Division or Unification of Labor

I heard a fairly convincing two-part argument against Smith's principle of the division of labor.   I use the qualifier of "fairly" because I don't think that it entirely disproves the theory, nor does it replace it, but it does cast some doubt as to whether it is universally applicable.

Division of Labor

As a refresher on the original principle: Smith suggested that dividing a task into a number of simpler tasks, and training workers to perform only a small step in an overall process, enabled considerably greater production with considerably less skilled hands.  It's the theory that was best demonstrated by assembly-line production, and the astounding level of output that could be achieved.

Smith's example was that of a pin factory: each worker performs a simple task: one draws the wire, another cuts it, another straightens it, another grinds the head, and another sharpens the point.  And by this method, a small group of illiterate farmhands could, with very little training, far outproduce an equal number of master blacksmiths.

Strictly speaking, this was not the origin of the division of labor.  Even before Smith's theory was put into practice, separate tradesmen mined the ore, smelted the metal, and fashioned it into useful objects.   Blacksmiths also specialized according to the kinds of objects they were most skilled in fashioning, and there were separate "smiths" for various kinds of metal.    So in essence, Smith was merely observing an existing phenomenon, but he took it to the extreme.

First Rebuttal: Divisibility

The first rebuttal to Smith's theory draws on an example that seems patently ludicrous - but which seems no less extreme than the example proposed by Smith: that of typing a letter by using fifty or so workers, each of whom uses a machine that has a single key: one person types a lowercase "e", another an uppercase "T", another a comma, another the spacebar, etc.

This follows the very principle of the division of labor, and illustrates the dysfunction of taking it to extremes.  The amount of time to load and position the paper, strike the key, remove the paper, and transport it to the next workstation overwhelms any benefit of specialization, and requires a tremendous amount of coordination to move it from one worker to the next in proper order.

Plainly, the task can be performed most efficiently by a single skilled worker than a team of unskilled workers, and it makes little sense (and much waste) to subdivide it.  As such, the unification of the tasks involved into one skilled typist makes far more sense than the division of labor among many unskilled hands.

And to that point, the unification of labor is taken further by the elimination of the typist - as it is presently common for the author of a letter to do his own typing, and likely far more efficient, as the speed at which keys can be struck is of relatively minor effect.

Second Rebuttal: Skilled Labor

A second rebuttal, more philosophical than functional, is that Smith's theory was created at a time when skilled labor was in short supply: it was the perfect solution to the problem of manufacturing when there was not a ready supply of skilled blacksmiths and illiterate laborers had to be trained to perform tasks in a short amount of time (five years of apprenticeship was not feasible).

But in the present day, most workers are at least semi-literate and considerably more intellectually sophisticated than their eighteenth-century counterparts.  So it lo longer seems necessary or desirable to break down a task into functions that are so simple that they could be performed by a trained chimp or replaced by a mechanical device that repeatedly performs the same sequence of basic motions.

It's also worth considering that the manufacturing of goods has become a relatively minor part of the present economy.   Most workers in the post-industrial world are involved in what is generally called "knowledge work," in which the intellectual component is of far greater importance in the generation of value than the manual tasks involved in producing physical artifacts.

Many of the most highly-compensated professions produce no physical artifact at all: a doctor, an attorney, and an executive do not (directly) produce any physical object.  And neither is this work adaptable to the division of labor beyond a certain degree.   It can be argued that the work of a physician has been divided into specialized forms of medicine (podiatrists, oncologists, cardiologists, etc.) and that some of the tasks are handed off to labs or clinics (blood work, X-rays, etc.), but there remains the need for a general practitioner to manage the care of a given patient in a holistic manner, calling in specialists when it is beneficial.

In more abstract terms, an intelligent person is capable of performing more sophisticated tasks, and more tasks, and there is not a practical need to subdivide labor.  In addition to being counterproductive, it is patently unnecessary in a culture in which there is ready availability of an educated workforce.

Reconciling the Extremes

Each of these arguments takes an extreme position - the complete division of a process into minute tasks, or its complete unification onto a single workbench.  As in many things, my sense is that neither extreme is right, but represent the ends of a continuum along which an intelligent choice must be made.

When a task is too divided, there arises a need for administration and control, and the number of errors as a result of miscommunication and misdirection rise.  I vaguely recall an article that suggested it can be assessed by the ratio of non-productive employees who supervise and audit work as compared to the number of employees who are directly productive.   Where that ratio is less than 1-to10, the overhead cost of administration likely exceeds the efficiencies of division of labor.   Seems a bit arbitrary, but likely a good indicator that consideration is necessary.

When a task is too unified, productivity decreases.  I don't expect the argument of an educated workforce holds much weight here: I don't think education would have been a factor in Smith's pin-factory example.  The most educated and skilled blacksmith would still be capable of producing far fewer pins than an assembly-line of uneducated workers.  I can recall no guidance on this issue, but it seems to me that it has much to do with task-switching, and perhaps the same ratio of 1-to10 could be used to assess whether the nonproductive time of a given worker (switching from one task to the next) as compared to the productive time (performing a given task) likely exceeds the efficiency that can be gained by specialization, when considering both the time to move work from one person to the next as well as the overhead costs of coordination and control.

In terms of managing knowledge-workers, the educational factor likely comes into play.   While some individuals claim to be able to perform all tasks related to a given act of production with expert precision, it is very seldom true.   I've found this to be the case in print and Web production, where individuals arrogantly claim to be experts at every facet - only to find that they are very good at one or two things, marginally competent in a few others, and woefully inadequate at the rest.  There does seem to be a finite number of things a person can do very well, and when a task requires expertise in too many dimensions, unification of labor produces a low-quality and low-volume result.

As such, it seems necessary to determine the point between the extremes for any given task or process: division or unification of labor to the logical extreme is not a panacea for all forms of work - but instead, the idiosyncratic nature of the task, and the idiosyncratic capabilities of the individuals who perform it, must be intelligently assessed to arrive at a point where productivity, cost, and quality have achieved an optimum balance.