Saturday, October 1, 2011

Effective Communication

This article appears in slightly different form in the Sept/Oct 2011 issue of IEEE Micro © 2011 IEEE.

Trees, Maps, and Theorems: Effective communication for rational minds by Jean-luc Doumont (Principiae, Kraainem Belgium, 2009, 190pp, ISBN 978-90-813677-07,, $96.00) 

If you can read only one book on technical communication, pick this one. In a magnificent illustration of his own methods, the author lays out basic principles, then applies them to the main types of communication that we all struggle with: documents and oral presentations. He also applies the same principles to producing graphic displays, procedures, meeting reports, scientific posters, websites, and email. 

Jean-luc Doumont is a professional speaker -- in English, French, Dutch, and Spanish. He has a degree from the Louvain School of Engineering in Belgium and a PhD in applied physics from Stanford University. He wrote this book in English. 

I first met Doumont when we were fellow speakers on a lunchtime panel at the 2006 IEEE Microwave Theory and Techniques Society Symposium in San Francisco. We invited engineers to bring their lunches into a large presentation room at the Moscone Center and listen to our advice about making effective presentations. I don't know how much the microwave engineers learned, but I was impressed by Jean-luc's professional approach to the project. In the months before the event, he raised many issues about our presentations and the room's logistics. Now, having read his book, I understand what lay behind those issues. His theories and practices have evolved from his long experience as a speaker.


Doumont believes that effective communication entails optimization under constraints. We want our audiences to get as much as possible from our communications, but we can't control everything. This leads to his three laws of communication: 
  • Adapt to your audience, because you can't usually control its composition. 
  • Maximize the signal-to-noise ratio, so others can receive your message with a minimum of loss. 
  • Use effective redundancy, to compensate for losses you can't control. 
Adapting to your audience is the most important of these rules and the hardest to follow. The audience members are constrained by language differences, their background knowledge, and even the amount of time they have to spend. You must anticipate their needs and expectations, and you must structure your arguments along their lines of reasoning. If they fail to understand, then you fail. If your first efforts don't succeed, you must try, try again. 

Maximizing the signal-to-noise ratio is tricky because it goes against some of our instincts. Everything matters. What does not help detracts; words, gestures, typos, flashy graphics, the keys in a speaker's pocket -- all qualify as noise. Question the relevance of anything you plan to present. Emphasize by suppressing, not by adding. 

Effective redundancy arises from simultaneously providing the same message in different media. Doumont's exemplar of effective redundancy is the stop sign. The octagonal shape, the red color, the position in the driver's field of view, and the word STOP all combine to reinforce a single message. Many speakers, however, make the mistake of providing different messages simultaneously. I have seen highly intelligent people project paragraphs of dense text onto a screen and say, "You can read that for yourself," then begin immediately to discuss the topic in different words. Doumont makes the point that if you simultaneously speak and project words on a screen, the text in the two media should be the same. 

Doumont explores the role of visual elements in communication. To do so he generalizes the distinction between pictures and words; he creates a taxonomy of communication types. The taxonomy has three axes: verbal/non-verbal, rational/intuitive, and sequential/global (that is, serial/parallel). A string of text on a screen is verbal, rational, and sequential. An audience, seeing that text and hearing you repeat it, can simultaneously process a picture on the screen. They might receive a different, even contradictory, message from the picture. This is bad for communication, though it might be effective as irony, or for sublimnal persuasion. We have highly developed filters for evaluating text, so it is hard for someone to mislead us with words. We have far fewer filters for images, which we process intuitively. 

Unfiltered intuitive processing can mean that a picture of what not to do can fail to have the desired effect. A graph with a widely spread out axis can make a small difference look deceptively large. This extends to other non-verbal cues. Your clothing, stance, voice, and movements as a speaker can affect your credibility. Doumont's bottom line is that every non-verbal message that does not reinforce your verbal message is noise. You should eliminate it. 

In 1956 the cognitive psychologist George Miller published a paper that led to the notion that our short term memories can accommodate between five and nine items. This has given rise to the widely promulgated rule of thumb that says that lists, procedures, and so forth should be limited to six items. Doumont uses a different numerical analysis to arrive at a smaller number. He also notes that humans can process balanced hierarchies more easily than linear lists. He gives guidelines for the width and depth of such hierarchies. I enjoyed reading Doumont's numerical theories, but I trust his conclusions, which are based on experience, more than his explanations for why they are true. 

The book's title describes Doumont's basic approach to structuring information. Trees are another name for the balanced hierarchies discussed earlier. Reasoning that we handle recursion with even more difficulty than lists, Doumont prefers breadth of structure to depth. His ideal document might have five chapters, each with five sections, and each of those with five subsections. He recommends various combining strategies to limit depth and breadth. 

The term maps refers to the forms of navigation you provide to your audience. Doumont believes that readers can navigate an effective structure effortlessly if you make it visible to them. For example, in a book structured as in the previous paragraph, you might provide a global map, and at each level tell readers what the divisions at that level contain. A running header or footer helps readers know where they are, and cross-references tell them where they can go. Oral presentations and websites need similar navigational aids, but each of these has its own problems. Without help, you can easily get lost on a website. During an oral presentation, you're stuck on a single track, but if your mind wanders, you need help to find your way back quickly. 

The reference to theorems is a shorthand way to say that the busy professionals who might read your book or listen to your presentation want to know in advance what you intend to prove. Then they can decide whether they're interested in hearing the proof. This models the way mathematicians present theorems. It is, of course, not the way they discover them. Many presenters want to lead readers along the path they followed to reach their conclusions. They want to save the punch line for the end. For almost all papers or presentations, that is the wrong approach. It will frustrate your audience and potentially waste their time. They wish to be informed, not taught. 


Having laid out a set of principles, Doumont applies them to developing many forms of communication. The main ones, of course, are written documents and oral presentations. For each of these he defines a process for applying his principles systematically to create a communication in that medium. For written documents, his process is deceptively simple: plan, design, draft, format, revise. 

Planning starts with asking why, who, what, when, and where. The last two are about logistics, and you can go badly wrong if you ignore them. The first three are a shorthand for determining the purpose of the document, the audience, and the necessary content. The purpose is what the audience should do (or be able to do) after reading the document. Doumont defines a two dimensional breakdown of the audience: specialist/non-specialist and primary/secondary. The primary audience consists of those in the To field of the email, those attending a talk, or those reading a research report when it first comes out. Members of the secondary audience are in the Cc field, hear a recording of the talk, or look up something in the report months later. Each quadrant has needs, and in some cases they conflict. The trick is to provide something for everyone without annoying anyone. 

Designing the document addresses content and its order of presentation. Using the theorem model, Doumont shows how to give readers the information in a useful and effective manner. For documents of varying sizes, he shows how to reformat the author's chronology into a nested structure of situation, problem/solution, and work done. He then shows how to present this structure using the traditional components: abstract, foreword, summary, introduction, body, and conclusion. He uses the idea of fractals to convey the idea of a document whose structure at each level reproduces its global structure. This can be carried to extremes, but Doumont remains practical. 

Doumont provides detailed advice about drafting and revising, but it is hard to summarize, so you'll have to read it for yourself. A main principle underlying these phases is to strive for clarity, accuracy, and conciseness. Don't simplify the ideas, but convey complexity simply. 

Formatting is about structure, not aesthetics. Doumont's main example of formatting is his own book. Each two-page spread has four columns of equal width. Doumont tells readers:
The book was designed to propose a logical flow for the discussion while enabling selective reading of individual parts, chapters, or sections. Feel free therefore to read the complete discussion linearly or to jump ahead to the themes of interest to you. Topics are discussed in one double page each time (or in a small integer number of them), to facilitate their direct access or out-of-sequence processing.
The pages, too, are formatted for selective reading. The right page is reserved for the main discussion, with illustrations, limited examples, or comments placed left of the text. In relation to this discussion, the left page answers frequently asked questions collected at the occasion of lectures and workshops, set on a gray background. In the remaining space it lists typical shortcomings, offers practical advice on specific subtopics, or broadens the discussion.
The resulting book has a clean, regular design with a large amount of white space. You rarely have to turn the page while reading about a topic. 


Doumont's specialty is public speaking. Like his book, his public appearances provide excellent examples of his methods. If you have a chance to attend one of his talks, don't miss it. This book explains many of his methods, but seeing the methods in action is doubly powerful. 

Doumont boils oral presentations down to this:
[Engage] the audience with mind and body, conveying well-structured messages with sincerity, confidence, and, yes, passion. 
As with documents, he has a procedure for achieving this end. The steps are plan, design, create slides, rehearse, present, answer questions. The planning phase is similar to that for a written document, but the design phase must confront the specific challenges of oral presentations. 

Oral presentations are synchronous. The speaker defines the sequence and rhythm, and the audience can't skip around. They need visual cues and clear transitions. A strong opening grabs their attention and helps them see what's coming. A strong closing tells them that they can relax and clap, but also leaves them clearly aware of what you want them to do next. 

Audiences cannot follow detailed evidence, so you should concentrate on a convincing delivery and leave the details to handouts or references. A convincing delivery relies on a good rapport between speaker and audience. Don't try to say too much. Stick to one main message. Decide on that message early in your planning phase, and state the message early in the talk. 

Creating slides is a place where many presenters fall down. The slides provide visual cues to where you are, and they convey the same message that your spoken text conveys. In fact, Doumont says the spoken text and the slides should each stand alone. A person who cannot see should be able to follow the spoken presentation, and a person who has difficulty hearing or understanding your spoken words should be able to understand the presentation from the slides. Slides are potentially helpful extras, but bad slides can substantially detract from a presentation. 

Speakers who create bad slides often do so because they are creating the slides for themselves, not for the audience. Dense, cryptic text can help speakers remember what to say, but it detracts from the talk as readers struggle to understand the slides. Some speakers think their slides should double as a written report. What they produce serves neither purpose. Other speakers, in a hurry, copy paragraphs from written documents onto slides. Such slides are worse than none at all. 

Delivering a presentation is a performance in real time. You can practice parts of it in advance -- what to say, how to say it, how to stand or move, and when to change slides. It is important not to have to struggle with those aspects of the presentation, so you can focus on receiving and reacting to the cues your audience sends you. At the same time, you should reduce the noise caused by clothing, jewelry, name tags, or excessive movement. 

Doumont has much more to say, and it's all worth studying, but you'll have to read the book. I hope you do.

Friday, April 1, 2011


This article appears in slightly different form in the March/April 2011 issue of IEEE Micro © 2011 IEEE.

This time I look at two books that delve deeply into technology and our relationship to it. One, by Brian Arthur, is concerned with how technology works and how it interacts with economics. The other, by Kevin Kelly, tells a poetic, almost mythic story of where technology came from and where it's headed. Their stories overlap and reinforce one another. I highly recommend both of them.
The Nature of Technology: What It Is and How It Evolves by W. Brian Arthur (Free Press, New York, 2009, 256pp, ISBN 978-1-4165-4405-0, $27.00)

Brian Arthur is an influential theorist of economics and complexity theory. His work has influenced many Silicon Valley entrepreneurs. He is a member of the research faculty of the Santa Fe Institute and a visiting researcher at the Palo Alto Research Center (PARC). This book grew out of lecture series Arthur gave in 1998 and 2000.

Arthur sets out to investigate, from an economist's point of view, what technology is and how it evolves. He observes that we use the word "technology" with three distinct meanings:
  • A means to fulfill a human purpose (for example, an iPhone or a legal system).
  • An assemblage of practices and components (for example, the semiconductor industry).
  • The entire body of devices and practices (what Kevin Kelly calls the technium).
For the first of these, Arthur uses the term purposed systems, reserving the term technology for systems (for example, the iPhone) that capture and exploit physical phenomena. A legal system shares many of the characteristics of a technology, but the phenomena it exploits are social and behavioral.

Arthur uses the term domain for an assemblage of practices and components. Domains are toolboxes. They form a language of rules and practices that define the possible technologies that can develop within that domain. Domain change is a major force in technological advance. For example, in the 1970s, aircraft designers moved from the mechanical and hydraulic domains to the electronics domain for controlling wing and stabilizer surfaces. Domains can have vast supporting structures, so a complete domain change can take decades and cause economic dislocation.

Arthur calls on the principles of combination, recursion, and phenomena to find a common structure for all technologies. A technology exploits one or more natural phenomena. The technology's internal structure is a main assembly that carries out its base function and a set of subassemblies that support the function. Each subassembly is a technology with the same overall structure. It isn't turtles all the way down -- the recursion is finite -- but most technologies have many levels.

This modularity makes it easy to see one of the main ways technologies advance and develop. Substituting a new subassembly for an old one can solve a problem or bring an improvement: lower cost, better performance, more functionality, suitability for more environments, and so on. For example, in the 1920s, aircraft designers wanted to achieve greater speed by flying higher, but they needed to invent the turbojet engine to make that possible. They were able to replace the engine subassembly while leaving many other subassemblies intact, at least at first.

This is a high level view, but Arthur gives a detailed picture of how standard engineering works to exploit and reinforce this model. In so doing he shows that the evolution of technologies does not follow a Darwinian model. Swapping one subassembly for another rarely happens in the biological domain. Darwin relies on long sequences of small steps and survival of the fittest.

In engineering, many factors besides fitness determine choices. For example, early nuclear reactors used light water for both the coolant and the moderator. This was not necessarily the best approach. The US Atomic Energy Commission made the decision for reasons largely irrelevant to fitness, but once it had a foothold, this design became the standard.

Arthur examines the mechanisms of technological revolutions. A new domain arises from an existing one. Perhaps an investment mania and a crash follow, but the new domain grows well beyond pre-crash levels. The economy follows as existing structures adapt and re-architect themselves over the course of decades. Change is expensive, and often the necessary changes are not obvious. The switch from steam-powered factories to electric ones required a total redesign, but factory designers did not grasp the needs and possibilities of the new technology.

The leading edge tends to be concentrated in one region. Silicon Valley is a prime example, but there are many others. A virtuous cycle emphasizes the region's initial advantage. Practitioners are drawn there. Informal networking spreads undocumented knowledge and provides unofficial channels for solving problems. This mechanism explains why Akron, Ohio, was able to turn itself into Polymer Valley when the tire industry moved away.

Arthur has advice for politicians and entrepreneurs who are concerned about competitiveness: build basic science without a stated purpose of commercial use. Encourage and remove obstacles from the small startups that naturally develop. Creating a leading edge region requires gardening, not central planning. Water it, weed it, and let it grow.

Arthur's model of the ongoing evolution of technology is as follows. A long history of technology has left us with an active set of technological components. A new technology enters the active set and becomes available to replace subassemblies of existing technologies and may enable additional technologies. The economy adjusts. The new technology creates new problems and opportunities. A domain may grow up around it. The process repeats many times, often in parallel. Like a coral reef, technology is a living thing. It has needs and a kind of autonomy, but so far it still requires human intermediaries.

Like many other writers, Arthur sees our relationship with technology as ambivalent. Our deepest hope lies in technology, but our deepest trust is in nature. We need challenge, meaning, purpose, and alignment with nature. Technology can help us meet those needs or thwart us by subjugating us to its purposes. Like good and evil, these aspects of technology are part of our world, and we must constantly engage them. We should not accept technology that deadens us, and we should not confuse what's possible with what's desirable.

Brian Arthur has produced a thoughtful and coherent account of how technology evolves. He has deliberately done so using plain language to make it available to a large audience. If technology plays a significant role in your livelihood, you should read this book.

What Technology Wants by Kevin Kelly (Viking, New York, 2010, 416pp, ISBN 978-0-670-02215-1, $27.95)

Kevin Kelly was editor-in-chief and publisher of Whole Earth Review. He had roles in launching The WELL and the Hacker's Conference. He later became Wired 's first executive editor. He is on the board of the Long Now Foundation. He co-hosts a monthly seminar series with Stewart Brand.

Kelly spent his early years as a freelance photographer, wandering around Asia with very little technology: clothing, a sleeping bag, a pen knife, and his cameras. After a religious experience in 1979 (which he described on NPR's This American Life in 1997), he resolved to act as if he had only six months to live. He gave away his money anonymously and bicycled 5000 miles around the USA to visit his relatives. Along the way he encountered the Amish, and admired their contentment.

When the six months passed and he found himself still alive, he began his new life with a unique perspective. He discovered online communities and came to see a benefit of technology that strongly influences a thesis he presents in this book, namely, that technology increases opportunities for people to realize their potential. What would Mozart, Van Gogh, or Hitchcock have been if they had lived before the piano, cheap oil paints, or film?

Kelly admires the Amish relationship to technology. An Amish community is selective about what it embraces. They try new technologies, then use known criteria to arrive at a communal evaluation and decision. Kelly embraces a similar approach in his own life. He strives to increase his personal contentment by minimizing the technology in his life. At the same time he wishes to maximize the contentment of others. This requires embracing and helping technology's growth.

To understand how he should relate to technology, Kelly looked at it from many angles, finally arriving at a grand view of an epic battle between entropy and exotropy (anti-entropy). From the undifferentiated starting point of the big bang, exotropy rapidly outpaced entropy in a continuing act of self-creation and self-organization that led to galaxies, planets, life, minds, language, and beyond. Exotropy pushes toward ever more abstract and immaterial expressions. Brains and language are important steps along the path.

At our current stage, billions of years into the process, humans have significant influence on the direction of technology's evolution, but Kelly does not see that as as an eternal truth. He sees the body of technology, which he calls the technium, as exhibiting many characteristics of an autonomous living thing. The book's title reflects this point of view. But Kelly chooses a surprising spokesman to articulate it.

Ted Kaczynski, aka the Unabomber, sees technology as increasingly restricting freedom because it strengthens society and imposes its own needs as well. Kaczynski supports an ideal of nearly complete freedom, only admitting a few rules necessary to make basic human interactions possible. Unfortunately, these rules allow mailing bombs to people you disapprove of.

Kelly cites the Unabomber Manifesto because it is an extreme representative of the opposite of what Kelly believes. When it comes to the same conclusions Kelly does about aspects of the technium, it strengthens Kelly's confidence in those conclusions. Kelly and Kaczynski both believe that the technium pursues its own needs with a kind of autonomy. Both also believe that the technium creates serious problems.

Kaczynski believes that the technium does more harm than good, and that compounding that differential over time will lead to its destruction. Because humans are increasingly dependent on complex technologies that they no longer understand, they will perish too.

Kelly believes that the technium does more good than harm. Compounding that differential over time leads to greater happiness and opportunity for humanity and to ever better ways for things to improve. Each of these men is expressing a judgment about the balance between the good and the harm that the technium does. Kelly adduces many examples to support his conclusion. For example, the long term trends toward greater longevity, health, and wealth suggest that the compounding factors are positive.

If you accept that the technium has a sort of autonomy, then it makes sense to ask what it wants. Kelly believes that technology wants what life wants. The summary, which Kelly develops with many examples and much discussion, is that technology wants increasing efficiency, opportunity, emergence, complexity, diversity, specialization, ubiquity, freedom, mutualism, beauty, sentience, structure, and evolvability. These terms define the trajectory and mechanisms that the technium has followed from the big bang to today and seems likely to follow into the future. Kelly says, "a single thread of self-generation ties the cosmos, the bios, and the technos into one creation."

At this point you might feel that Kelly has devised a creation story for an age that feels uncomfortable with the traditional ones. Kelly is quick to say that the technium is too small to be God. But he speculates that we may see a new Axial Age spurred by technology. He says,
I find it hard to believe that we could manufacture robots that actually worked and not have them disturb our ideas of religion and God. Someday we will make other minds, and they will surprise us. They will think of things we never could have imagined, and if we give these minds their full embodiment, they will call themselves children of God, and what will we say? When we alter the genetics in our veins, will this not reroute our sense of a soul? Can we cross over into the quantum realm, where one bit of matter can be in two places at once, and not believe in angels?
In this new Axial Age, Kelly believes, we may seek spiritual refuge not just in ancient redwood groves, but also in ancient networks.

This review might not convey the excitement I felt while reading Kelly's book. The book is too dense to summarize easily here. The language is poetic, but hard to paraphrase, and many of the most interesting points are hard to lift from their complex contexts. I spent months digesting this book, and I enjoyed doing so. I hope you'll do the same.