Michael Polanyi
Doubleday & Co., 1966
Reprinted Peter Smith, Gloucester, Massachusetts, 1983
ISBN 0844659991 (boards)
Buy it from Amazon.com
Buy it from Amazon.co.uk
This quiet little philosophy book, by a religiously-inclined research fellow of Merton College, Oxford who went to America to study further, back in the early 1960's, is a surprising thing to treat as a classic of modern requirements engineering. But such it undoubtedly is.
Referenced by many learned papers written by people who never read a page of his writings, Polanyi is now unquestionably the father of Tacit Knowledge - at least, if you discount the massive contribution of Wittgenstein, who is perhaps even less read and understood despite the clarity of his work.
The Tacit Dimension is a terse and clear exposition of Polanyi's thinking, developed over more than 20 years in lectures, books, and essays. The care pays off: the book stands alone admirably as an account of what tacit knowing actually means. Why it matters in our domain is, of course, another matter (see, for instance, Nonaka and Teece).
The nub of Polanyi's argument, and a much-quoted aphorism, is that, quite simply,
"we can know more than we can tell." (his italics)
He goes on:
"This fact seems obvious enough; but it is not easy to say exactly what it means. Take an example. We know a person's face, and can recognize it among a thousand, indeed among a million. Yet we usually canot tell how we recognize a face we know. So most of this knowledge cannot be put into words."
There in a nutshell is the key problem in requirements elicitation: people know, and even know that they know, but cannot readily tell us in detail what their knowledge is.
Fortunately Polanyi at once gives us some better news. In the case of facial recognition, the police have
'a large collection of pictures showing a variety of noses, mouths and other features. From these the witness selects the particulars of the face he knows... This may suggest that we can communicate, after all... provided we are given adequate means for expressing ourselves.
The situation, then, is that we possess knowledge we cannot tell, tacit knowledge, but that skilled elicitation enables us to talk about at least some of it and effectively share it when necessary. The good news is that requirements engineers shouldn't be out of work any time soon, as the task is difficult and requires skill. The tacitness is not destroyed by tricks such as the police's Photofit, as we cannot tell how we 'match the features we remember with those in the collection' - it is the same mystery expressed in a different form.
Polanyi reflects on the tacit dimension in various ways, and thereby derives an ontology of this strange way of knowing. In particular he shows that the knowledge is embodied. For example, when we use a tool of any kind to explore something, we very soon don't feel the tool, we feel the thing through the tool. For instance, if you take a pencil and tap various surfaces with it, you can quickly tell whether you are tapping something hard and hollow like an empty tin, or firm, like a desk, or yielding, like a pile of paper. The tool becomes an extension of your hand, and you interiorize its behaviour as a probe. To put it even more strongly, you start to 'indwell' or inhabit the tool, just as you inhabit your clothes and your body. How does the brain know anything about the world? Through the fingers, the skin, the ears, the other senses ... it suddenly becomes obvious that most of our knowledge is tacit, giving rise to most of the classical problems in philosophy. We no more know how we stand up or walk than we know how we strike a tennis ball with a racquet, ride a bicycle, play the violin, or design a program. We have a convincing illusion of being aware of all around us, while all the time we have practically no idea of how we actually operate, or what skill is. Perhaps the thing that most needs explanation is not tacit knowing at all, but the belief that knowledge is explicit.
After this introduction, you will not be surprised to hear that Plato comes into the story. In the Meno, Plato points out that it is contradictory to say that you can see a problem but that you don't know the solution. How do you know there's a problem if you don't know what it is?
"To see a problem that will lead to a great discovery is not just to see something hidden, but to see something of which the rest of humanity cannot have even an inkling."
Let me just say in a few words what Plato's argument is, in case you aren't familiar with it. Plato argues that you never discover anything, you just remember what you knew already, as when he questions a slave until the poor fellow 'remembers' the theorem of Pythagoras. This shows, says Plato, that the perfect form of all things is eternal, and we are born with complete knowledge of it.
"The solution which Plato offered for this paradox was that all discovery is a remembering of past lives. This explanation has hardly ever been accepted, but neither has any other solution been offered for avoiding the contradiction."
There is something delicious in realizing that Polanyi is cheerfully saying that all the philosophers have been wrong for 2000 years, simply because they didn't see that knowledge was often tacit. A scientist who has a tacit grasp of a problem may indeed be well on the way to a solution, but can't state it, yet, and certainly can't measure it. 'We can have a tacit foreknowledge of yet undiscovered things'.
The practical effect on handling requirements is profound. System users cannot tell us what they need in the new system, because they don't explicitly know: our job is to tease out that knowledge and have different people agree it explicitly. Interviewing is good at eliciting current problems (as Lauesen shows) but poor at identifying the shape of a desired solution. Scenario workshops, prototypes and demonstrations are better for that. Elicitation must address different types of tacit knowing, from deeply ingrained skill to simple facts that were too obvious to be worth stating. If you think requirements engineering is just a matter of documenting and tracing known requirements, you urgently need to study Polanyi.
This is a splendid book, short enough to read in a few sessions, but strong enough to change perceptions for a lifetime. Every requirements engineer should read and interiorize it.
© Ian Alexander, December 2001
You may also like:
