affairs at George Washington University. His short book demonstrates, with robust evidence from the USA, how wrong such a belief is. While everyone can speak on the Internet, few of us are heard.

"In this respect, the Internet diverges from much of what political scientists have grown to expect from the literature on political behaviour. In many avenues of political participation, scholars have noted that once the initial barriers to participation are overcome, citizens' voices get considered relatively equally. When citizens vote, each ballot carries the same weight in deciding an election. When citizens volunteer for a political campaign or an advocacy group, they all face similar limits; at the extremes, no volunteer has more than twenty-four hours a day to contribute towards a campaign. The greatest exception to this rule has been political fund-raising; among the relatively small set of citizens who donate to political campaigns and interest groups, disparities in wealth make some citizens' voices much louder than others. Even here, though, there are important (albeit imperfect) limits that constrain inequalities in who gets heard. Under federal election law, no citizen could donate more than $2,000 total to any one candidate over the course of the 2003-2004 election cycle.

A central claim of this book is that direct political speech on the Internet – by which I mean the posting of political views online by citizens – does not follow these relatively egalitarian patterns. If we look at citizens' voices in terms of the readership their postings receive, political expression online is orders of magnitude more unequal than the disparities we are used to in voting, volunteer work, and even political fund-raising. This book also shows that by the most commonly used social science metrics, online audience concentration equals or exceeds that found in most traditional media."

In the review below I have highlighted some of the points that I found particularly interesting but there is much more to read in the book. I recommend it to anyone who is interested in how real people use the Internet.

"Googlearchy”: The Link Structure of Political Websites

Physically the Internet consists of large numbers of computers connected by wires or by wireless links. However in logical terms the World Wide Web consists of pages which are connected by hyperlinks. Google has become one of the world's most valuable companies by exploiting the insight that the structure of these hyperlinks conveys meaningful information about which sites users want to see ranked highly in Internet searches.

The author and his team carried out detailed research on political websites.

"Millions of Americans have now created their own blogs or websites. Hundreds of thousands of businesses and organisations have followed suit. Creating a link to another website hardly conjures up the energetic activity that "activism" assumes, and those linking to other sites may not even be advocating the political views they reference… The way in which these website owners link to each other is anything but random."

The author points out that existing research has shown that over the entire web the distribution of inbound and outbound hyperlinks follows a power law. This means that the probability that a randomly selected web page has K links is proportional to K-a for large values of K, where “a” is a constant. The author illustrates this with a numerical example:

"Imagine a hypothetical community where wealth is distributed according to a power law. At one end of the spectrum, there is one millionaire, 10 individuals worth at least $100,000, 100 people worth $10,000, and 1,000 people worth at least $1,000. At the opposite end, 1 million people have a net worth of one dollar. In this hypothetical community, wealth is distributed in proportion to the function K-a, where a = 1.”

Studies of the web indicate a value of a=2.1 for inbound hyperlinks which is much more unequal than the example above where a=1. This means that a few popular sites have enormous numbers of links while most personal webpages receive hardly any links at all.

A closer look at the mathematics of power law distributions

Proportionality is something that is not easy to visualise without a closer look at the numbers. It is better to think of the author's example as an equation:

N = Number of people having a particular amount of wealth

K = The particular amount of wealth that one has in mind

Then N = 1,000,000 * K-1

The results of the above equation for the wealth values mentioned by the author is set out in the able of numbers below:

Wealth People
 $                      1          1,000,000
 $              1,000                  1,000
 $            10,000                      100
 $         100,000                        10
 $      1,000,000                           1

The above table shows that a power law with a=1 produces quite a high level of inequality. To have one person owning $1 million one needs to have 1 million people who have only $1 each.

However this degree of inequality pales into insignificance when you compare it with the unequal distribution of inbound links on websites.

The author states that for inbound links on websites, the value of "a" is about 2.1. This is demonstrated by the table below; for comparability of the tables I have "normalised" the figures so that the number of websites having 1 million links is 1.

Links Sites
                          1          3,981,071,705,535
                 1,000                          1,995,262
               10,000                                15,849
             100,000                                      126
         1,000,000                                           1

At the level of 1,000 dollars or links, the internet is approximately two thousand times more unequal than the hypothetical wealth distribution given above.

Empirical test on political websites

Traffic in the form of readership follows a similar power law distribution. The author and his team tested this by generating lists of political sites from search engines and then building web robots to crawl the sites following links three deep. Accordingly starting with 12 lists of 200 sites, their robots downloaded about 3 million pages which were classified for relevance to politics and the distribution of links ascertained. This detailed work confirmed the applicability of a power law for links to these political websites. He devises a new word "Googlearchy": rule by the most heavily linked.

Author's conclusions

The author makes three key points from his work:

  1. The number of links pointing to a site is the most important determinant of its visibility. Site with many inbound links should be easy to find and should receive more traffic.
  2. For every category of website, a small portion of the group should receive most of the links and most of the traffic so that winners take all.
  3. The dominance of a niche becomes self-perpetuating. Heavily linked sites continue to attract more links, more visitors and more resources while sites with few links remain ignored.

Political Traffic and the Politics of Search

The author and his team obtained data from a company called Hitwise. As of May 2006 the Hitwise database included over 1 million websites with details of traffic from 10 million US households who were customers of its Internet service provider partners. The Hitwise data allowed the author to see (in anonymised form) which sites users visited before and after a particular website. 20% of visits to news and media websites and more than 25% of the visits to political websites came directly from search engine queries.

The first key finding was how little traffic goes to political websites. In the sample, more than 10% of total website visits were to "adult" sites while political sites received about 0.1% of visits, a factor of 100 less. News and media sites received about 2.9% of web traffic which is still 30 times as much as the political sites.

"Discussions of the online public sphere have imagined that political blogs, advocacy organisations, and other non-commercial outlets would challenge the monopoly that commercial media have had on public discourse. Judging by traffic, this challenge does not seem to be especially strong. News and media sites still receive 30 times as many visits as political websites do. That level of readership is large by the standards of traditional opinion journals, such as the Nation, the New Republic, or the National Review, all of which are minor print publications. Yet political sites remain a small niche amid the larger web."

The author's findings regarding the demographics of users are also interesting. While younger people are generally over-represented on the Internet, that is not the case with online politics. 18-34 year-olds account for 43% of all web traffic but generate just 30% of the business to news sites and only 22% of the visits to political sites. The converse is true for older people.

Online Concentration

Apart from legal monopolies, concentrated markets arise when there are economies of scale. In other words the more the firm produces, the lower are its average costs. The newspaper industry is one such business as newspapers face high fixed costs but low marginal costs, so the more copies printed the lower the average cost per copy.

The author points out that the Internet dramatically reduces distribution costs; for example instead of printing physical newspapers a newspaper website makes things available for electronic access. However the Internet does not eliminate the costs of producing content; even if every reader of the New York Times switched to the electronic edition, the company would still need to employ the same number of journalists, writers and photographers. Accordingly Internet publishing is an industry that has economies of scale since there are fixed costs of producing the content. The larger the readership the lower the average cost per reader.

The author’s team measured concentration of news and media sites and found the high level of concentration that would be expected from the above discussion of the industry's economics.

Blogs: The New Elite Media

This chapter starts with an overview:

"Those who have been enthused about the Internet's political implications, as well as those who have looked at the new medium suspiciously, have begun by assuming that the Internet will funnel the attention of the public away from traditional news outlets and interest groups and toward countless small-scale sources of political information. As previous chapters have shown, this assumption is problematic. Winners-take-all patterns in the ecology of the web – both in its link structure and traffic – do not fit with what many have assumed."

Political blogs were virtually unknown in 2000 but a major feature of the US political scene by 2004. Some commentators even argued that blogging and "citizen journalism" would displace the “old” media. The author demonstrates the incorrectness of this view and that successful bloggers are not representative of Americans at large.

"Bloggers fit poorly into the narrative that has been constructed for them. Though millions of Americans now maintain a blog, only a few dozen political bloggers get as many readers as a typical college newspaper. Yet the problem is not just the small number of voices that matter; it is that these voices are quite unrepresentative of the broader electorate."

According to the Pew Internet survey in November 2004, of the 120 million Americans who were online by then, about 8 million had created a blog while about 32 million reported reading a blog. However, blogs remain a niche; only 2% of Internet users visited political blogs daily while 77% never visited political blogs.

The author points out that the most important claim about blogs in political discourse is that they amplify the political voice of ordinary citizens. He cites a number of quotations from other commentators:

  • "Blogs enable anyone with an opinion to be heard."
  • "When you have a theory or concern you can put it on your blog and you can tell the whole world."
  • "Blogs allow anyone with an opinion the ability to reach millions of people instantly and simultaneously."

He demonstrates that readership of blogs is highly concentrated. In March 2005 the top five blogs together accounted for 28% of blog traffic; the top 10 blogs accounted for 48%. Sites that received over 2,000 visits a day in aggregate received 74% of the total traffic.

The author itemises the personal profile of the top 10 bloggers at that time, demonstrating that five of them were current or former professional journalists and that all of them were highly educated representatives of an elite. His team extended the work by conducting a census of all 87 political blogs that received more than 2,000 visitors per week, receiving responses from 75 of the 87. This demonstrated the elite education of the successful bloggers; in no sense could they be regarded as representative of the average American.

Elite Politics and the "Missing Middle"

In the concluding chapter, the author recaps on his key findings.

  1. Political traffic is a tiny portion of web usage. As mentioned above, pornographic content receives roughly 100 times the traffic that political websites receive.
  2. The link structure of the web limits the content that citizens actually see. The basic pattern is the winners take almost the entire readership; there is equality in the freedom to publish but publishing does not mean you have readers.
  3. Much use of search engines is superficial. When the team reviewed the Hitwise database they found that people normally use very simple search terms and rarely look beyond the first few results from the search engine.
  4. Even with the Internet, quality content is expensive to produce. Most online news is supplied by traditional news organisations with large journalistic resources.
  5. Even in a new area like blogging where initially there were no incumbents and content can be produced cheaply by a single individual, social hierarchies have quickly emerged. A small group of bloggers receives more traffic than the rest of the blogging population combined; these top bloggers are a highly educated elite even less ethnically diverse than the traditional elite media.

As the author says, "It may be easy to speak in cyberspace, but it remains difficult to be heard."

Reflections on my own website

Although I have just written this review, I originally read the book over a year ago before I started constructing my personal website. I did not let the book deter me from starting!

My experience has been that it is possible to become noticed on the web even from a standing start. The key requirement is to create content that is better than anything else available on the web, in the eyes of the readers that you are aiming at. The point here is that there is no absolute measure of quality; my website does not contain academic papers but rather relatively short items which are intended to add to the reader’s knowledge or to make him think about an issue differently.

An example is my paper "Triangulating the Abraham Faiths.” This looks at the relationship of Judaism, Christianity and Islam in a very simple yet distinctive way. At the time of writing it is the number one search result on Google if you type any two of the words Judaism, Christianity and Islam followed by the word "closeness."

Accordingly I have been pleased by the way that my site has slowly attracted more inbound links, higher rankings on Google for a growing number of searches and a steadily increasing readership albeit one that started from a base of zero.

Kindle edition above


The Disqus comments facility below allows you to comment on this page. Please respect others when commenting.
You can login using any of your Twitter, Facebook, Google+ or Disqus identities.
Even if you are not registered on any of these, you can still post a comment.
comments powered by Disqus


Custom Search

Follow @Mohammed_Amin

Tap for top of page