I've been fascinated by the ways that the Web can be used to predict real-world events, and it seems I'm not alone. Back in August, I created Google the Vote, an article on my Wiki that tracks the changes in the number of Google search results for phrases like "Barack Obama for President" and "John McCain is a leader" for the two major party's Presidential candidates. That's been fairly mixed, but in recent weeks has started to show a clear margin for Barack Obama (mirroring what the polls have been saying).
Meanwhile there are many other approaches. One of the most interesting
is intrade. Here's their stock tricker for their futures market on the
Presidential election:
By turning the election into a futures market, they are using the collective wisdom of their investors to predict the outcome, and because there's real money involved, "stuffing the ballot box" isn't as much of an issue. The outcome is at worst, a mirror of the popular news media, and at best a fairly short-term predictor of some elections.
Google Trends
I really love the idea of Google Trends, and over time, I'm sure it's going to improve, but right now it seems a bit broken. First, let me explain the idea. Google Trends is supposed to be a way of analyzing the frequency of searches or the popularity of terms on Web sites over time. This allows you to graph the popularity of a search term over time. Sounds great, right?
The problem is that some terms seem to decline over time in ways that don't reflect real-world changes in popularity. My suspicion is that this reflects a change in Google's user-base and even the makeup of the Internet community, but there's no way to confirm this without access to Google's raw data, which they guard jealously.
However, Google Trends does have some interesting data. Take a look, for example, at the search stats for "... for President" with each of the candidates name pasted in.
There's a lot to be learned from the Web and when it comes to social phenomenon like voting, I'm sure it has much more to teach us.
By turning the election into a futures market, they are using the collective wisdom of their investors to predict the outcome, and because there's real money involved, "stuffing the ballot box" isn't as much of an issue. The outcome is at worst, a mirror of the popular news media, and at best a fairly short-term predictor of some elections.
Google Trends
I really love the idea of Google Trends, and over time, I'm sure it's going to improve, but right now it seems a bit broken. First, let me explain the idea. Google Trends is supposed to be a way of analyzing the frequency of searches or the popularity of terms on Web sites over time. This allows you to graph the popularity of a search term over time. Sounds great, right?
The problem is that some terms seem to decline over time in ways that don't reflect real-world changes in popularity. My suspicion is that this reflects a change in Google's user-base and even the makeup of the Internet community, but there's no way to confirm this without access to Google's raw data, which they guard jealously.
However, Google Trends does have some interesting data. Take a look, for example, at the search stats for "... for President" with each of the candidates name pasted in.
There's a lot to be learned from the Web and when it comes to social phenomenon like voting, I'm sure it has much more to teach us.
Leave a comment