For various reasons I have recently found myself
doing some focussed reading on an idea that the IT community knows as Conway's
law. The idea is this:
“Any organization that designs a system (defined
broadly) will produce a design whose structure is a copy of the organization's
communication structure.”[1]
But Conway’s law itself
is not the subject of this post. This post is about another observation that I
made along the way and for which Conway’s law makes a handy example. The observation is, that amongst IT professionals, Conway's law is
one of a number of ideas that are generally accepted as truths, statements of
something that will inevitably happen. Using Conway's law as our example, assuming
that it is true that an organisation and its systems will inevitably align, leads
to a whole range of other ideas around whether or not alignment is a factor in
success. My quick review of the academic
literature reveals that we really cannot say for sure. What we can say is that
Conway's law is not a law, nor is it a theory, it’s a hypothesis.
Laws, theories and hypotheses
Let’s tackle the difference first; a hypothesis
is an educated guess supported by an explanation of why the guess might be right.
Hypotheses can be tested (usually quantitatively) to see whether they can be
proven correct or not. In the case of Conway’s law a number of researchers have
proposed tests of the idea and reported some results, some of which say this
can happen [2, 3].
 |
| http://evolution.berkeley.edu/evolibrary/images/teach/parachute.gif |
A theory provides us with both a mechanism for
and an explanation of something we observe. We may develop a theory as a result
of testing one or more hypotheses and if as a result the hypothesis
holds true we may have a solid theory which we can then use to predict why and
how something happens in a particular set of circumstances.
A law is something that is true in all cases
that meet its criteria, laws are supported by strong empirical evidence; laws can
be explained by theories and laws generally do not change when new theories are
developed. There aren’t actually many things outside of hard science that are
laws, although there are some things we think of as laws in computer science, for
example, Moore’s law, which has generally proven correct up to now. There are however,
plenty of good theories that explain and predict that Moore’s law is about to
run out of steam, as well as lots of hypotheses as to why this might be true. What
is true though is that a hypothesis cannot ever become a theory and a theory
isn’t a law, nor can it ever become one!
Why does it matter?
It matters because Conway’s law is not a law, it
should not be relied upon to argue that something will always happen. Nor is it
a theory, it cannot be used to predict what will
happen or explain what has happened if an organisation does one thing or another. It’s an educated guess
that something might happen which says nothing about how it might happen or
whether it’s a good thing or not. Understanding this matters a lot if you are
about to make a business decision or buy something expensive.
What should we do?
We should stop calling things a law when they
are not. In the case of Conway’s law, the underlying idea is based on an older
hypothesis, better known as the mirroring hypothesis[4] and it comes from the field of organisational design [5]. It was first referred to as Conway’s law by Fred Brooks [6], who incidentally is often held responsible for the declaration
of another non-law; Brooks law. Even
Fred Brooks does not think Brooks law is a law, since there are a whole lot of
exceptions to it, but that’s a whole other post [7].
Instead, consider what would happen if we
stopped using the term Conway’s law and reframed our conversations to use the
term mirroring hypothesis instead. My hypothesis is that a whole lot of slide
decks might need to be re-written.
1. C. M, “How do Committees
Invent?,” 1968; http://www.melconway.com/research/committees.html.
2. S.E. Bailey, et al., “A Decade of
Conway's Law: A Literature Review from 2003-2012,” Proc. Replication in Empirical Software Engineering Research (RESER),
2013 3rd International Workshop on, 2013, pp. 1-14.
3. I. Kwan, et al., “Conway's Law Revisited:
The Evidence for a Task-Based Perspective,” Software,
IEEE, vol. 29, no. 1, 2012, pp. 90-93; DOI 10.1109/MS.2012.3.
4. L.B. Colfer, Carliss Y., “The Mirroring
Hypothesis: Theory, Evidence and Exceptions,” Book The Mirroring Hypothesis: Theory, Evidence and Exceptions,
Series The Mirroring Hypothesis: Theory, Evidence and Exceptions, ed., Editor
ed.^eds., Harvard Business School, 2010, pp.
5. A.S. Herbert, “The Architecture of
Complexity,” Book The Architecture of
Complexity, Series The Architecture of Complexity 106, ed., Editor
ed.^eds., 1962, pp.
6. F. Brooks, The Mythical Man Month, Addison-Wsley, 1975-1995.
7. S. McConnell, “Brooks' Law Repealed,” IEEE Software2014; http://www.stevemcconnell.com/ieeesoftware/eic08.htm.
