# Combining Simplicity and Complexity

There’s an ongoing debate about the merits of simplicity vs. complexity. “People want things simple and easy to use!” proclaim some.

“Balderdash – they want complex and powerful behavior!” exclaims the other side. And back and forth they argue, gnashing teeth and brandishing keyboards.

I think the problem lies in the confusion of terminology, which the authors hint at but don’t state explicitly. Pitting simplicity against complexity in a virtual cage match creates a false dichotomy, or the belief that you must choose one or the other. Both are possible.

This isn’t a cop-out, “can’t we all just get along” answer. I think the real issue is that we are mixing terms. Simplicity and complexity really can be friends, and don’t have to fight to the death (like Kirk and Spock they are best friends, and even if they do fight it’s only a charade).

When we argue about a thing being simple or complex, we are unknowingly asking two higher questions:

• How easy is it to understand?
• What can it do?

These are the questions we ineffectively try to answer using the words “simple” and “complex”. Unfortunately two words aren’t enough; we need four to answers these two questions:

How easy is it to understand?

• Simple: Easy to understand, straightfoward
• Complicated: Difficult to understand, convoluted

What can it do?

• Advanced: Does a lot, powerful
• Basic: Doesn’t do much, simplistic

Framing the problem this way lets us separate out the good and bad answer for each question.

Being simple or complex is a good thing. Being simplistic is an ok thing. Being complicated is a bad thing. Let’s see why.

## Case 1: Simple and Basic

Simple and basic is the stereotype of simple: We think that if something is easy to understand, it isn’t capable of much.

This is true a lot of the time. Think about a rock, an oar, or a spoon. These are basic tools and easy to use, though they don’t seem to accomplish much. In software world we have notepad: it’s easy to understand, but not very powerful (no spell checking, embedded graphics, etc.).

Are simple, simplistic things good? You bet. They get their (simple) job done. And their ease of understanding is a great benefit: it’s excellent for education purposes, and often times we don’t need the power we think we do. In fact, having simple behavior often leads to increased reliability — how often does an oar “break down” compared to an engine?

Also, there’s nothing stopping you from taking multiple “weak” items to create a powerful one, like using thin threads to make a thick rope. I’ll touch on this later.

Rocks, spoons, and notepad are fine in my book: they have their uses.

## Case 2: Complicated and Basic

Ah, now this is a strange beast. What item could be hard to understand yet not accomplish much?

A Rube Goldberg machine. It’s a contraption built for a basic task, such as lowering a sign, using absurdly convoluted and intricate means. Here’s what I mean:

These devices are complicated (can you immediately tell what it will do?) and basic (lowering a sign isn’t very awe-inspiring). They stink from a practical viewpoint, though they do have redeeming value for entertainment, artistic or educational purposes (how not to build a device).

Unfortunately some software is like a Rube Goldberg machine, like setting the clock on your VCR. Setting the clock should be a simple task, but it often involves a complicated, unweidly procedure because the interface of a VCR is not designed for it (“Press channel up to pick the date…”).

Strangely enough, people don’t seem to get entertainment value from seeing how complicated this is. They get frustrated, which is a bad thing. Items in this category should be avoided.

## Case 3: Complicated and Advanced

This is the stereotype of most powerful devices: Sure, they can do a lot, but they are really hard to use.

A real-life example is a helicopter. It can fly straight up, backwards, and manuever in any way imaginable: it’s extremely powerful. Unfortunately, you need extensive training in order to operate one. I’ve never flown in one, but apparently it requires the use of all 4 limbs and your brain in order to operate it.

Items in this realm are often on the cutting-edge. They are our most advanced technology that works, but we are still in the process of figuring it out, discovering patterns and optimizations that make it easier to use.

Computers were originally complicated and advanced. Early computers were powerful (they could perform any computation, albeit slowly), but were very hard to use. You had to use punch cards or even enter information manually using switches. As time went on, we developed keyboards, graphical interfaces, and better programming languages. We were able to input instructions in a more simple (easy to understand) manner.

Programming is still not dead-simple: simplicity and ease of use is a range. But it’s clear that computers are vastly simpler (easier to use) than they used to be. The underlying technology, microchips, has become more advanced (powerful) and also more complicated (difficult to understand) in order to make computers simpler. There was a time when a single person could understand a microchip or operating system — no longer.

Complicated and advanced devices are “ok” — they do get the difficult jobs done, but when using them you often think there’s got to be a better way. And there often is.

## Case 4: Simple and Advanced

Ah, this is the holy grail. This is what mathematicians seek by “elegant” equations, what scientists yearn for with beautiful theories, what designers seek when creating products. Consider Einstein’s famous equation:

$\displaystyle{E = mc^2}$

Energy and matter are equivalent – you can convert one into the other. This concept is astonishingly easy to understand (simple) yet describes extremely powerful behavior (nuclear reactions). Many physics equations are like this: Gravity, Newton’s laws or Maxwell’s equations. They are 1-liners that guide us through powerful behavior, the workings of our universe.

Google is a simple interface for incredibly powerful behavior — finding any document on the Web. The ipod (I’ve never owned one) is claimed to be an amazingly simple device to manage and play music. Even programming can be like this: A Turing machine is a simple model of computation (writing symbols on a ticker tape) that can do the same calculations we can perform on a modern computer. Heck, MacGyver can create a bomb out of simple, basic parts like a toothpick, comb and bottle of shampoo.

## Use Simple Building Blocks

Advanced behavior often comes from simple parts. The beauty of the Unix design philosophy was to have many simple, even basic programs that did a single task well: combining files, sorting them, counting lines or searching for words. Each tool was basic, but when linked via “pipes” could lead to very powerful behavior. If you are a programmer, I urge you to learn about Unix if you don’t already.

Much of the world is made from simple, easy-to-understand building blocks. Simple atoms make any object. Simple DNA (only 4 bases) is stretched into long sequences, creating the instructions needed to make a human being. The most intricate video file is still a sequence of 1′s and 0′s.

It didn’t have to be this way. We could have had thousands of elements in the periodic table. DNA could have had millions of different bases. We could have designed computers to store files with 0′s, 1′s and 2′s.

But that’s not what happened. The best-designed, most elegant systems are simple and advanced. Simplicity gives them reliability, and a clever arrangement of parts gives them power.

## Simple Isn’t Easy

There’s one giant caveat here: “easy to understand” does not mean “easy to do”. Running a marathon is easy to understand. It is not easy to do. Similarly, actually creating powerful behavior from the easy-to-understand parts can be a challenge.

It’s tough to find underlying patterns in chaotic, complicated behaviors. It’s easy to get something working and leave it at that. But looking at our natural laws, there’s inspiration that nearly any complex phenomenon or design can be built simply.

## Parting Thoughts

Have I resolved the debate? You be the judge. To me, simplicity and complexity coexist peacefully by thinking about two separate questions:

1) How easy is it to understand?
2) What can it do?

Don’t confuse a simple interface with basic behavior. Don’t assume a complicated device is powerful. Think about these questions independently and you’ll be fine.

## 12 Comments

1. That’s a nice way to break it down, but your argument would be more persuasive if you chose clearer words: complex doesn’t mean powerful; it means complicated, composite, or intricate. How about weak/powerful and simple/complicated?

2. Kalid says:

Hi Jonah, thanks for the comment. Yes, thinking about it more, I think the complex/complicated argument may be too subtle. To me, “complex” can take a positive or at least neutral connotation, while complicated is typically negative.

I like weak/powerful because it polarizes the issue, but unfortunately weak has a negative connotation; perhaps “basic” and “advanced” can work.

Something along these lines may do the trick:
* What can it do? (basic/advanced)
* How easy is it to understand? (simple/complicated)

Update: Yes, I like this distinction better. I’m going to update the article to make this more clear.

3. Anonymous says:

A turing machine is simple & advanced, but is it easy to use/productive? Not for any sort of real computation, it’s more labour intensive than assembler! I think you need to clean up your model for that kind of case.

4. Thanks, that’s a good observation. A Turing machine is “simple” in the sense that every step is easily understood, but it hardly “simple to use” or expressive. I may have to replace it with a better example

5. creaothceann says:

nitpick: the guy’s called “MacGyver”.

6. Kalid says:

Ah, thanks for the correction — just fixed it :).

7. Specialist says:

Try using the lambda calculus as an example instead of Turning machine.

8. madkilla says:

Simplicity is best, you guessed
but really not sure
thoughts are impure
in your path theres a detour

9. Charmkite says:

As a math oriented person, you might be interested in the Akaike Information Criteria (AIC). It basically pits complexity and simplicity against each other when making choices, and finds a happy medium. It is used for selecting models when fitting data, but the thought can go a long way.

10. Paul says:

Your thoughts expand on a throw away discussion with one of my computer science teachers many years ago. It has always stayed with me and whenever I’m in the depths of analysis paralysis I go back to it for guidance:-

“Complex and complicated are not the same thing”

A long piece of straight wool is simple and not complicated.

Tangle that piece of wool up and it remains simple (a single entity), but is now complicated (entagled and not easy to discern what is going on).

Take that original piece of wool and chop it up into pieces. It is now complex (multi faceted) but not complicated (each piece is readily understandable).

Only once those multiple pieces of wool are tangled together do we get a complex and complicated system.

11. Kalid says:

@Charmkite: Cool, I’ll have to check that out.

@Paul: Thanks for that awesome physical analogy, I like it!

12. john says:

simple, complicated, explanation, symplectic etc.
Sometimes etymologies are misleading, but they can be suggestive and illuminate root images.
Cf. protoeuropean ‘plek’ having to do with weaving, braiding, etc.
If the consensus of theoretical physicists is correct, the best understanding we have of the uni-/multiverse covers less than 5% with “dark matter” and “dark energy” making up the rest 95+%.
“Method” from the Greek meta+hodos comes from “along the way”. “Problema” from “(obstacle)thrown up in the way” (Tao). Clarity vs obscurity; distinction vs confusion (Descartes) vis-a-vis “theoria”, “to see as the gods see”.
Subscribe (free) to online “Complexity Digest”. Look at R. Onians, The Origins of European…REV. ed. Bruno Snell is also useful. Onians covers Chinese and Icelandic etc. Cf. Santa Fe Institute and
mezoscopic epigenetics. Dance of “happiness” involves complex steps. Enjoy!