Good Design Requires Conceptual Integrity by Jon Pittman

The POV Dispatch is our Autodesk internal newsletter, published monthly, where we discuss the big ideas that are important to us and our customers. It is published by our Corporate Strategy & Engagement (CS&E) team of which Autodesk Labs is a part. Jon Pittman is the VP of CS&E, so it should come as no surprise that Jon routinely makes submissions to issues of the POV Dispatch. Jon is also a Lecturer at the Haas School of Business of the University of California at Berkeley. Jon contributed this to a recent issue, and I thought I would share it with you.

---

What is good design?

Most agree that good design has value in that it is more functional, easier to use, and more pleasing to users — ultimately creating economic value but one challenging question is "what is good design?" This seems like a simple question. Most people would probably say they know good design when they see it, but find it difficult to define. Robert Pirsig had a similar difficulty when he tried to define "quality" in one of my favorite books — Zen and the Art of Motorcycle Maintenance¹. When he tried to define quality, Pirsig encountered a dilemma — with two seemingly conflicting views of quality. One view was what he called a classic view — in which quality of something is defined by a clear breakdown of a whole into components with clear and logical relationships among the components. Further, the classic view of quality is parsimonious — everything exists for a purpose and there is no wasteful overlap, duplication, or redundancy. Classic quality is very much a product of the logical mind.

He defined romantic quality, on the other hand, as about the whole or the gestalt of something. It is more about the feeling you get from something at a very visceral level. In contrast to the classic view, the romantic view is very subjective and very much a product of the emotions. Pirsig resolved his dilemma by asserting that one cannot understand quality sufficiently through either the classic or romantic view — both are required.

So what does this have to do with good design? In some sense good design and high quality are similar. Without getting into debates about "intelligent design," the only real distinction I can think of has to do with human intention. Natural systems and constructs — think of a tree, an organism, or a landscape — can have high quality but are not typically designed by people (yet). Good design has all of the characteristics of high quality, but someone made design decisions intentionally. Thus, good quality and good design are identical — with the stipulation that good design is the result of an intentional human act.

Here are three other characteristics of good design:

• Good design tells a story. Something that is well designed embodies and communicates a strong point of view. It does not try to be all things to all people. The entire design reinforces the story and omits things that detract from that story. If a design communicates a coherent story it elicits a great user experience. This is the essence of the romantic part of good design.

• Good design balances form, function, and performance. Good design looks good, does the job well for its user, and does it in a sufficiently robust and efficient way. America's most famous architect, Frank Lloyd Wright, said that "form and function are one," that they are inseparable. This is the essence of the classical view of good design.

• Good design has conceptual integrity. All of the parts of a design relate to each other and relate to the whole. There is nothing unnecessary and all of the parts have an internal consistency that contributes to the point of view. This is the magical part of a good design that brings together the classic and romantic views.

Why is it important?

Aside from the fact that Autodesk is a design company and ought to exemplify good design in our products, we are at a particularly critical juncture where good design is even more important to our success. As we move to a new cloud/mobile/social platform, one of the consequences is that users will be likely to consume our software-as-a-service in proportion to the success they have in using our software. This is welcome news for customers and us. If customers are successful in using our software — meaning that it serves their purpose, they will continue to use it and may well want to use more. If they are not successful, they will stop using it — or use less of it — and our revenue from those customers will decline. Thus user success becomes an imperative for Autodesk business success — and good design is one way to ensure shared success.

One design characteristic that leads to user success is software that is "easy to use." Ease of use is yet another elusive concept. Fortunately, Don Norman, one of the early pioneers in human-computer interaction, brings some insight to this problem. In his book, The Psychology of Everyday Things² , Norman asserts that something is easy to use when it is consistent with and reinforces the user's conceptual model of the thing. In other words, we all have a mental model of how the software works. If the software is consistent with that mental model, we think it is easy to use. If it is not consistent with that mental model, we find it hard to use. The conceptual model of a piece of software is, in fact, the story that the software reinforces. We need to intentionally design and build a logically consistent user model — that is the essence of conceptual integrity.

The very nature of the cloud/mobile/social platform requires good design and conceptual integrity. Our desktop applications such as AutoCAD, Inventor, Revit, and Maya are all large, monolithic applications. Traditionally our users tended to focus on just one application, thus, they could learn the "story" or conceptual model behind that one application and be productive. With suites, more and more users are using multiple applications. The new model of the cloud enables us break our software into smaller modules; we could build each with its own story or conceptual model, but that would make our overall portfolio much more difficult to learn and use. We really need one story for our whole system. That means the entire system needs to have conceptual integrity and be well-designed. To achieve this, it needs to be designed as a whole, rather than piecemeal.

How does one achieve this conceptual integrity?

Given a large and sprawling product — and development organization — how do we achieve conceptual integrity for a whole system? Fred Brooks, author of software classic, The Mythical Man Month³ (and architect of the IBM 360 operating system — one of the first successful large-scale software systems — was a big believer in conceptual integrity. Brooks asserted that conceptual integrity only comes from "one mind, or a very small number of resonating minds". He very much felt that design by committee was a road to failure. Brooks advocated putting in place system architects who were the keepers of conceptual integrity in a system and made sure that the only things that made it into the system were consistent with the conceptual model.

So how does that work in a large organization? We do have some clues from the design profession. When designing at urban scale, someone is responsible for the overall urban design, and the individual buildings, public spaces, and transportation systems are designed by others. There is a cascading of responsibility, and ultimately it is everybody's responsibility to ensure the conceptual integrity of their piece and its relationship to the whole. Ultimately the system needs to be designed with a strong point of view and a consistent set of principles as to what it should be?

But how does that approach apply in the age of the crowd? Isn't crowdsourcing an example of designing without top-down direction? Yes and no. While we once thought that an earlier version of crowdsourcing, open source software would yield innovative designs, one astute observer⁴ of open source has noted that it produced mostly robust versions of very prosaic and well-solved problems — operating systems, compilers, web servers, web browsers, and databases. It has not produced much innovation; however, it has produced robust software with high conceptual integrity. That is because, even with a crowdsourced ethos, there are gatekeepers who ensure that new submissions maintain conceptual integrity. Furthermore, since most of these systems are well known, there are rules and standards about how one does things. That also helps create and reinforce conceptual integrity.

We have seen that good design can be defined, and is critical to our business success — now more than ever. We have also seen that it requires conceptual integrity and that must come from one mind or a very small number of resonating minds. As we transform Autodesk's products and business, the need for good design have moved from a nice to have to a must have.

Footnotes

1. Wikipedia entry for Zen and the Art of Motorcycle Maintenance
2. review of the Psychology of Everyday Things
3. review of The Mythical Man-Month
4. news room for our own CEO, Carl Bass

---

Thanks Jon.

Autodesk Labs provides a vehicle for various divisions to share technologies as part of integrating them into the overall experience. Your feedback helps determine if these plug-ins or services maintain the conceptual integrity Autodesk is going for. Your overall experience shapes the future of our integrated technologies.

Integrity is alive in the lab.