the laws of managerial motion

Chandler as physicist

Once more, I find myself writing something because I want to refer back to it in the piece I really want to write – I guess this is a better problem to have than being out of ideas.  This week, I attempt to reformulate the thesis of “Strategy and Structure” by Alfred D Chandler, in the form of a few cybernetic “laws of motion”.

Chandler’s book was written in 1962, before “management cybernetics” had really become a thing (Stafford Beer only left United Steel in 1961).  And he was fundamentally a historian rather than an economist, operations researcher or theorist of any kind.  But nonetheless, I would make a case that Chandler rather than Beer could be considered the “father of management cybernetics” (a title given to Beer by Norbert Wiener in a book blurb).  Chandler also, in my opinion, ought to be seen as the starting point for management science itself – this would be good for the soul of management science as, unlike FW Taylor’s time-and-motion studies or Elton Mayo’s Hawthorne Experiment, “Strategy and Structure” isn’t a massive case of research fraud.

What it is, is a compendious study of internal memos at DuPont, General Motors, Sears Roebuck and Standard Oil, looking at how they developed and how they thought about their internal organisation.  He then expanded on the historical lessons with another book called “The Visible Hand” on corporate management.  This is my interpretation of his view, rewritten into terms broadly consistent with information theory.

Zeroeth Law: (This is actually W Ross Ashby’s “Principle of Requisite Variety”).  For a system to be stable, the regulatory and management system must be able to process at least as much information as the operations generate.

Lots of technical stuff hiding in the background to make this rigorous, obvs, but I won’t go into that here because the unrigorous statement is, I think, reasonably easy to understand and quite intuitive.  Management is decision-making, which is a cognitive operation, and so for something to be “manageable”, it has to not generate decisions at a faster rate than the capacity of the management system to carry out those cognitive operations.

Importantly, this means that technological advances which improve the cognitive bandwidth that management can bring to bear (yes, computers, but also things like a good accounting system) will increase the size or complexity of the thing it can manage.  Or perhaps more importantly, things that decrease that bandwidth (rapid staff turnover, the distracting effect of a corporate merger) might make a previously manageable system temporarily or permanently unmanageable.

First Law: The complexity and cognitive demand generated by a healthy company grows over time

As things get bigger, they get more complicated.  And things, in general, get bigger.  Sometimes, they might get more complicated and bigger in their operations even as the revenue shrinks.  This is something of a generalisation, and won’t be true at absolutely every moment, but as a broad brush description of large enterprises from 1945 to today, I think we should just assert it and admit there are exceptions.

Second Law: Adding resources to an existing management structure is subject to diminishing returns

Lots of reasons why this might be the case – I find myself wanting to digress again on Fred Brooks’ “Mythical Man-Month” and the fact that when you add people to a team, the number of communication issues scales multiplicatively while the cognitive power only grows linearly.  But Chandler found this to be empirically true and in my view there’s no real reason to search for reasons – practically everything does have diminishing returns in this naughty old world.  It’s entirely possible (and maybe not even all that rare) for marginal additional management resource to make a negative contribution to overall cognitive capacity.

Third Law: Cognitive supply and demand are brought back into balance by reorganisation.

This, I feel will be the controversial one, because reorganisation itself imposes a significant demand on bandwidth.  But we should think of this as an investment, because the kind of reorganisation that Chandler is talking about is one which reduces the demand on the central management function in the same way that Fred Brooks characterises as the only way to get a late software project back on schedule – to accept that you are going to do less. 

So the big change in corporate structure that happened during the period of the memos that Chandler studied was the development of the “M-Type” multidivisional corporation.  This pushed budgeting and detailed control to the operations, reducing the information coming in to the top management to a set of defined reports and making the system manageable again.  Later on, after the period studied, a lot of big M-Type corporations found that they were becoming overwhelmed once more, and so they split up into smaller companies and abandoned the task of managing them as a single unit – “deconglomeration”.

The basic idea of the industrial economy you get from this kind of consideration is that companies will grow (first law), and will tend to get more inefficiently managed as they do so (second law).  Eventually some kind of crisis will happen when it becomes clear that the management problems cannot be solved by adding more resources to them and (third law) a change in structure will happen.  As an abstract description of the world, I think it fits a lot. 

It also suggests that technological change, particularly in communication and the organisation of information, will greatly influence the speed and timing of how things happen – it might put off an inevitable crisis, or it might trigger one.  That’s what I really wanted to write about, and hopefully will do so later this week.