Sunday, May 8, 2011

Obliquity


Browsing at my local bookstore I picked up a copy of Obliquity by British economist John Kay (right), and was quickly taken in. It takes on a very profound subject with admirable brevity (200 pages). He notes the story of seeing three stone masons working on a medieval cathedral, and one says 'I am cutting stones to shape', one says 'I am building a great cathedral', and one says 'I am working for the glory of God.' These highlight that actions are guided by a hierarchy of objectives some short term, some long term, and some prove better guides than others.

Putting these various tactical and strategic objectives in harmony is tough work. Enjoying the journey and knowing you are on the right path removes the pressure and angst associated with an excessive focus on the end-goal, making that same goal more achievable. I think the key point he makes is to have all levels working for you, by adjusting all of them in light of feedback. An example is golf, a sport which for many is a 'game worth playing', making it worthy, and which benefits from practice and coaching, yet ultimately you can only swing well if you don't think when you are doing it.

Obliquity is his term for what someone called 'muddling through', and involves experiment and discovery in uncertain environments. Higher-level objectives, intermediate goals, and basic actions are constantly reassessed. The problem is that the big goals--a fulfilling life, maximizing shareholder value--are too imprecise to have any clear idea how to do them. As GE CEO Jack Welch noted, maximizing shareholder value is not a useful guide because it doesn't help you know what to do when you come to work.

It's a great book, though ultimately there is no clear answer, just another version of Feyerabend's 'Against Method': whatever works. Plans should be tentative and adjusted based on feedback from intermediate objectives, acknowledging our imperfect understanding of the solution, or even, the adequacy of our goal.

The problem is basically how to know when you should think outside the box, how to use common sense. To always think outside the box is to ignore the great toolkit of algorithms and methods smart people have given us so we do not have to reinvent, say, ordinary least squares or t-statistics. Common sense may be a straight jacket, but letting every wacky idea into your mind leads to insufficient focus. Quantification is imperfect, and can lead to useless reification as with macroeconomics, but its still the ultimate benchmark, and indeed, allows us to know that macroeconomics does not work. I know success, happiness, justice, are impossible to objectively and uniquely quantify, but I think modest attempts to provide proxies are helpful.

I wish it were easier to know how to solve important problems, and more importantly, know what problem are important and soluble. I do know there isn't a simple cookbook, and Jacques Barzun's observation serves a cautionary warning:
Again and again, the bright thought has occurred, "if we can only define our terms, if we can only find the basic unit, if we can spot the right 'indicators', we can then measure and reason flawlessly, we shall have created one more science."

This has usually been a failure in the social sciences, including business, in part because of Goodhart's law: targeting a variable changes its meaning and significance. Yet, as a full-time quant that's what I do--measure, model--and often I find that simple rules dominate the vague, obfuscatory, self-serving, and disingenuous descriptions many workers apply to their tasks. It's a paradox that makes my life interesting.

No comments:

Post a Comment