From Predictive to Automatic

Predictive analytics is about making statements about the future, and then enabling either an individual or a group of people to make decisions. The idea is that better decisions will be made if people have a better idea of what to expect in the future.

I’m not quite as optimistic in the whole relationship between better information and better decisions. The intervening variable, the person, or worse, the group of people, are make poor choices in the face of good futures. They have their own preferences and expectations. Judgement is certainly a trait that is distributed normally in the population. Besides, Arrow’s Theorem.

That’s where automatic decision systems can help.

Remember the idea of Website Morphing? You can read the full paper here. Hauser, Urban, Liberali, and Braun won an award for that paper. Basically, the put forward a model by which a website would adjust in response to the user. Think of a personalization engine combined with a recommender engine. It’s that awesome.

The design of such systems requires a fairly sophisticated digital team. They have to know what a design pattern is. They have to understand responsive design. And, they have to have a degree of trust with their own users.

We’re starting to get there. I’ve been impressed with the direction that Parse.ly has been going in lately.

It’s out there, and not so far off.

***

I’m Christopher Berry.
Follow me @cjpberry
I blog at christopherberry.ca