Monday, August 27, 2012

Generalists versus specialists

#entarch Following a series of blogposts from @tetradian under the general heading "No Jobs for Generalists", @ruthmalan chips in to ask whether the labels "generalist" and "specialist" represent useful stereotypes, or whether they merely reflect another one of those "petty antagonisms".

Tom Graves, No Jobs for Generalists (August 2012)
Ruth Malan, A Trace in the Sand (August 2012)

In places, Tom's description does imply a dichotomy between generalist and specialist, which could well lead to antagonism between job roles (where "antagonism" probably means mutual incomprehension and discomfort and maybe schismogenesis rather than outright hostility). Elsewhere he talks about "specialist-generalists", but I think it may be more correct to think of the relationship between generalist and specialist as a dialectic one.
How exactly does the specialist/generalist dialectic work? I found an interesting framework by Gillian Stamp, in a book by Elliott Jaques and others. Levels of Abstraction in Logic and Human Action (1978).

Mode One: Proceduralists or Pragmatic Specialists (competent, persistent, attention to detail)

Mode Two: Practitioners or Pragmatic Generalists (good at organizing both their own work and that of others)

Mode Three: System Setters or Theoretical Generalists (good at gathering and organizing quantities of information, good planning ability)

Mode Four: Structuralists or Theoretical Specialists (intellectually very able, subtle, creative and very self-contained in their work)

Mode Five: Originators (poor at routine work, usually taking an original approach to a problem even when this may not be appropriate)

In this framework, generalists are not at the highest level of abstraction, but sit in the middle.

The popular EA frameworks (such as TOGAF) seem designed to support Mode Three. But there may be many enterprise architects whose personal inclination may be more towards Mode Four or Five; and it may be awkward for them to pretend to be in Mode Three in order to sell EA services to managers who are in Mode One and Two.

This might go some way to explaining some of the difficulties addressed in Tom's blog.

In a Jaquesian management hierarchy, higher levels of management (and higher status) is associated with higher cognitive ability, which Jaques associated with levels of abstraction as well as longer time horizon. Superficially, this seems to provide a theoretical justification for the old civil service idea that generalists (people with university degrees in ancient Greek) should manage specialists (people with university degrees in science and engineering) - although the old civil service way of implementing this idea reflected a rather perverse interpretation of the generalist/specialist dichotomy, since it is unlikely that one's abstract cognitive ability depends solely on the subjects studied at university.

Within large organizations, people often aspire to more general roles than they currently occupy. There are two distinct reasons for this. One is that people may develop more abstract capability as they mature; the other is that the more general roles are perceived as carrying higher status. Companies often designate selected employees as "high-fliers", and move them rapidly around to prevent them becoming bogged down in any particular specialism. However, it has often been observed that people drift upward until they reach a level at which they are no longer competent. (This is known as the Peter Principle.)

In a Jaquesian hierarchy, the management function needs to be at a higher level of abstraction than the function being managed. New employees and external consultants are generally subject to closer levels of supervision than long-serving employees, whose actual work (defacto job) is often much more interesting and varied than the official job description. From an HR perspective, it makes perfect sense to recruit specialists and allow deserving employees (if they wish) to grow into the more sprawling (and ill-defined) generalist roles. And from an outsourcing perspective, it makes perfect sense to subcontract the (more easily specified) specialist work and retain a smaller inhouse staff of generalists to make sure it all fits together.

But the idea that any old generalist can perform an abstract architectural role - especially architectural control over a complex procurement environment - seems troubling. Is Mode Three good enough?


Tom Graves said...

Many thanks, Richard - lots of good points here, in fact too many to answer in a single reply-comment! :-)

The Stamp/Jaques framework looks useful, though we need to be wary about using it as a hierarchy: they're correct in saying that these are modes, not levels, though I can see that those who are a bit too fond of hierarchies might misuse the framework in that way - as you do warn above.

In my own case, yes, I'd probably migrate naturally towards Mode Five, though in my own field of metamethodologies and the like I'm actually pretty solid in Mode One - 'competent, persistent, attention to detail' - and to be honest I'm probably weakest in Mode Two or Mode Three (particularly around 'organising work of others' and 'planning ability').

The key point, really, is that none of the Modes is 'good enough' on its own: we need an appropriate balance of all of them in order to get the work of the enterprise properly done, especially over the longer term. The Modes can (and do) occur in just one person, and in principle are available in every person. In a larger organisation, though, it makes sense for people to do work that more closely aligns with their preferred Mode(s) - which is where distinctions between 'specialist' versus 'generalist' tend naturally to arise.

In answer to one of your earlier points, I don't see specialist versus generalist as inherently antagonistic in any way: they're just different forms of focus, different ways of working. The only time that they become directly antagonistic is when those differences are bundled together with some arbitrary and often delusionally-egotistical 'hierarchy', where one type of Mode is purported to be 'above' and 'more important' than another. And although each Mode has its own distinct disciplines, there does need to be strong discipline throughout: hence, yes, I would strongly agree that "the idea that any old generalist can perform an abstract architectural role - especially architectural control over a complex procurement environment - seems troubling".

Richard Veryard said...

Thanks Tom

Jaques would agree that the Modes are not themselves hierarchical levels, but suggests that the number of modes increases at each level: thus supervisors should master Mode One only, and top management should master all five modes.

A Jaquesian hierarchy is by no means arbitrary, and is not necessarily delusional or even egotistical, but probably has some other features that you would disagree with.

Jaques's theories are a bit unfashionable nowadays, and I disagree with many of his conclusions. But he is a significant thinker with a clearly articulated position on organizational design, and because I always strive to explore ideas that challenge my current beliefs and practices, I think it is worth the effort to understand his attempt to define the perfect hierarchy.

Meanwhile, since few if any organizations have a management hierarchy that comes anywhere near satisfying the Jaquesian criteria, this at least gives us a lens for identifying specific causes of management dysfunction within existing organizations.

What is of particular interest to me here is how any of this applies to the architectural management and governance of complex systems of systems, especially where the day-to-day control is spread across different companies and continents. What is the right level of abstraction for effective governance, and what are the implications of this for the "job" of architect?