It’s official. Social Media can predict elections. A revolution is coming.
Since the first second that social media came into existence, people have used social media to question the power of social media.
This week we launched an interactive social scraping tool, built by our digital agency WebNarrative, that we claimed could predict the outcome of the US Election.
It was featured on the Daily Telegraph website under the headline “Can Social Media Predict the US Election?
The answer is yes. Yes it can. With the right algorithm, the right tools and the right expertise it can.
After analysing over a million pieces of information our tool predicted that the re-elected President Barack Obama would pick up 303 votes and Mitt Romney would end up with 235.
How it actually ended was Obama took 303 votes and Romney 206. Our tool also predicted who would win which States and we currently have 50 out of 50 correct with Florida still left in play. If all Florida’s 29 votes go to Romney, we will have all predictions perfect.
What the tool can also offer, however, through Twitter Cross Crawling Indexing is the ability to display the social comments that have powered the predictions. So, if you click on one of the States it brings up the top comments – effectively offering the context to the decisions the algorithm has made.
As you can see, if you look at Ohio on the 6th of November where Obama was campaigning the top tweets (including one about Jay-Z) tell an interesting story about the mood of the voters with the wind very obviously in Obama’s sails.
Whereas, the comments from Florida – where both Romney and Paul Ryan, his nominated Vice President were campaigning, are much flatter.
The tool looks at various social interaction channels including Facebook, Twitter and social betting platforms. The system will rank messages based on the social traction that each one gains, along with cross referencing each message’s keywords to find the commonality between them.
It takes these different sources and lays them of top of each other, with the algorithm working out probability of how that social noise and interaction will translate to votes in real time.
What I also found really interesting, though, were the comments below the story on the Telegraph about whether Social can be used to predict anything.
As you can see it’s a divisive subject with the sceptics being particularly vitriolic (some to the point of stupidity) about the role social media plays, both in our private lives and the wider world in general.
However, in an industry that is becoming increasingly besotted with measurement, algorithms and predictions we believe this is pretty amazing and a glimpse at the future of social as a valid prediction tool.
Imagine, years into the future where this tool, or something similar, has become so accurate and displayed such a short margin of error in other important Elections, the effect it would have on candidate’s campaigns.
Would a candidate for President/Prime Minister be able to sit back and relax two days before the actual vote safe in the knowledge he has an insurmountable lead?
Maybe that’s a little far-fetched, but imagine having real-time information about voter’s thoughts in important swing States/Boroughs and being able to change campaign plans accordingly in a matter of minutes and see the results of those changes in a matter of seconds…..
Now think of that kind of responsiveness in other sectors and industries. Ads that can be changed between ad breaks, the positioning of news stories in the news cycle, policy making decisions that don’t need Quangos and focus groups.
With the right thinking, the right passion, the right idea, the right talent this technology is ready here and now and I predict a revolution in how look at, use and interact with social.