must we bury the Renaissance Person?

Conventional wisdom has it that today the Renaissance person is no longer possible, because of the sheer volume of knowledge needed to master any single discipline.  Instead, we have the “Renaissance Team”, in which experts in each discipline work closely together. And so we have Bill Moggeridge’s “T-shaped people” – individuals who are experts in one domain (the vertical stroke of the T) and yet able to communicate with their expert peers in other domains (the horizontal stroke of the T).

The Renaissance Team idea has much to recommend it. For one thing, it’s not disruptive.  We can stick with our traditional form of higher education that focuses on a single discipline.  We merely teach collaboration to the single-domain experts (T-shaped people), so they can collaborate effectively with their T-shaped person peers.  It suits management, because capabilities are conveniently quantized: if one T-shaped person isn’t performing on the team, another one with the same domain expertise can be found to replace him.  It’s also an easy idea to understand.  Everyone already knows what it means to be an expert in their discipline; the Renaissance Team idea just emphasizes that people need to work together.

Design Thinking in Engineering Education

Recently, “design thinking” has achieved currency in higher education circles. I recently attended an NSF-supported workshop on graduate education in engineering design.  Engineers understand that design is an important skill, and after fifty years in which analysis has beat out synthesis for space in the curriculum, design is back.  It began some years ago.  The first moves were the freshman course introducing design through engaging engineering exercises such as the “emergency egg launch” (students work in teams to design a device that can throw an egg 20 meters without breaking it), and the senior capstone class (students work in teams to solve a design challenge in their discipline, usually for a real client).  These courses made their way into undergraduate curricula over the past two decades in response to the realization that engineering education was failing to teach design. Now there is a cry for more.

The goal of this new graduate education is to produce engineers who can design. One proposed approach (let’s call it the “domain first” approach ) would take graduates of a traditional four-year engineering program, and emphasize design in their Master (or PhD) education.  The domain-first approach conserves the traditional values of the domain. If we dilute the undergraduate curriculum by substituting a design course for a core discipline course, we risk producing dilettantes rather than domain experts.  And anyone who has tried to meddle with a core disciplinary curriculum knows the difficulty of removing anything.

The domain-first approach graduates professional master students who have all the knowledge and skills of the modern engineer, to which we add the magic ingredient of “design thinking”.   So, Master students might study design methods (e.g.,  brainstorming, user-needs analysis, requirements specification, and so on).  They might take a literature review course,  (e.g.,  Simon’s seminal article “Science of Design,” Rittel and Webber’s “Wicked Problems” article, various pieces by Nigel Cross).   And perhaps they take a project course in which students work in teams to address a real world design challenge posed by a corporate sponsor or a humanitarian need – much like the senior capstone course mentioned above.

Three year (graduate) design education

Yet it is possible to teach “design thinking” in three years.   Schools of Architecture take into their professional Master programs students who have completed undergraduate degrees in non-design disciplines—history, mathematics, Italian. They teach architectural design to non-designers in three years and produce practicing architects.

This model contradicts the “domain-first approach”.  It suggests, that to become an effective engineering designer one may not need to take a four-year undergraduate engineering degree before going on to learn “design thinking”.  If a French major can enroll in a 3-year professional architecture degree program to learn design, could she also become a engineer in a similarly structured engineering design program? Engineers will surely dispute this, on the principle that four years of undergraduate engineering education provide indispensable core knowledge.  Yet, is engineering more knowledge intense than architecture?  If architects can do it, why not engineers?

Yet a professional Master degree in architecture is quite different than today’s Master degrees in engineering. The three-year architecture program focuses on design—studio is the heart and soul of professional architecture education. All else exists to support the studio education. Studio is king: Architectural history, structural engineering, or digital media are all “support” courses. For engineering to successfully apply the model from architecture, design studio—or something close to it—would have to be the central activity of a three-year education.  This is easier said than done: because the professional norms and values in engineering are not focused on design the way they are in architecture.

I have argued so far that if we want to educate “design thinking” engineers, professional architectural education with the studio at its core, offers a model.  This model has been explained and discussed by others, notably in Donald Schön’s book The Design Studio, and to some extent also in Educating the Reflective Practitioner. The design studio—for all its virtues—has also some disadvantages of which those who seek to adopt it should beware.  Most obviously, the studio model is based around solo performance.  Although in some advanced studios students work in groups or teams, they mostly learn to design alone. Designing is a skill that one must first master individually—only after one has the basics does it makes sense to work as a member of a team.  Imagine, by analogy, that we started to learn a skill like reading or arithmetic through group project-based learning.  Learning the skills of design is similar—first master the skills alone; later they can be applied in teamwork.

Another disadvantage is that design studio is a costly enterprise— student-instructor ratios are high (architecture studios in North America cap at 1:15 with a skilled teaching assistant, and 1:12 is considered a more reasonable ratio.)  Studio courses typically make up 12 contact hours each week, and the instructor spends much of that time meeting one-on-one with students in “desk critique” sessions.  The studio also demands dedicated space for each student to set up a personal work area for the entire term.  Because these studio courses are demanding of faculty time and energy, instructors are not expected to meet the expectations—fundamental in the sciences, engineering, and humanities—of a research career.  In many instances, studios are taught by adjunct faculty members who maintain a professional practice outside the university.

Must we really bury the Renaissance person?  I think not.

designing considered domain-independent

At lunch I sat next to C*— a well known roboticist. We began to chat.  He said, “You’re an architect, aren’t you?”  I said, “No, not really; but I am a professor in the School of Architecture. My main interest is design.”  He asked me, “you mean building design, or graphic design, or what?”   And I replied, “Actually, I’m more interested in the process of designing, in general, than in design of any particular kind of thing.  I think there’s something to be learned about how design and designing works, that’s more or less independent of the specific domain — buildings, electronic circuits, whatever.”   “Foo!” retorted C*, “you can teach everything there is to know about design in one semester; the rest is domain-specific.  And actually, design is nothing more than what used to be called ‘proper thinking’”. Dismayed at this turn of the conversation, and not knowing what to say next, I turned back to my pudding.

Now, C*— is known to be a smart fellow, so I don’t reject what he thinks out of hand.  Moreover, as a roboticist, he has quite a lot of experience designing, engineering, and building robots, which as we know, are fairly sophisticated machines.  So his opinions about design are worth taking into consideration, even if I think at the outset that he’s wrong. But I do think he’s wrong.

It’s certainly true that to be a good designer of anything—rocket engines, hydroelectric dams, water purifiers, software, elections, posters, or poems—you must know a lot about your subject.  And surely, you can’t expect a rocket engine designer to design a computer unless they take the time to learn about computers.  In any serious design domain, you need a lot of specific knowledge about the class of thing you are designing.  But I would also argue that designers share a common expertise.

Most people are not naturally designers; good designers learn a set of skills and ways of working.  Architects spend years to acquire those skills as they also learn domain-specific knowledge about making buildings.  What makes studying architecture (or any of the explicitly labeled “design” disciplines like industrial design, product design, graphic design) different from studying engineering is that you learn both the specifics of the domain as you practice the general skills of designing.  So by the end of the three or four of five years that an architect studies in school, she’s not only mastered the domain-specific business of assembling materials on a site, planning for people’s activities and movement, accounting for weather and climate, heating and ventilating, and so on.  She’s also had three to four years (8-10 studio courses) of continual practice designing.

This is remarkably different, by the way, than how we teach engineers.  For most of the past sixty years, until the 1980s, engineering was taught almost entirely as an analytical discipline.  Engineers learned the laws of nature as they applied to the specific domain they were studying, and how to analyze the behavior of specific artifacts: the electrical and magnetic performance of a circuit; the kinematics of a mechanism; and so on.  As a consequence, when they went out into the world to practice engineering, they had to learn on the job how to apply their theoretical and analytical knowledge to the business of actually making something.  Then, in the 1980s, under pressure from ABET (the accreditor for programs in applied science, computing, engineering, and technology), engineering departments began to put design back into their curricula.  So nowadays, they involve students in design exercises during their first year (for example, the popular egg-launching assignments that one sees everywhere on college campuses), and again in their senior year students take a “capstone design” course in which in groups on a design problem, often for a client.  So even with this newly design-infused curriculum engineering students spend time in at most two courses in their four-year degree programs actually doing any design.

What are the advantages of the intensive design education that architects and other kinds of designers get?  For one, they learn to look at whole problems, and to engage in task-setting behavior for themselves, rather than being given a tightly circumscribed problem and working within those bounds.  They learn to generate alternatives quickly, and they are less prone to the tunnel-vision syndrome in which a designer picks an approach and then develops it without looking at alternatives.   There are other things too, like the experience of making good use of criticism; an integral part of learning to design is being critiqued by others.  Those who survive learn to listen and to apply the critique in their work.

I’ve worked with both kinds of students, and I’ve found that it’s much easier to for someone who has studied design through architecture to learn the basics of mechanical engineering than for a mechanical engineer to learn to design. I’m not sure why that is; maybe it just takes longer than one semester.  Perhaps the domain lens is so powerful that once you learn to look at the world through it, it’s hard to see things any other way.

There are other reasons to think that there’s more to designing than common sense. Herb Simon devotes a chapter of his book “Sciences of the Artificial” to what he calls “the Science of Design”, and in it he describes a set of topics that make up this intellectual agenda.  Forty years have passed since he gave those Karl Taylor Compton lectures; still much remains to be done to realize the vision he articulated there.  And NSF recently decided that it might be smart to study “science of design of software intensive systems.”   But science of design and software engineering is another story.

meta-design pioneer N. John Habraken

N. John Habraken began developing methods for what is now called “meta-design” in the 1960s. Responding to the post-war need for housing in Western Europe, Habraken and his Eindhoven research group (the SAR) developed a method for designing housing that, as opposed to cookie-cutter mass-housing approaches, provided variability in floor plans and flexibility over time. His method distinguished the “support” or shared infrastructure parts of a design from the “infill” parts that could be changed freely. This distinction allowed designers to work independently on different “levels” of a design, and to evaluate the capacity of a design at one level to support or contain a variety of lower-level designs. A designer at one level provides a context for lower-level designers to make decisions, which ultimately determine the freedom that “end users” have to operate—in the case of housing, the freedom of inhabitants to arrange the furniture.

Habraken showed that organizing a design process this way produces designs that can more easily be changed over the building life-cycle—a relevant property for designs that must last decades or even centuries. (Stewart Brand’s popular book How Buildings Learn celebrates this principle). But Habraken went beyond theory:

To implement his ideas Habraken developed technical notation—a system of grids and charts—for documenting designs and design rules and procedures for systematically developing and testing designs formulated in this notation. He showed how, using his methods, designers could systematically explore a design space. This enables a designer to develop supports (meta-designs) that allow more infill variability and flexibility. He began with buildings, but soon expanded his scope to urban design.

Habraken’s design methods, first described as the “SAR method” in the 1960s later became known as “Open Building” because unlike most kit-of-parts modular coordination systems, the method does not restrict designers to a single closed system of components from a single manufacturer.

Habraken’s technical design method was predicated on observations of the built environment: the existence of dependency hierarchies. He recognized several hierarchies or orders in the built environment: Structure, Territory, and Supply. The order of structure or gravity deals with how the world is assembled (e.g., the foundation supports walls above, which support the floors). The order of territory deals with spatial containment (e.g., the city block contains the lot, which contains the building, which contains the apartment). The order of supply deals with the directional networks of pipes and wires that provide water, gas, and electricity. These three physical hierarchies are properties of the real world that in turn imply hierarchies of control in design. Therefore a method that acknowledges these hierarchies can be more efficient and produce better designs.

Habraken’s notation was adopted in the Dutch building code, and his design methods were employed in many developing countries for building housing. But Habraken’s ideas never made it in mainstream architecture. Perhaps his systematic methods and ways of talking about the built environment were too technical for the culture of professional architecture. And although he developed methods for designing built environments, Habraken was always interested in whether his observations might apply to design fields beyond architecture (such as VLSI) and he was always interested in cross-disciplinary campus conversations on design.

Habraken’s work on meta-design is relevant to collaborative design and participatory design, but it never gained a foothold in either of these fields. His method for collaboration in design was based on a hierarchical partitioning of concerns. During the 1980s, as the emerging field of computer-supported collaborative work (CSCW) gained momentum, this hierarchic way of working together by separating concerns did not fit mainstream CSCW. And although Habraken’s original motivation was to empower end users in the design of their housing, his systematic view also did not fit the prevailing tenets of participatory design.

why a phd in (computational) design?

As part of my professor job, I review many applications to Carnegie Mellon’s PhD programs in the School of Architecture, and I field a fleet of inquiries about pursuing doctoral work in design.  Here are some of the reasons people apply:

1. I already have a master degree and I want to take the next step.

2. Everyone in my family has advanced degrees, so I should get one too.

3. I can’t get a job, so I may as well get a PhD.

4. I want to teach in a university and a PhD is required.

5. I have a scholarship that will pay for a PhD.

These are all understandable reasons to pursue doctoral studies, but they’re also terrible reasons. Entering a PhD program for the wrong reasons is likely to lead to misery.  As one of my advisors (Aaron Fleisher) said to me years ago: “It’s easier to get in to a PhD program than to get out.”  Think about it. To get in, you must only write a few page statement-of-purpose that convinces an admissions committee that you have the capacity to do the work.   To get out, you spend several years doing research and write a lengthy dissertation.  Too many people find themselves midway through a PhD program when they realize that they don’t really like it, but at that point, they’re too far committed —emotionally, financially—to drop out.

The best—perhaps the only good—reason to pursue PhD study is because you’ve found something that you deeply, passionately want to understand and work on.  It’s an itch you must scratch.