Book chapter summary - The Nature and Art of Workmanship, chapters 1-4, by David Pye

What I read

In the first four chapters of this book, author David Pye describes the nature of workmanship, differentiating it from other similar words and ideas, and juxtaposes work of risk with work of certainty.

In the first chapter, Pye speaks of the distinction between design and workmanship. The difference, for the author, is clear: “Design is what, for practical purposes, can be conveyed in words and by drawing: workmanship is what, for practical purposes, can not.” Pye indicates that the workmen have an equal, or perhaps even more important role than the designer, who creates an idealized goal that may not actually be attainable. Pye discusses that in popular culture, material is given as equal standing as design (and more than that of workmen), but this is also not giving accurate recognition to good workmanship. He describes that mass production has led to products where workmanship is excellent, but is “dismally restricted.” The “workmanship of a motor-car is something to be marveled at, but a street full of parker cars is jejune and depressing.”

In chapter two, Pye introduces a primary frame that is used for the remainder of the text: a workmanship of risk, as compared to a workmanship of certainty. He notes that he will try to remove the word craftsmanship, which is a “word to start an argument with.” But it gestures to the quality of the result of the making activity, where the workmanship is based heavily on the abilities of the maker. This is contrasted with the workmanship of certainty, which is always present in mass production and automation. Pye illustrates the two through the example of writing with a pen, as compared to printing.

Printing initially requires the “workmanship of certainty,” in that it is largely a manual process first in establishing the machinery to do the working; this leads to the mass produced output second. Using a typewriter is intermediate in its risk of use, but certain in its lack of risk of output: “the N’s will never look like U’s.” Workmanship of risk, present in most trades, is typically not visible in its output, because a workman uses things like jigs or tools to reduce the risk.

Workmanship of risk has historically (sometimes) been valued. Sometimes, it produces things that are objectively bad and more expensive than a certain workmanship, and Pye indicates that “it is fairly certain that the workmanship of risk will seldom or never again be used for producing things in quantity.” But he also notes that there is, and will continue to be, demand for things that, as a result of the workmanship of risk, are individual.

In chapter three, Pye works to introduce meaningful definitions for guiding a discussion of workmanship, and starts with the action of making as compared to the output of making. Drilling is used as a preliminary example, comparing a dentist drilling (with a mass produced object) and a man drilling a piece of wood; both are drilling by hand. But the dentist is a workmanship of almost 100% risk, given the context of the action, not the actual object being used. Pye explores this distinction by taking a more and more focused view of the phrase “made by hand.” Discussing the amount of by hand activity starts to work down a linguistic rabbit hole; is it “partly by hand,” “singly by hand,” “in small numbers,” “guided by tools”?

Pye lands at a definition-like view related more to the context in which something is made and less to the making. He compares a “water-and-wood complex,” a “coal-and-iron complex,” and a modern “electricity-and-alloy complex”, where each are found in certain eras of time.

In chapter four, Pye discusses quality, again with a focus on both defining and framing the ideas theoretically, and showing their practicality through usage. He starts by looking at the words that are commonly used (mostly for workmanship of risk), such as bad, good, precise and rough.

Good and bad presents and is judged by soundness and comeliness. The first is the degree through which the designer’s technical specifications were achieved. The second is the degree of which the aesthetic specifications were achieved. Workmanship fulfilment of design is an approximation, and there is a level of appropriate acceptance to the approximation that is often driven by contextual (historic and cultural) expectations of the output. Pye describes that there are no perfectly straight lines, but the straight-ness that is acceptable are related to the expectations of risk and certainty. Perfect, as an idea, is a viewer or consumer’s comparison of output to the expectation of a very straight line, although that is not always the intention based on the designer’s goal and the workman’s execution. “Rough” may be an intention. Similarly, a designer may create a specification of perfect (in the use of straightness) without understanding the risk and certainty of the workmanship, and the risk of workmanship may be high (rendering that specification unrealistic or set poorly.)

Pye describes that workmanship of certainty in mass production is “incapable of freedom,” and is more and more common. But, “There is no substitution for the aesthetic quality of [workmanship of risk] and the world will be poorer without it.”

Pye breaks his text to show examples that support the ideas of risk and certainty.

Pye then returns to discuss the language of design: designer and intended design. The designer “decides what information [the drawings and specifications] are to convey.” The intended design is the “ideally perfect and therefore unattainable embodiment of his intention.” This is because the workman interprets the design intention, and even with the best interpretation, the output cannot be a perfect match. Workmanship can be judged based on the intention of the designer and the context of the “degree of regulation” (the amount of risk or certainty). Good workmanship is contextual to the expectations, and is not uniformly consistent.

Next, Pye defines technique and technology. Technique is the knowledge of making, while technology is the scientific study of technique. “Workmanship is the application of technique to making… As opposed to design, workmanship is what for practical purposes the designer cannot give effective instructions about by drawings or words.” Pye indicates that he avoided the use of skill on purpose, as it’s not useful for the discussion. But, he describes that with certainty, the production operations are predetermined, and with risk, they are up to the workman’s care, judgment and dexterity. He also notes that he is using workman in a way to distinguish the maker from the designer, and to note that, if one is more important than the other, it is the workman rather than the designer who ultimately must interpret, and therefore is the actor with the most value to provide.

What I learned and what I think

This is the type of text I’ve been looking to read and absorb, and I’m glad Carl suggested it (and let me know that he didn’t actually read the previous mess, so he’s redeemed himself with this one.) The subject matter is practical, in that the results have real impact and bearing on the way creative making occurs. It’s also theoretical or contemplative, and focused on producing knowledge. One of the differences in this approach than in some of the others is that the theory is almost defensive, rather than trying to be purely objective, and through that defensiveness comes realism and applicability. Pye’s goal is not just to muse about an idea so that it can be further mused on later, and his pursuit is not just to generate new knowledge about the phenomenon of making things. He actually feels his profession is at risk, and so his intellectual argument has a pragmatic intent. I value this in the same way that I value a vocational school as much, or more, than a university; the pursuit of (or creation of) knowledge for knowledge sake has a long-term value to humanity, while the pursuit of a skill has real value for a real person. It’s localized and populist.

The relationship that Pye paints between designer and craftsman is unique, not one that I’ve seen before. A designer makes a plan that is a goal, but not a mandate. The craftsman aims to achieve the goal, but better understands the risk vs certainty in the material, the production capabilities, and the role of their dexterity and skill. If a designer’s plan is naïve of the risk and certainty, they’ve done a poor job of designing; it isn’t the maker’s “risk of fault” or “responsibility” to transform that beyond the space of risk. Early in the article, he identifies that referencing material isn’t helpful because it implies that the material holds control over the workman, but I’m not sure what the material is in a massed produced object; I don’t think he would claim that, in injection molding, the material is the plastic. He would focus on the materiality of whatever was necessary to make the injection molder in the first place: that’s where the workmanship occurs. This means that a designer is responsible for the noun, but the workman for the verb—the way things are made, instead of what is made. He needs to be an expert in injection molding as a production method, so that he can make a “jig” of an injection molder. The abstraction layer is always there—a hammer to bang on hot metal, with an understanding of the use of the tool being more important than the “nature” of the metal.

Maybe the writing doesn’t hold up in the face of more and more modernization. At what point is the discussion of the abstraction of the bespoke making so removed that it’s not useful? A developer made software to make Figma to make UI to make designs to make workmanship of making software. Forget the weird circularness; which part of this is really the “workman” and “designer” and “workmanship” in action?

One way to play it out is that the designer is the designer, and the workman is the developer. This is probably the most obvious way the disparity or disconnect that Pye has discussed plays out. The designer shows intention. They are naïve of the way the workmanship of development occurs, or the abilities of the workman; they design something with high risk, and it is produced poorly. The blame occurs as Pye describes:

“The designer is apt to imagine he has more control over workmanship than he has. Standards of workmanship become established in each kind of manufacture. The designer gets used to them, expects them, asks for them, and comes finally to believe is getting them because he asked for them. Then he comes up against a firm who do not know their work and finds he is helpless. All he can do is say ‘do it again.’”

Another way it may play out is that the design language system designer, or the component library designer, or the platform designer, is the designer, and the person moving components around in Figma (or “designing the interface”) is the workman. It’s the same problem again in practice, but with the poor workman. Pye doesn’t use the word skill, but that’s basically what this would be: the workman—the interface designer—has poor dexterity, knowledge, judgement (probably because of poor experience), and the team responsible for providing the components designed a specification that was unrealistic, or that led to workmanship of risk even though they thought they would be getting workmanship of certainty. They were certain, because they share the name designer, but the interface designer is really an interface workman. In this case, the “blame” of poor output is on the poorly designed language system, which was based on the expectations of a different level of experience of the workman.

There’s another way, where the product manager is the designer, and the workmanship is happening through the creation of the components, which are jigs on the way to the creation of the mass production of software.

This last idea is fundamentally what “magic” is missing in the creation of software right now. Designers (as designers) are specifying software with a want for and expectation of craftsmanship of certainty. Designers and developers (as craftsman) are then creating software “jigs” of components, controls, modules, templates, objects, in order to aim for that workmanship of certainty. The result is the loss of the “one-offness” of the workmanship of risk, where the risk has reward, and the reward is the nature of a beautiful, special output. Have we componentized ourselves to bland?

I will try to continue this one over the next few weeks, but the two weeks of Arizona Cocodona Race Crazy is probably going to mess up the consistency I’ve had so far.