Copyright: 2008
Publisher: Houghton Mifflin
ISBN: 0-618-78460-8
Stephen Baker is a senior writer for Business Week magazine and a while back I read an article he wrote entitled "Managing by the Numbers". Turns out much of the material in that article was research for this excellent book, The Numerati. The book explores the world of mathematicians who are charged with parsing the growing databases of information that contain information about everything from what kind of cars we like to drive to which co-workers we are most likely to share the latest gossip with. Vast troves of rich information are only useful if there is someone who can make sense of it all and apply it to decision making processes. Bakers book is well written, easy to read and fascinating in its depth, breadth and clarity.
Modeling Humans
Samer Takriti works for IBM and his job is to model the IBM work force and come up with the best strategy for deploying their human resources. By factoring in available information (some information from Human Resources is off limits even for internal use) Takriti and his crew are developing a system where a manager can enter in some project specs and get back the Dream Team of IBMers to do the job. They model the workers by their internal social networks, their salary, their physical location and a host of other information. Someday they hope to include more personal information such as health information that could help determine whether say, a particular worker might not be very productive in the hot climate of Mexico City.
The idea seems a bit far-fetched until Baker takes us through a logical progression at IBM. Back when IBM was known for their hardware, they developed state-of-the-art supply chain management. They did this by reducing all the physical artifacts of the supply chain to numbers. So a manager could easily get on the computer and find out that a shipment of cicuit boards was sitting in customs waiting clearance so she might need to adjust her schedule.
Now though, IBM is a services company. The resources that they manager are human and not physical. So just like they reduced their hardware components to numbers in order to manage the supply chain, they are now working on turning their human resources into numbers that computers can parse using the algorithms that Takriti and company are designing.
Just like companies around the world hired IBM to help them figure out their supply chain management, IBM is hoping companies will use their "math" to manage their human resources as well. The possibilities are endless but as it turns out, modeling humans is a little tougher than modeling a shipping container of electrical components!
How Much is Too Much?
The basic problem with modeling humans is that... well humans like their privacy! We might not care if our company factors in our peanut allergy to its resource allocation model, but we might not appreciate them accounting for our children's poor performance in school as a possible detriment to our performance. Baker thoroughly explores the privacy issue and he does an incredibly fair job. A few years ago I read Database Nation : The Death of Privacy in the 21st Century in which the author looks at these same issues. I came away from that book feeling as though my privacy was being invaded maliciously. Baker takes a more balance view, pointing out that the use of personal data is a two-way street. For instance, if my kids are having problems in school, is now really a good time to go off to China for a 3 month project? In that case maybe it really would be a good thing if they had that information. Or maybe not!
The real question is: Where are we going to draw the line? That question is going to be the subject of much debate over the next few years as more and more companies gather more and more information about us. Inevitably the government is going to want to use some of that same data. They are already using some of this data in tracking down terrorists. Most of us probably don't mind that, but what if they start using personal information to keep an eye on say our religious practices in order to head off another Waco?
Shopping Demystified
You know those rewards cards that most of us have from the local grocery store? I have a Safeway one myself and I use it whenever I go there. It's nice because I get decent discounts sometimes and I also get a discount on gas when I purchase it from their gas station. Turns out those cards are creating immense databases of our collective buying behavior. Work is in progress to create "smart shopping carts" that can deliver personalized deals to us while we shop. By noticing say that I am pretty fickle in my brand choice of bread, but that I like whole wheat bread, they can send me a coupon to get me to try a new brand that they just started carrying.
A more sinister application might be to try to disuage me from shopping in their store if they consider me a "barnacle". A barnacle is a classification of shopper that mainly shops at a store to get the discounts. They are the kind of shopper that goes in the store, buys only things on heavy discount and then leave to go to the next store and do the same thing. So when a barnacle swipes his / her card on the smart cart, they would get ads for high priced olive oil or some other thing that would likely turn the shoppers cart around sooner.
The biggest drawback of the "smart cart" technology right now is that the data has not yet been sliced enough to provide outstanding value to both the store owner and the shopper. Unless both stakeholders find great value in using the technology, it won't have very high adoption rates.
Demographics and Tribes
Baker covers some fascinating ground when he gets into the retail world. He points out that in the old days we were classed into demographics. Using very broad strokes, we were painted as "18 to 35 while male". With the amount of data we have available now, more and more marketing is done by "tribe". So "late 30's males interested in computers and hiking" factors in not just my age and gender, but also my behaviors. With this knowledget it is possible to truly customize ad campaigns, be they political or merchandise.
Even with the sharper defined tribes and buckets, it is important to remember that the numerati as Baker refers to them, are not interested in "the truth". The truth about me is that while I am very interested in computers, I am not a geeky gadget guy. Most of the people in my tribe would probably respond favorably to a new ad about a new mobile phone. That's not an area of strong interest for me though. The key though is that marketers, politicians and employers can get better, quicker and cheaper answers to their questions even if it isn't the absolute truth.
Conclusion
This is a very fascinating book that covers a broad swath of knowledge. There is good depth to the book, but Baker keeps it from being a ponderous read with his engaging writing style and fascinating tidbits of information. I recommend this book to anyone who is interested in how companies are using information to make better decisions.