Crowdsourcing and Predictive Analytics turn out to be cousins
My colleague, Michael Westgate, did a very cool analytics review of the top bands that will perform at Austin City Limits in October, 2013. Via Michael’s work, we already know which bands are most likely to be the hits six months from now. We’ll keep tracking and refining our view, but we won’t do it in the more traditional way. We’ll use our own algorithm to judge how each band is picking up influence across 107 metrics in ten channels of online communications.
The algorithm enables us to judge real overall influence and to do it over a long period of time.
And that is where the rather obvious point has hit me that crowdsourcing and predictive analytics are cousins and have been all along.
One of the common mistakes made to judge behavior is to study how a person travels online from site to site. Somehow, we think, if we can see where they are going on a single journey, we can judge what they like. That’s a bit of a crapshoot. It’s what you would expect of a linear-focused advertising world geared to get you to buy something.
One individual’s journey is just that……one individual’s journey.
Now, if we follow, let’s say, a million of this person’s friends, making a similar journey, I would argue we could start to see patterns that are meaningful. For Austin City Limits, we are and will be looking at millions of data points over time.
The crowd is predicting who the leaders are. We don’t need to guess. As summer festivals start, it will become even more clear. Since ACL is the last of the summer festivals, we’ll know exactly where to spend our time. And yes, we all go to ACL so we really do like that benefit!
What is even more intriguing is that we can validate our data by dropping a lens over summer festivals pre-ACL. We can then see what folks are saying publicly, via mobile phone, by band. This technology, from our partner, SnapTrends, shows activity by band at last year’s ACL.
The crowd always knows what’s next. The trick, however, is following the right crowd. Austin City Limits is easy. But if we want to know which features of a computer will drive its sales a year from now, its more sophisticated and, as a result, far more important that we get this right. One of the best examples I’ve heard on selecting the right crowd relates to the space shuttle. Ask all of the residents of Florida how much fuel is in the space shuttle and they have no clue. Ask 300 rocket scientists and they will get within an ounce. Tracking the right people matters.
It is one of the reasons why we are building a software and a services firm together. We need to innovate how we monitor the cloud, so we can make the right decisions on the ground.
Meanwhile, out of Michael’s top 20, my personal favorites so far are Eric Church and Atoms for Peace (RadioHead’s Thom Yorke, Red Hot Chili Pepper’s Flea and Beck’s Joey Warnocker). My only problem is Michael’s algorithm doesn’t really care what I think…:)
Check out his post to see more of the top 20.
All the best, Bob