Shared publicly  - 
 
A mathematical explanation of why corporations almost necessarily must become stupid and incompetent as they grow. See also j.mp/hehgooglers cc +John Baez 
7
3
James Salsman's profile photoRyan OGrady's profile photovictor steinbok's profile photoDavid Guild's profile photo
25 comments
 
Read thru it.  Guy's got a point.

I can't help but figure Peter's Principle is a huge nasty wrench in those same works he analyzes, too.  Makes the 95% probably more like 25%.  On a good day with the wind in the right direction.
 
I like that this is effectively the business version of well loved childhood game Chinese Whispers.
 
+James Salsman I don't find this to be particularly sophisticated analysis, but I am not interested in expending energy and time to explain why.  But, thank you for the cc!
 
This reminds me of a Lewis Carroll piece which I doubt if I could dig up quickly... (edit: the important bit is excerpted here http://tinyurl.com/cttadx4 starting at the end of the previous page, p.54. ...though my memory has it as "Ho for Ginny!" not "...Jenny!")

More seriously, while I think there is some truth to this, it strikes me as a major oversimplification of corporate communication lines at best.

From my (limited) experience with corporate communications and its problems, you pretty much never have the supreme executive reviewing every piece of information and making decisions which are passed down through subordinates. That would be too obviously idiotic even for management-types.

What you do have at any given level is delegation of various areas of authority to various individuals who are then evaluated over time (i.e. in aggregate, rather than per-decision) on the basis of (a) their own reports and (b) less subjective evaluations of their performance, such as production or sales figures.

In other words, the key communications loops are generally not long chains, and there are some corrective mechanisms in place.

In the one place I worked where there were serious stupidity problems, those problems were actually exacerbated by too much delegation -- there was one guy who had all the puzzle pieces needed to solve a communication problem between the IT people and the production people, and he just sat there waiting for us to solve the problem on our own.

(He later got "kicked upstairs", thus confirming the Peter Principle... actually, I think he was promoted above his level of incompetence, since he had clearly displayed incompetence prior to the promotion).

I'm sure we can improve vastly on the corporate-hierarchy management methodology, but I don't think that starting with an unrealistic model of how it works is going to be of much help in identifying its weaknesses.
 
I have a different, perhaps simpler explanation. In a dynamic model, even minor perturbations can knock you into a different trajectory. When companies are small and tightly controlled, it is possible to catch this shift early and take corrective actions -- they don't always work, but they are worth trying. In a larger corporation, the situation is more chaotic. There are more trajectory variations and determining the direction of correction is much harder. At the same time, because of the overall complexity, what fails on one end may well be compensated on another -- it's more about how different parts integrate together than about any individual part. Just think of Borders, Circuit City and CompUSA as exemplars of small decisions having a significant effect on the outcome. If you compare the general policies of these giants with their surviving competitors, the differences were few and appeared fairly minor at the time. But some of these decisions (e.g., Circuit City's firing of their most experienced employees) ended up having a profound impact on the company's future. There was simply not enough corrective power left in management to change the trajectory.
 
Thanks for pointing this out, +James Salsman.  The blog article is interesting, but what we really need are models of corporate behavior that people then go out and test experimentally.  I have no idea what the state of the art on that is!  It may seem 'too complicated to study scientifically', but people in biology study incredibly complicated systems.  There should be scientific journals about the failure modes of corporations, because these affect us enormously.  But I don't know if there are! 
 
I don't really buy the initial argument that corporations are stupid and incompetent.  They very nearly rule the world, you know.  They are doing something more efficiently than anyone else.
 
+Sterling Wright Since when are rulers of the world required to be competent and intelligent? Haven't we had a share of stupid and/or incompetent rulers fairly recently? Besides, it's just the choice of words. Someone else might choose to say "inertial"... or something else.
 
 Hi +victor steinbok.  "Ruling the world" was a bit tongue-in-cheek on my part.  But, I do argue that there are few (if any) organizations of any substantial impact that can beat the production and distribution efficiency of a for-profit global corporation producing and delivering millions of product Xs all around the world, and managing to make a profit while doing so!  "Stupid and incompetent" are certainly not adjectives I would attribute to them.

As far as governments....they are certainly inefficient, as are all bureaucracies.  Nobody is going to argue against that claim.  But, this truth may be why so many people think government should be as small as possible.
 
Governments are just a particular type of corporations. I probably would have said "single-track", "inertial", or "nonresponsive" rather than "stupid" or "incompetent", but the point is the same. A large system is more likely to be chaotic and partially counterproductive than a small one. Calling it "stupid" or "incompetent" implies that the negative decisions are deliberate, which may not be the case. Think of  ant parable -- everyone assumes that ants are highly cooperative creatures, managing to get large useful objects (e.g., food) into the nest. The reality is quite different -- if you throw a piece of cheese on the ground and let ants start carrying it, it looks like they are deliberately carrying it toward the nest. If you cut the piece into smaller pieces while they are doing that, the illusion is destroyed -- each piece will be dragged in a different direction. The reason for dominant direction being the correct one is simple probability -- more ants are coming from the nest than from other directions, so they are more likely to pull in the same direction. Ants coming from other directions also pull back in the direction from which they came. In reality, the process is highly inefficient -- if ants were cooperative, the objects would have moved much faster in the correct direction. Does that make ants "stupid" or "incompetent"? Not really. But it does not make them any more efficient either.
 
Governments are not a particular type of corporation.  They are a legal entity, formed by the people. 
 
Democratic governments are optionally reconstituted periodically, causing them to have additional immunity against the failure mode due to yes men.
 
And this is why we are having so many problems these days, when people don't even know what governments are.  That allows governments to be coopted by other entities such as corporations or obscenely wealthy/powerful individuals.

No, corporations are not people.  Good grief.

Governments are compacts made (and financed for that matter) by the people to protect and extend their rights.  They exist for the sole purpose of serving their citizens' needs.  You do remember the "WE THE PEOPLE" preamble, yes?  THey are organized through a constitution

Corporations are (at present -- there have been other definitions, and will likely be again) for the purpose of making money, generally by filling some sort of demand.  They have no such constraints -- except the basic legal sort, which they routinely try to circumvent -- in promoting the general welfare.

The early American government was in fact so hostile to companies (because of contemporary ones such as East India Company) that it limited corporations to a very specific maximum life span and to very specific purposes (such as building a specified structure), after which it was disbanded.
 
(To be more precise/head off complaints, I am using "government" as short hand for democratic or even republican government, as opposed to a monarchy, oligarchy, plutarchy, fascism, etc.)
 
"They very nearly rule the world, you know.  They are doing something more efficiently than anyone else."

They have cut out all that inefficient compassion and social benefit stuff.

The point is valid, though; a cheater who is consistently stupid still won't get ahead... and as I said, the model seems fundamentally inaccurate to me.
 
"...it limited [each corporation] to a very specific maximum life span and to very specific purposes (such as building a specified structure), after which it was disbanded."

Can we point this out to the Tea Partiers the next time they talk about respecting the wishes of the Founders? This sounds like a tradition worth returning to. (Oh dear, I'm advocating a return to tradition. Does this make me a conservative?)
 
I think that 13 levels of management is proof of incompetence. Not surprisingly, A implies A.
 
+John Baez we might be able to find some natural experiments. For example, it should not be too dificult to determine the different proportions of people inside and outside of an organization who believe the organization is blundering and/or malicious as the size of the organization changes. Also, there is a strong implication from the first paragraph of www.jstor.org/discover/10.2307/27800836?uid=3739568&uid=2&uid=4&uid=3739256&sid=56300894943 that heiarchical governance often fails.
 
One of the simpler features of mathematical biology is 'scaling laws': for example, how the ratio of height to weight of a land mammal changes as the animal gets larger.  There could be some scaling laws that govern, e.g., the fraction of workers in a company that belong to 'management', or the amount of time a typical worker spends on email, as a function of the number of workers.  Some of these things would be fairly easy to do experiments on.  Probably someone already has.

Here's a blurb on scaling laws in biology:

http://www.sciencedaily.com/releases/1999/02/990202072253.htm

just to give the flavor.  There are much more substantial things to read out there, but you can probably imagine some analogies to corporations.
Add a comment...