Simplicity: A New Model

Note: This article will be is part of the book Management 3.0: Leading Agile Developers, Developing Agile Leaders.

Ah, simplicity.

We all seem to want it, but we rarely seem to get it.

Many experts have discussed simplicity and complexity. But their contributions often confuse various terms, which hasn’t led to a simplification of the discussion itself.

Here is my attempt to clear things up a little…

What is simplicity?

Simplicity usually relates to the burden which a thing puts on someone trying to explain or understand it. Something which is easy to understand or explain is simple, in contrast to something complicated.Wikipedia

If you’re going to discuss simplicity, it is important to know the difference between complex and complicated. Not knowing the difference means you might apply exactly the wrong approach to the right problem. (Or, the right approach to the wrong problem.)

I believe the difference needs to be explained using two dimensions. The first dimension is about the structure of a system, and how well we are able to understand it:

  • Simple = easily understandable;
  • Complicated = very hard to understand.

The second dimension is about the behavior of the system, and how well we are able to predict it:

  • Ordered = fully predictable;
  • Complex = somewhat predictable (but with many surprises);
  • Chaotic = very unpredictable.

The Structure-Behavior Model

My underpants are simple. I found it easy to understand how they work. But my watch is complicated. If I took it apart. it would take me a long time to understand its design and its components. And yet, neither my watch nor my underpants hold any surprises. (At least not for me.) They are ordered, predictable systems.

A three-person software team is simple too. It takes only a few meetings, dinners, and beers to get to know everyone on a team. A city is (usually) not simple but complicated. It takes taxi drivers years to know all its streets, alleys, hotels, and restaurants. And yet, both teams and cities are complex. No matter how well you know them, there will always be surprises. They are predictable to a degree, but you never know for sure what will happen tomorrow.

A double pendulum (two pendulums attached to each other) is also a simple system. It is easy to make and easy to understand. And yet, it undergoes unpredictable chaotic motion due to a high sensitivity to the initial setup of the pendulum. And stock markets are also chaotic. They are by definition unpredictable, or else everyone would know how to make money on stock exchanges, and the whole system would collapse. But, unlike pendulums, stock markets are also extremely complicated. The many different businesses and types of financial properties and transactions make them utterly incomprehensible for a simple guy like me.

Complicated refers to a system’s construction being too intricate to understand, unless you’re an expert, while complex and chaotic refer to a system’s behavior, which is unpredictable to a small or large degree.

What is complicated is not necessarily complex, like two cars in a garage. And what is complex need not be complicated, like two people in a bedroom. (But these people’s behavior in their bedroom can be quite unpredictable.)

  • Simplification is the act of making the structure better understandable (moving it from top to bottom in my model.)
  • Linearization is the act of making behavior better predictable (moving it from right to left in the model.)

Unfortunately, linearization is (in laymen’s terms) usually confused with simplification. And that’s where the complications start…

How Does This Differ From Other Models?

Figure03-3Ac Cynefin is a model devised by knowledge management scholar David Snowden. It describes a typology of contexts using four domains: Simple, Complicated, Complex, and Chaotic, and is used to guide approaches to decision-making and policy-making.

However, Snowden doesn’t distinguish between the structure and the behavior of systems. This can make discussions a bit complicated (not complex).

Figure03-3Bc Management professor Ralph Stacey created something similar, called the Agreement & Certainty Matrix. It shows Simple, Complicated, Complex, and Anarchy (Chaos) as four areas based in two dimensions: the degree of agreement and the degree of uncertainty.

This model suffers from the same problem. I don’t see complicated and complex as two discrete domains. Instead, I see them as parts of different dimensions. That’s why my Structure-Behavior Model identifies six domains instead of four. And some systems (like cities) can be both complicated and complex.

OK, Now We Can Revisit Simplification

I believe my Structure-Behavior Model is able to simplify discussions around simplicity, and to clear up some misunderstandings…

Everything should be made as simple as possible, but no simpler.Albert Einstein

With this quote Einstein meant that a system must be made understandable, which means moving it vertically, from the top of the model to the bottom (simplification). However, his addition “but no simpler” maps to the behavior of the system. Einstein warned not to change the system horizontally, because that would change the kind of system (linearization.)

Simplicity is a myth whose time has past, if it ever existed.Don Norman

In an inspiring article Don Norman discussed the value of having more features in a product, instead of fewer. More features means different behavior, and (often) also a different structure. In my diagram it is both a horizontal and vertical issue. (For example: Google just added Priority Inbox to GMail. This made GMail’s behavior more complex for me. It also complicated the user interface, but I still seem to understand it well enough.) Unfortunately, Don Norman used the term simplification both for linearization of behavior (horizontally) and simplification of structure (vertically).

And so Don complicated his message, which is exactly why many people didn’t understand him. Maybe it would have helped if Don had used pictures:

The goal of visual thinking is to make the complex understandable by making it visible, not by making it simple. – Dan Roam

In his bestselling book The Back of the Napkin Dan Roam suggests to use pictures to make things understandable by drawing pictures. He clearly refers to moving things from complicated to simple (vertically). However, his warning “not to make things simple” is, again, a confusion of terms. What Dan means is that pictures should not change the complexity (behavior, meaning) of something, because that would mess up people’s ability to predict what the pictures are trying to say.

And so…

Yes, by all means, simplify everything that is hard to understand!

But no, you may not want to linearize (“simplify”) something, because the reduced behavior of what you offer may not be what your user had expected.

Thanks for reading. I hope it was simple enough to understand.

Jurgen is an experienced author, trainer, and speaker. Why don’t you hire him to add some spice to your company event or seminar?

Twitter TwitterRss SubscribeEmail NewsletterLinkedIn LinkedInSlideShare SlideShare

Latest, greatest and favoritest posts:
When You Are Rejected
Agile Is a Risk Management Strategy
Where There's People, There's Problems
  • Iterate Your Flawed Models
  • The Best Used Books in My Library
Related Posts
free book
“How to Change the World”
  • Paul Boos

    This is an outstanding post! I just sent it out to my entire team to read. Thanks for sharing your thoughts…
    Another excellent model that covers both complexity and complicated in a project context is Reinventing Project Management by Aaron Shenhar and Dov Dvir. Complicated though gets scattered across several dimensions though, so I still find your model great.
    They help assess projects and the types of approaches you may need to using a radar chart; the four domensions are Technology, Complexity, Pace, and Novelty.

  • Misty Funk

    Fantastic article. I love simplicity articles and quotes. Thanks for sharing.

  • Clinton Keith

    Excellent article. I see your point about it communicating the distinction of complicated and complexity.
    However, I don’t quite understand the value this added detail has in communicating what the Stacey diagram does so elegantly; helping decide management/leadership strategy based on certainty and agreement. The Stacey diagram is a very useful image for explaining that.
    Perhaps some examples of how this is applied would create this understanding. I’m probably being missing something obvious…

  • Jurgen Appelo

    Hi Clint,
    Each model is wrong, but some are useful. And each model solves a different problem.
    I agree the Stacey model is useful for some problems. Bu it’s no help in the heated discussions about what simplicity means. Because it doesn’t distinguish between structure and behavior. (Same with Snowden’s Cynefin model.)
    I think my model does help in such discussions. But I’m sure it’s not meant to help managers with their leadership strategy. Mine has different dimensions.
    So, each model has its own strengths. Use them where you think they can help.

  • Leiderleider

    Thank you Jurgen for supplying words to this topic. It expresses quite clearly what I have been thinking of for a long time. It surely will help me communicate simpler with others about this topic.
    The only thing that is not really easy ist the word “linearization”. One aspect to me is that in your horizontal dimension to me it feels like crossing steps whilst on the other hand moving vertically is rather a linear process with every “value” possible.
    I believe you describe something that covers the dimensions Quantity (vertical) and Quality (horizontal).

  • Jurgen Appelo

    Thanks! But I don’t think it’s quantity vs quality.
    For example: some cities are easier to comprehend and navigate than other cities. What counts is not the size of the city, but the way the street plan are structured.
    And it’s true that both dimensions (horizontal and vertical) are themselves simplifications. 🙂

  • Florian Hoornaar

    Great post. However I would like to offer an alternate explanation to Einstein’s quote: “Everything should be made as simple as possible, but no simpler.” One should strive to make everything as understandable and predictable as possible without it changing function. Shifting horizontally (to the left) is no problem. The more predictable the behavior of a system the better. The thing that should not change is (the amount of?) capability. This introduces a third axis.

  • Ike

    You were right to break it down according to structure and predictability, because they answer such different questions. Replicating the machine is very different than replicating the likelihood of the results.
    I wish I had written this instead.

  • Cynthia Kurtz

    Hi, great post and great examples. I’ve been thinking about the same issues and have come up with remarkably similar solutions – great minds think alike – see especially here – happy to connect and discuss!
    Cynthia Kurtz

  • Clinton Keith

    Agreed. So it’s a model of a model then? 😉

  • Don Norman

    Thanks for the excellent treatment of simplicity, complexity, etc., including several references I was not aware of.
    You properly critiqued my early paper on the topic. But in my new book (to be published in October), I take a slightly different stance. I also distinguish between being complicated (which means confusing) and being complex. Complexity, I argue, is necessary and good. What we need is NOT simplicity, but understanding. We want complex without being complicated.
    Chapter 1 is available at
    Thanks for the nice column.
    Don Norman
    (At KAIST, Daejeon, S. Korea)
    (I’ll be in Delft in December)

  • Taotwit

    @ Don @ Jurgen,
    I think I understand that complexity is not the ‘enemy’ and that some systems are inherently complex. And I can see that the ‘complicated’ might be replaced with ‘confusing’.
    But with Dan Roam’s assertion that the antonym of Simplicity is Elaborate in mind, do we need space for the concept an elaborate system design which is not the same as a complicated (confusing one) nor is it fundamental complex (partially unpredictable)?
    BTW I believe we should strive to communicate the structure and behaviour of systems *Simply* – so I love Jurgen’s pictures and engaging words!
    @taotwit nigel

  • Matthew T. Grant

    When you lay out your schema for Simplification and Linearization, I think it should read “more understandable (or, perhaps “comprehensible”?) and “more predictable.” “Better predictable” sounds odd to this writer’s ear.
    Thanks for the very thought provoking, and thought clarifying, post.

  • Chris Sterling

    I think this is a great article supplying a model that can be helpful in certain contexts around simplicity. I am a little worried about the discussion on Cynefin since it does not provide the additional aspects of the model: disorder and catastrophic change. Cynefin provides guidance on how to approach situations that are in each domain and also how a particular situation can move between domains. For instance, response to a simple situation would be sense->categorize->respond and to a complex situation we probe->sense->respond. It is my opinion that Cynefin separates structure and behavior in a different way than how you are doing so in your model. I want to get deeper into discussion some time with you about this since it is hard to have such a deep conversation over comments or online.
    Again, there seems to be some great stuff here and I can’t wait for what is to come.


    Hi Jurgen
    I like the additional distinction but you are over simplifying the cynefin framework to do so. Cynefin has five domains, not four and the central ‘disorder’ domain is equally important as the other four. Also the boundary between simple and chaos, which on your model only works if you create a cylinder, is crucial to the model to explain systems that tend towards and often topple over the cliff-like chaos boundary.
    Otherwise keep challenging, exploring and blogging.
    Cheers, Ron Donaldson

  • Jurgen Appelo

    Ron, thanks for the update.
    You’re right about the 5th domain in Cynefin.
    Though I must admit I have serious doubts about it.
    I am also skeptical towards the idea of things
    “falling over a cliff-like edge into chaos”
    I think it’s a metaphor that doesn’t work for me.

  • Davide ‘Folletto’ Casali

    As a long time disciple of simplicity there are a few things that sounds strange to me. I’ll try to do my best in explaining, and I hope to publish something more articulated soon.
    The examples you make are for me really troublesome.
    You say that “my underpants are simple but my watch is complicated”, and the reason is how much more complex is the watch when opened. I’m sure that we could get into a similar complexity if we “open” the underpants too, textile’s experts will have truly something to say about it. 😉
    Another example: you say also that the double pendulum is more unpredictable than a group of three people, I really beg to differ! With enough knowledge and maths about the double pendulum I’m able to define its position in any moment (it’s chaotic *and* deterministic), while I’m really not able to do the same for a group of three.
    If you “open” a clock, you’re jumping to a completely different level of perception. It’s an open clock, non a clock anymore. It’s like comparing the UI of an application and the source code of it: you will never compare the source code of Windows to the UI of OSX. Don’t jump the point of view when you compare things! 🙂
    —> I’d suggest you to pay more and more attention to the examples you’re using, because I don’t think they map correctly to the concepts you’re expressing.
    The theory is somewhat in the right direction, but I think you’re making some technical distinctions that are more part of the effects of simplicity and simplification than being simplicity itself.
    For example, you talk about “ability to understand” and “ability to predict” but what does it mean? If I’m able to understand – that’s something really different from “knowing” – you are also able to predict. If I understand the double pendulum and the maths governing it, I’m also able to predict. If I understand my underpants, I’m able to predict it, whatever it could mean :D. If I understand how a clock works, I’m perfectly capable of predicting its evolution.
    Still I agree with you that there’s a lot of confusion, but the counter-thesis you’re making I don’t think is nailing correctly the issue. As a designer, more than often I’m in a position when I’m able to simplify – or linearize in your definitions – even the processes (behaviour), that doesn’t mean that it isn’t simplification: it’s just that I’m just simplifying a different thing.
    The term you then choose to express the different metric you’re simplifying is up to you, but I won’t use a different term.
    Try thinking as simplicity in a way more related to the cognition processes in an emergent dialogue between the subject and the world, in a way similar to Maturana’s idea of emergency, and I think that you’re going to improve significantly your definition accuracy.
    That said, the suggestions and the conclusion you’re making are quite on spot. I don’t agree in how you get there or on the definitions you’re making for the reasons I’ve expressed, but still those conclusions represent useful tips to whoever wants to improve his simplification skills. 🙂
    I’ll try to clarify better my take on simplicity soon. I’ll link you back then. 🙂

  • Jurgen Appelo

    Hi Davide,
    Thanks for your reply. Your input is very good and highly appreciated. I agree that it should be possible to come up with other (better) examples.
    Like you I think I’m heading in the right direction. But my thinking on this issue is only a few months old, and I appreciate your feedback!
    I will think about it some more, and I might return to this topic some time later. In the meantime I look forward to your writing, and how you could improve on my ideas.
    Thanks again,

  • playmobil

    dieren playmobil

    dieren playmobil

  • Niels Pflaeging

    At least field 5 and 6 do not exist.
    Makes the double axis graph a bit illogical.

  • Fabrice Aimetti

    Hello Jurgen, I’ve translated to french your great article :

    Thank you, Fabrice

How to Change the World - free Workout - free