Coolhunting on the Web for Stars

Doing a degree-of-separation search is a quick way to find the most influential nodes in a local context. Combining multiple datasets, each collecting the Web sites collected through querying Google for the name of one “star” permits to find the most central star. Comparing the betweenness scores of the stars predicts the “coolest” star. By combining the nodes returned by different degree-of-separation searches, we can compare betweenness of different people, identifying the ones with the highest betweenness. What this means is that they are the most linked, or most “talked about” on the Web or in the Blogosphere. This process will now subsequently be explained through tracking the trends of political candidates.

Figure 3 displays the Google query results of doing a degree-of-separation search for the leading 7 republican and 7 democratic contenders to become the next US President, as of end of August 2006.

Figure 3 Combined degree-of-separation search results for 14 US Presidential hopefuls

Each of the colors identifies the set of nodes returned by the Google queries for one of the presidential candidates, e.g. the Web sites returned to “Al Gore” are shown in blue. The red nodes are the Web sites returned by more than one query. Table 1 lists the results of the two most recent presidential polls as of end of August 2006.

Table 1. Results of US presidential polls Aug 2006 (source Wikipedia)

Based on table 1, we would expect Hillary Clinton and Rudi Giuliani to be the most between actors in our Web coolhunting analysis. The result is slightly different, however. While there are no surprises for Rudi Giuliani, Hillary is not really the top ranked democratic candidate by betweenness. This honor falls to Al Gore and John Edwards, which are tied for first place (Figure 4).

Figure 4. Visualization by betweenness of 14 US Presidential Hopefuls

The same evaluation by betweenness also permits to find the most relevant Web sites discussing presidential candidates. These Web sites also double up as “kingmakers.” Kingmakers are the Web sites that, through linking to a query, increase the betweenness of the original query through their own high betweenness. In our presidential polling analysis, en.wikipedia.org and www.ovaloffice2008.com are the most between Web sites. While it is not surprising that wikipedia is very central, as all candidates take care to get their profiles entered and updated there, the central position of ovaloffice2008 comes as somewhat of a surprise. Looking at the individual query results for each candidate, wikipedia, the candidate’s own Web sites, and the sites of national newspapers such as the New York Times or the Washington Post rank higher. If the Web sites returned to the 14 different degree-of-separation queries are combined, however, a different picture emerges, with wikipedia and ovaloffice2008 by far having the highest centralities. While the Google page rank of Wikipedia is 9 (out of 10), ovaloffice2008’s Google page rank in August 2006 was only 5. Its betweenness in the context of presidential elections however is the second highest of all Web sites included in this analysis of presidential hopefuls. Ovaloffice2008 also includes a very active forums where citizens of inclinations and party colors discuss strengths, weaknesses, and chances of the various candidates, further adding to the centrality of this Web site.

Collecting the contents of the Web sites returned to a degree-of-separation query permits to obtain the top ranked terms describing the “hot topics” for each “star”. Figure 5 shows the hot topics around Hillary Clinton.

Figure 5 Key terms in the context of Hillary Clinton

As figure 5 illustrates, “California”, “veterans”, “John Kerry”, “New York”, “Texas”, and “avian flu” are the key topics coming up when analyzing the context of Hillary Clinton.

Repeating degree-of-separation searches over longer periods of time permits to measure changes in betweenness of the different “stars” to detect trends early.

Figure 6. Changes over time in betweenness of 14 presidential contenders (x-axis is time in days, y-axis is betweenness)

Figure 6 illustrates the changes in betweenneess of the 14 presidential contenders over 9 days in August 2006. As the blue line shows, non-competing candidate Al Gore’s lead is growing, while other leading democratic candidate’s John Edward’s fortunes are declining. The big winner of this week is candidate Russ Feingold, whose absolute betweennees and thus Web popularity is more than doubling in this time period. Leading republican candidate Rudi Giuliani is keeping his lead, in a neck-on-neck race with Al Gore. The overall centrality of the combined group analysis is slightly diminishing over the time period, indicating that there is no clear leader emerging thus far.

To see how incumbent but rather unpopular President George W. Bush fits in, figure 7 compares him on Sept 1, 2006, against his 2000 Presidential adversary, Al Gore, but also with his Vice President, Dick Cheney.

Figure 7. George W. Bush’s popularity Sept, 1, 2006.

As figure 7 illustrates, President George W. Bush generates more buzz on the Web than his Vice President Dick Cheney, as was to be expected. His 2000 competitor, former Vice President Al Gore, however, would easily beat both of them, at least on the Web.