Richard Feynman On The Folly Of Crafting Precise Definitions

One of my best friends in graduate school was a former physics major named Larry Ford.  When behavioral scientists started pushing for precise definitions of concepts like effectiveness and leadership, he would sometimes confuse them (even though Larry is a very precise thinker) by arguing "there is a negative relationship between precision and accuracy."   I just ran into a quote from the amazing Nobel winner Richard Feynman that makes a similar point in a lovely way:

"We can't define anything precisely. If we attempt to, we get into that paralysis of thought that comes to philosophers… one saying to the other: "you don't know what you are talking about!". The second one says: "what do you mean by talking? What do you mean by you? What do you mean by know?""

Feynman's quote reminded me of the opening pages of the 1958 classic "Organizations" by James March (quite possibly the most prestigious living organizational theorist, and certainly, one of the most charming academics on the planet) and Herbert Simon (another Nobel winner).  They open the book with a great quote that sometimes drives doctoral students and other scholars just crazy.  They kick-off by saying:

"This is a book about a theory of formal organizations.  It is easier, and probably more useful, to give examples of formal organizations than to define them."

After listing a bunch of examples of organizations including the Red Cross and New York State Highway Department, they note in words that would have pleased Feynman:

"But for the present purposes we need not trouble ourselves with the precise boundaries to be drawn around an organization or the exact distinction between an "organization" and a "non-organization."  We are dealing with empirical phenomena, and the world has an uncomfortable way of not permitting itself to be fitted into clean classifications." 

I must report, however, that for the second edition of the book, published over 20 years later, the authors elected to insert a short definition in the introduction: 

"Organizations are systems of coordinated action among individuals and groups whose preferences, information, interests, or knowledge differ." 

When I read this,  I find myself doing what Feynman complained about.  I think of things they left out: What about norms? What about emotions?  I think of situations where it might not apply: Doesn't a business owned and operated by one person count as an organization?  I think of the possible overemphasis on differences: What about all the times and ways that people and groups  in organizations have similar preferences, information, interests, and knowledge? Isn't that part of what an organization is as well?  I could go on and on.

I actually think it is a pretty good definition, but my bias is still that I like original approach, as they did such a nice job of arguing, essentially, that if they tried to get more precise, they would sacrifice accuracy. Nonetheless, I confess that I still love trying to define things and believe that trying to do so can help clarifying your thinking.  You could argue that while the outcome, in the end, will always be flawed and imprecise, the process is usually helpful and there are many times when it is useful pretend that you have a precise and accurate definition even if you don't (such as when you are developing metrics). 

Comments

[clk]

In many theories, it is easier to find examples or definition based on both extremes; be it right or left. However in reality, life like everything else is seldom at the extreme, and somewhere in between a continuum i.e. the grey areas rather than the black or white. Hence precision is somewhat elusive in reality as by definition, they tend to be either extreme.

Remember the former President's call "You are either with me or with the enemy". I tend to be in between 'unfortunately'.

[coaching employees]

Totally reminds me of Theory of Second Best. Great mind-tickling post. Thanks!

[Don C]

I think there are at least two underlying dimensions at play in the tension between examples and definitions.

One is individual preference or learning style.

The other is the amount shared experience between the people who are communicating (or at least attempting to do so ;-).

I recall reading something by Karl Weick in which, in my memory, he asserts that we are able to apprehend greater complexity if we don't force it to be explicitly stated. My inkingly is that he said in much more eloquently than my rendering from memory. And I think he was referring to an individual researcher.

I wholeheartedly agree that the process of moving between example and definition is most productive.

[Alan Fekete]

As pedagogy, explaining a concept by giving several examples is usually unsuccessful; the listener/reader has too many ways to develop a misunderstanding. It usually works much better to give both examples and counter-examples (examples that are not included in the concept); this helps the reader check their understanding.

[web design Landon]

I am sure that someone had the idea long before Feynman!.. I would argue that the definition of terms may often bring comfort to those of us who squirm under ambiguity, but sometimes what may seem like the beginning of wisdom is actually the start of a journey down a path that is wrong or overly narrow.

[Bob Sutton]

David,

Thanks! Your post is great and of course this yet another illustration of Sutton's Law, which we discuss in Hard Facts:

"If you think that you have a new idea, you are wrong. Someone probably already had it. This idea isn’t original either; I stole it from someone else."

http://bobsutton.typepad.com/my_weblog/2009/06/suttons-law-orginality-and-wisdom.html

So, of course, I am sure that someone had the idea long before Feynman!

[davidburkus]

I love this. I hate to be "that guy" but I wrote a guest post in a similar vein here: http://www.nevermindthemanager.com/2010/06/lost-in-definition/

As usual, Feynman makes a much better case in significantly fewer words. Thanks for sharing though. I'm working on making my post into a larger essay and will definitely be borrowing some of this. (With permission to poach of course).

[web design Landon]

I expect Simon would have appreciated the copious feedback required to distill a sprawling set of procedural descriptions into one concise declarative definition, and to note that that definition is synthetic, ..

[Larry Ford]

Gee, Bob, I don't have a specific recollection of having said that, but it does indeed sound like something I would say (and I am proud to have said it if for no other reason than that I confused some of our fellow students. This may be a related concept: "There is a negative relationship between wisdom and certainty." Cheers, Bob

Larry

p.s. Neither of these thoughts is original to me.

[Darbsnave]

It reminds me of "The General Theory of the Second Best" by Lipsey & Lancaster (http://www.jstor.org/pss/2296233). In certain circumstances, the second best solution is better than the best.

[Kevin McFarthing]

I agree wholeheartedly - precision can indeed be the enemy of accuracy. One way to distinguish when you're becoming overly precise is to apply the DIM test - "does it matter?". For most practical purposes you can stop whilst relatively imprecise, and move on.

Kevin
Innovation Fixer Ltd (UK)

[Dorian Taylor]

Perhaps March and Simon felt comfortable with the definition after letting their earlier framing sit for two decades. Or perhaps it took that long for the surrounding context to develop for the definition to make any sense.

I expect Simon would have appreciated the copious feedback required to distill a sprawling set of procedural descriptions into one concise declarative definition, and to note that that definition is synthetic, i.e. part of a model gelled for the purpose of formal discourse. Feynman likewise.

But then, the map is not the territory.

[Susanne Ramharter]

Reply to Bob's Comment - Alexander Pope on Critics in 1709: ".... for Fools rush in where Angels fear to tread."

This caustic line from 300 years ago sums up so much of what is being scientifically researched an proven today - see Ariely, Kahnemann and various other studies about Rationality. Daniel Kahneman's latest discussion about the 2 Systems of "thinking" says we believe that we are rational and analytic, but actually most of our reactions are based on what our associative memory or intuition tells us, and confirmation bias then lets us reinforce it with facts.

My point? as you say, ambiguity and uncertainty are hard to bear, our increasing need for speed as well as the "well, do something!" attitude demands immediate decisions and actions - how much better to base those on (supposedly) precise definitions than to state "I could be wrong, but let's try this..."

[Bob Sutton]

Arie and Don,

Thanks for the comments.

I agree that definitions are useful at times. But Larry's point to me was that there are times when the rush to precision is dangerous both because it can be too premature and too narrowing. I would argue that the definition of terms may often bring comfort to those of us who squirm under ambiguity, but sometimes what may seem like the beginning of wisdom is actually the start of a journey down a path that is wrong or overly narrow. There is value in being being specific and in being vague, and perhaps wisdom is partly about knowing when to do one versus the other. Of course, figuring out when to do one rather than the other is often mighty hard!

[Don C]

Mr. Ford's gambit must have been doubly befuddling if he refused to provide any definitions for precision and accuracy.

[Ariegoldshlager]

Bob,

Thanks for a very insightful note.

However the definition of terms is also the beginning of wisdom, or: "the beginning of wisdom is the definition of terms".

It appears to me therefore that the difference between "the beginning of wisdom" and "folly" is in the degree of precision.

Arie.
The Fine Balance Consulting Group