Can Google Predict the Future by Measuring Emoticons in Data Traffic?

Last week Google’s Gmail team last week announced support for displaying emoticons in their Gmail service.

Personally I couldn’t care less, I guess this feature was demanded by users with poor mental rotation skills. Don’t get me wrong, I use emoticons a lot in emails and I see the value, but I don’t really care whether they’re comprised of characters or an image.

But I digress.

The announcement revived a wacky idea I had years ago. I discussed it in a pub with friends and forgot about it again. But it’s a neat idea and here it is: Google has the means of measuring the number of smileys and other emoticons people enter and use in content created in the ever-ballooning Google galaxy of services (Gmail, Chat, Blogger, etc). We’re talking about millions of emails per day, hundreds of thousands (or maybe, millions) of Blogger articles per day, tons of chat messages (nobody uses Google chat, hence the unit “tons” is probably correct). I wouldn’t be surprised if Google could detect 10-20 million emoticons per day.

What could Google do with all this data?

• Present the aggregated data in a special version of their Zeitgeist.
• Create a “How’s The World Feeling At This Moment” indicator.
• Create a weekly forecast (“Better stay in bed next Monday, it’s super chagrin day”).
• Short sell on the Stock Exchange (now that would be “forward looking”!).
• Promote products via AdWords that make consumers happier if the “happiness trend” goes down.

Another example: it would be super interesting to see whether there was as spike in the number of :-( faces before the stock market crash that began around September 16, 2008. My take here is that tens of thousands of financial people knew something bad was going to happen, only the rest of the world didn’t yet realize it. If one could tap into the emails sent between the financial centers of the world (and I’m sure Google could do this if they wanted this) you would be able to focus the measurements on certain industries. Obviously, this would be evil so Google is not doing this.

As with any statistical data set it is important to correlate the data with other available data sets. In this case it would be interesting to see if spikes in Google’s data have any correlation with the data these fruitcakes are collecting.

As I said, a wacky idea. But an interesting one nonetheless.