Simplicity Is Not So Simple
“Keep it simple, stupid” is often repeated and invariably good advice. Nevertheless, it’s easier said than done. The truth is that simplicity is anything but simple.
Despite our best efforts, things seem to get complicated all by themselves. That’s not because we want it that way, but it’s the way the world works and it’s not going to get any easier.
Digital technology is undoubtedly making our world more complex. More powerful computers are ushering in an era of Big Data, while increased connectivity means that everything interacts with everything else, adding further complication. If we want clarity and simplicity, we need more than just platitudes.
The Information Problem
In the late 1940’s, as the post-war era had just begun, the digital world we know today was created at Bell Labs through two landmark breakthroughs. The first and more famous was the invention of the transistor in 1947, which made possible the small, cheap binary circuits that are embedded in every electronic device we know today.
The lessor known, but in many ways no less important, revelation was Claude Shannon’s 1948 paper A Mathematical Theory of Complexity which spawned the field of information theory. It established, among other things, a measurement unit for information – the bit – a binary piece of information.
In other words, information would be defined by a single choice – a coin flip if you will. We now regularly measure information by the amount of bits it takes to describe things ranging from one character (8 bits or a byte), a one page document (about 2 kilobytes,) a full length movie (1-2 gigabytes) or the Library of Congress (235 terabytes).
Not only are we measuring information now, but we are generating and storing massive amounts of it. According to a recent McKinsey report, medium sized companies today commonly store more data than the 235 terabytes that the Library of Congress has.
Information is also getting cheaper. You can store all the music in the world (literally) for about $600, so we’re only going to see more of it.
The Simple and the Complex
Shannon’s insight opened the door to understanding what makes things simple and complex: Simple things are easy to communicate while complex things are difficult.
For instance, a googol is a very large number but simple to describe (i.e. 10100). while 479,001,599 is much smaller, but a prime number that can’t be reduced to anything simpler, so it’s very complex.
That is the essence of Kolmogorov-Chaitin complexity, which defines complexity (and therefore simplicity) by the number of bits (or coin flips) that it takes to describe an object without losing information.
Sometimes, for security, we would rather things be complex, which is why web sites will reject passwords that can be broken down to just a few bits and credit card companies multiply large prime numbers to encrypt transactions. University professors use lots of big words so only other academics can understand them and therefore justify their tenure.
Usually though, we want things to simple, even if that means we lose some information, which is why web videos can be grainy and jump around. We compress information (i.e. simplify it) to make it easier to transmit. That, in large part, is what drives the communications industry today.
Interactivity, Fractals and Chaos
So far so good. However, our simple story remains incomplete. Kolmogorov- Chaitin complexity only refers to single entities. What happens when you have a bunch of them interacting?
That’s the question that Benoit Mandelbrot sought to answer in the early 1960’s. He noticed that the noise created in communication lines followed it’s own kind of order and inferred two types of effects that created it.
Joseph Effects: These are persistent trends. Just like in the biblical story, where Joseph predicted seven fat years and seven lean years, events in a time series are highly dependent on what precedes them.
Noah Effects: These are sudden events that create discontinuity. A storm comes and blows everything away, creating a new fact pattern that will be propagated through further Joseph effects.
It was these insights that led him to pioneer the field of fractal geometry, which can create shapes of infinite intricacy and complexity merely by repeating simple rules. The most famous example, of course, is named after the man himself: the Mandelbrot set pictured below. As you zoom in and out (see link above) new details seem to emerge without end.
Mandelbrot’s fractals were more than just a fancy mathematical parlour trick. In The Blind Watchmaker, biologist Richard Dawkins describes a how interaction between relatively simple things can create the greatest complexity imaginable – life itself. This amazing phenomenon is the essence of the new science of emergence.
Emergence
In 1980, Loren Carpenter applied Mandelbrot’s ideas to computer graphics. He realized that by repeating simple fractals, he could generate amazingly lifelike scenes, synthetically. His method became so successful that the company he helped found, Pixar Animation Studios, became a Hollywood hit machine.
Carpenter then took the idea a step further and applied the principle to social interactions. If order could emerge without direction, then maybe social action could as well. In the video below, he shows how hundreds of people can coordinate their actions well enough to play the computer game “pong” collectively without leadership or hierarchy.
Later, Duncan Watts and Steven Strogatz discovered that not only that the same forces drive ordinary social networks, but that they follow the same power law rule that Mandelbrot uncovered in his original research into communication lines and that Chris Anderson found to continually pop up in his book The Long Tail.
It doesn’t stop there either. Even in staid management circles, the idea is catching on with Henry Mintzberg leading the charge to challenge decades of corporate planning orthodoxy by advocating greater emphasis on emergent strategy.
Simple Rules For A Complex World
The evidence is clear: There’s nothing simple about simplicity. Our world is getting ever more complex. Computational power is increasing exponentially. Barriers are breaking down. Outcomes are extremely sensitive to initial conditions because events materialize out of unplanned interactions.
Therefore, our goal should not be to seek out ultimate simplicity, but maximum manageability and there are some basic principles that can help us do that.
Factor Down One (or possibly two) Levels: As I wrote in an earlier post, technologies are made up of combinations of other technologies. That applies to not only physical objects, but cultural objects as well, such as democracy, a body of law or even a corporation. So we can always break down a concept to its constituent parts.
However, our sense of purpose can quickly disappear down the rabbit hole if we allow every discussion to devolve into an examination of the components of components’ components.
Factoring a problem down one or (possibly two) levels will explain the vast majority of variables and will allow you to isolate and focus on one or two specific areas of concern. Any more than that and the amount of information you will have to deal with will be unmanageable.
Limit Variables: Centuries ago, medieval scholars gave us Occam’s razor, which urged us to use as few variables as possible.
It’s good advice. As this HBR article points out, less is more. If you have 5 steps in a process, see if you can narrow it down to three. If you have fifty slides in a presentation, try to cut it down to thirty-five or even thirty. Small teams generally perform better than large ones.
They key here is to use the absolute minimum number of entities that you can without losing essential information. When in doubt, leave it out.
Look For Patterns: Although individual events might be unpredictable, the fates usually consort in recognizable patterns. As mentioned above, systems with interaction and feedback tend to result in power law distributions, while independent systems generally follow a bell curve. Making those types of distinctions accurately is incredibly important.
In a similar way, industries tend to have their own rhythms and cycles. If you are an ad agency, for instance, and clients drop surprise holiday campaigns on you every year, then they shouldn’t be such a surprise.
Understand That More Is Different: In Steven Johnson’s book Emergence, he stresses that more is different. Changing scale means changing the rules.
A salient example is Dunbar’s number, which suggests that relationships in groups start to break down once the total number of people goes over 150. In a company, once you grow past that, management practices have to change drastically, just as you need to speak differently when presenting to a group than you do in a one-on-one conversation.
As I noted in the beginning, while things may start out simple, they will inevitably become more complicated over time. So while we might yearn for a simpler era and a simpler life, modern life demands more of us.
The best that we can do is keep entities simple, accept that systems become complex and seek to manage that complexity while understanding that it can not be controlled.
– Greg
If we consider this in terms of adult learning which is in my mind key to the success of social media and why it has seen the growth it has had. People want to learn and groups help individuals learn better and apply what they’ve learned. And I agree keep the groups focused and small in order to maximize effect. The greater the complexity the more difficult the application of knowledge.
“Adults need to participate in small-group activities during the learning to move them beyond understanding to application, analysis, synthesis, and evaluation. Small-group activities provide an opportunity to share, reflect, and generalize their learning experiences.” http://tiny.cc/72vh2
Interesting perspective. Thanks.
– Greg
“You can store all the music in the world (literally) for about $600” Wow! I had to Tweet that. Good stuff….but now my brain hurts. So much information! Thanks again, Greg.
Thanks James. I see your blog is going well, btw. Congratulations!
– Greg
Great post, Greg
I fully agree with your closing thoughts:
“The best that we can do is keep entities simple, accept that systems become complex and seek to manage that complexity while understanding that it can not be controlled.”
Most complex systems, particularly those involving human behavior, are inherently difficult to control as they’re not deterministic. Operating within those systems requires dealing with ambiguity and uncertainty. You’re right in saying that it’s necessary to keep things as simple as possible wherever we can. But it seems equally essential to avoid oversimplifying a complex reality through modeling or quantification.
This reminds me of Roger Martin’s knowledge funnel where he discusses the tradeoff between exploration and exploitation. Once, knowledge / insights have been explored and validated (‘heuristics’), they’re utilized by being translated into describing models and measures in order to assure control and repeatability (‘algorithms’). That’s what usually occurs when startups transform into bigger companies. As one consequence, e.g. their innovation capability changes as they increasingly focus on analytical management.
I think, managing companies in a complex business environment basically requires a parallel approach:
– understanding what works and what doesn’t through experimentation and agility,
– utilizing validated insights through mapping them by means of a manageable (simple) set of parameters / measures.
What do you think?
Cheers, Ralph
Ralph,
I think there’s some truth to that, but I also think that you need to be somewhat humble and realistic in what you can achieve.
There’s no way to control a big company (or even a medium sized company). I think that’s one reason that all the talk about “radical change” is bullshit. A complex system doesn’t change just because you want it to.
That’s why I think simplicity is important – for transparency – so that you can see patterns emerge and actually get some understanding of the system (that’s actually very hard to do).
So yes, I would say that I agree with what you said, except it’s not just parameters that need to be simple, but processes, org charts, etc. All to often, too much complexity is built in when people attempt to build the “perfect system.” You’re much better off building a simple one that you can more fully understand.
– Greg
No article which talks about simplicity should be without Einstein’s famous quote:
“Everything should be made as simple as possible… but not simpler.” 🙂
And the most complicated objects can arise out of the simplest rules. Cellular Automata which are simple objects with simple interaction rules can produce astonishing complexity when left to interact. They are the basis of much Physics and the popular Conway’s “Game of Life” which produces self-organizing higher level patterns when the cellular atomata are left to evolve on their own.
http://en.wikipedia.org/wiki/Cellular_automaton
Also has a few of the Conway patterns there.
– Robert
Thanks Robert.
To be honest, I’ve kept away from Cellular Automata because, quite frankly, I’ve never seen it explained in a form that I could understand.
However, one thing that has struck me is that the true pioneers of the computer age, John von Neumann and Norbert Wiener seemed to see quite early that the designs they had pioneered were not optimum. Given that we live largely in a von Neumann architecture environment, that seems a salient point.
I would be grateful if you could point me to some good resources to help me understand it better. Or has the idea basically been co-opted by emergence?
– Greg
Hi Greg and Ralph,
great conversation… Complexity is inherent to our environments, but there is no intrinsic opposition between complexity and simplicity. In fact, it is just the opposite, as ANY elements we are dealing with (humans, economy, business,…) are already complex, the simplest rules lead to the most complex super-systems.
Trying to build a “perfect system” too often means introducing complicated rules. Realism is not about control and understanding (alas) but much more about figuring out some of the simple rules which govern systems, and letting the rest “as is” until we are able to get some of these things more simple.
My guess is that our new motto should be “if it’s broken, don’t fix it”. Trying to simplify those imbricated layers built for illusory control would only lead to more complication. Building new systems is seldom feasible. Instead, we should try to seed simplicity wherever possible, and let the systems evolve until new steps are required.
Very good advice Thierry!
However, I do think that there is a middle ground between control and futility. No matter how complex the system, you can influence it, often to a great degree.
I also think that you sometimes have to ignore complexity. For instance, each human is complex, they have a personal life, hopes, dreams, etc. However, if you start going too far into the details of each entity in the system, you’ll get completely lost!
That’s why I recommended going down one or possibly two levels. It’s a fuzzy concept, open to interpretation of terms, but it’s important to go into just enough detail to explain the majority of forces while understanding that there is some stuff your leaving out. If you miss something important, you’ll know soon enough and be able to re-frame.
That’s much better than having to re-learn your own explanation every time you have to solve a problem:-)
– Greg
Of course you can influence system, even on a large scale. And I also agree on your take about fractals (what you called ignoring complexity, which is in fact staying at a more macro level).
But don’t you think that what mostly messes things up is not complexity, but the (often) inefficient structures we put in place in order to deal with complexity from control?
You tweeted a interesting post about the state of Agile sooner today, which corroborates this point. Control structures (hierarchies, processes, etc) are the most difficult to get rid with, but they are the most visible part of our organizations. Could we be able to distinguish between complex and complicated behaviors? To take your human being example, analysts know how difficult it can be to pinpoint human motivations, hidden under many layers of self-control… Apparently emergent patterns might in fact be induced from control artifacts…
Theirry,
It’s an important point. We create abstract structures, much like Plato’s forms, that don’t exactly conform to reality. They’re necessary, but any reasonable person understands that the truth lies elsewhere (unfortunately, there are many unreasonable people who don’t!).
So I think what is important is framing. You can never fully understand complexity (that’s how it got the name:-), but you can frame it within a certain context to solve a particular problem.
A solution and an answer are two different things.
– Greg
Cellular Automata is all about systems being defined locally rather than globally. The difference is something like Newton’s law of Gravity being used to calculate the obits of the planets of the Solar System and a scheme where each object was given local rules on how it worked and then the interactions of all of the locally acting objects produced the overall behavior. I don’t know if anyone has constructed a Cellular Automata based theory of Gravity but I wouldn’t be surprised.
The common example I’ve heard to describe Cellular Automata is the patterns that can be seen in people walking through a crowded train station. Each person has simple rules on avoiding other people while going their way and the interactions of that produce higher order flow patterns that can be seen if you’re standing on one of the overhead galleries observing. It’s surprising (at least to me) on how much Physics can be redefined and produced by such systems.
“Or has the idea basically been co-opted by emergence?”
It’s more a matter of degree if I understand it correctly. “Emergence” seems to be about surprising properties that wouldn’t have been predicted and only exist when the system gets complex enough. The patterns produced by Cellular Automata are “all the way down” if you get the Turtles analogy. There’s never a clear “break” between one level of complexity and the surprising “emergent” properties of the common emergence examples if you get what I mean.
I’ll keep an eye open for Cellular Automata sources. Stephen Wolfram, the prodigy who did the Mathematica program (among a lot else), has done a lot of work on them and you can find a bunch of his papers easily by searching. His Magnum Opus, “A New Kind of Science”, is all about them and their use in Physics but the 1,280-page tome is considered quite turgid.
http://www.amazon.com/New-Kind-Science-Stephen-Wolfram/dp/1579550088
– Robert
Richard,
Thanks for that. I’m not sure that there is a clear delineation between Cellular Automata and emergence (and chaos, for that matter). All three seem to evolve around the “micromotives and macrobehavior” to borrow a phrase from Schelling. If there is one, I’m missing it:-)
– Greg
Great conversation, Greg and Thierry – thanks for expanding the initial thoughts!
You’ve been indeed touching the field of Design Thinking and Wicked Problems here. This gives me some good inspiration for the second part of our joint post, Thierry 😉
I think, Greg summarizes it quite well by saying ” You can never fully understand complexity, but you can frame it within a certain context to solve a particular problem.”
Structures – what I called (simple) maps in my comment above – help to describe and handle certain parts of reality, but are not the reality itself. Depending on knowledge and applied context, these structures can be (re-)framed. It’s often forgotten that models do not come without limitations.
That’s similar to what I learned in my studies about an electron. Is it a particle? Is it a wave? It’s neither / nor and both / and. Depending on the context, it can be described as a particle or a wave. But neither of theses models fully comprehends the actual nature of the electron.
Cheers, Ralph
After starting with simplicity, we somehow arrived at Heidegger. How’d that happen?
As to your physics analogy, I think you can also use the equivalence of matter and energy. Essentially, they are the same thing, but you don’t go around saying, “Hey, buddy, can you get me a million calories to help prop up this door?” In that context, the relationship is completely irrelevant.
Great discussion guys. Thanks!
– Greg
Great post and discussion, thanks! May I add a few comments?
1) humans can indeed increase the complexity of a system by messing things up. That is however not to mix with complexity as an intrinsic caracteristic of systems.
2) the difference between complicated and complex systems can well be understood applying the cynefin framework http://en.wikipedia.org/wiki/Cynefin
3) “factor down levels” and “limit variables” may bear the risk of destroying the complex system by rendering it complicated thus leading to a trivialisation of the system. Trivialisation of a system makes it complicated or even simple, at least to my understanding (it remains in reality complex). How could I know that these variables are not of great importance for the whole system?
4) only complicated systems may be controlled, complex systems may only be influenced.
-Philippe
Thanks for your input.
– Greg
“our goal should not be to seek out ultimate simplicity, but maximum manageability”
The bottom line here for me is filtration.
These models, methods, mindsets, processes, are all dynamic forms of filtration.
As long as your maximizing your ability to manage the “filters” you come closer to simplifying as Ralph says exploration and exploitation of that complexity.
But exposure must come before anything…
Social has “exposed” lots and lots of complexity 🙂
The only way to sustain simplicity is having a “filtering system”
Filtering systems require observation, capturing ability, adaptation, calculating, designing, and it’s continuously improving on “filtering” what’s exposed.
Mindsets, models, methods transition understanding.
for example Outcome Driven Innovation is a framework that filters the complexity of consumers jobs to be done.
Spiro,
Filtering is certainly important. However, even then you still have to worry about how things work together.
– Greg
In my view, complexity is like entropy. It naturally increases unless ongoing energy is put into the system to keep it simple. It is also interesting to note that complexity is very similar to heat. There is no limit to how much of it you can have, but there is an absolute limit to how little of it there can be. And the window that sustains life is fairly close to absolute zero.
You might be interested in some of my white papers on related topics:
The Mathematics of IT Simplification
A Cross Disciplinary Look at Complexity (coauthored by Nikos Salingaros)
Both of these are readily (i.e. no registration required) available at http://www.objectwatch.com under the white papers tab.
Roger,
I think too (as did Claude Shannon). Thanks for your comment.
– Greg
I would have made it simpler, but I didn’t have enough time. 🙂
Nice one:-)