Simplifying strategy

9 May 2021 Filed under: knowledge

Simplifying strategy

We know that less is often more. So why we still create overly complex models and strategies, and how do we simplify them?

Dan Ward observed countless engineering and R&D projects during his time with the United States Air Force. He noticed that the simplest projects were most likely to be successful, in every domain: the simplest logistical plans, the simplest plane designs, the simplest components; they always ended up being more reliable, more efficient, and more successful than more complex alternatives. Ward mapped the relationship between simplicity and efficacy, and how those two qualities interrelate in the lifetime of a process; he called it the simplicity cycle.

In the beginning, when designing a system – whether that’s a new fighter jet or a spanner, a project management process or a marketing plan – we add a bit here and a bit there, and the system improves with each new addition. The system is getting more complex, but it’s getting more effective too; the two qualities go hand-in-hand.

Diagram showing the relationship between complexity and efficacy — In the early stages of a project, increases in complexity can bring with them increasing efficacy.

(The reality is never quite as straightforward as this – it’s usually more of a messy, looping spiral as experiments fail or succeed – but the general principle holds.)

This path, though, cannot be continued forever. The upper-right of the graph is inaccessible. The most effective design is never the most complex design. There comes a point at which any additional complexity will make the system worse: less usable, less efficient, more wasteful, less reliable, harder to understand when it’s working and harder to fix when it’s not. As its designer, at that point you’re faced with a choice of two directions: the path of complication, or the path of simplification:

You can continue adding to the design, making it more complex and in doing so make it worse. Or you can begin the process of subtraction and simplification, and potentially improve it.

Not all subtractions will make the system better, of course. It’s possible to take away something critical, and in doing so make the system simpler, but also less good.

In the 1940s, the engineer and science fiction writer Genrich Altshuller documented patterns of problems and solutions he’d observed in engineering, and ended up with a surprisingly complete catalogue. The resulting toolkit, known by its Russian acronym TRIZ,¹ contains a solution to exactly the problem of how to navigate down the slope of simplification: “trimming”. Simple and intuitive, trimming asks us to:

Remove a piece from the system
Test whether the system works
If it does work, discard that piece, since it was demonstrably unnecessary; if it doesn’t work, replace the piece
Repeat with another piece until you’re satisfied that the system is as simple as can be

It’s relatively straightforward to see how this process unfolds when you’re designing a physical object – a plane, a circuit board, a chair, indeed anything tangible. But it also holds for the design of more intangible things: strategies, plans, explanations of the world, processes within organisations, and so on.

In devising a strategy, you follow the same initial course that Ward outlines. You create a model or a narrative that you think explains the world in some way, diagnoses a particular challenge, and suggests a policy for dealing with that challenge. You discover new data and incorporate it into your model or narrative, making it more complex but increasing its power; it becomes harder to understand, but better at explaining the world, an acceptable trade-off. Things are getting both more complex and more good, rising up the initial line on the graph.

At some point, though, you hit the same impasse. You add more data and more complexity, but you don’t actually gain anything. You pay the cost of the extra complexity – your strategy gets harder to understand and to communicate – but without any benefit. It’s easy to see what following the complication slope looks like in that situation: add more data, more detail, more narrative. Look for more angles; apply your thinking to more situations. Keep pushing, keep adding. That way leads inarguably to overstretched, overly complex, impenetrable strategy: the sort of thing that Richard Rumelt uncharitably calls “fluff”, “a form of gibberish masquerading as strategic concepts or arguments.”

Luckily, we can use exactly the same trimming technique that makes sense in engineering. Remove paragraphs from your narrative, subjects from your qualitative research, data sets and variables from your models, threads from your arguments. Replace paragraphs with sentences, sentences with phrases, words with pictures and diagrams. Does your narrative still work? Is it, in fact, stronger? Context and complexity that helped you arrive at this point might no longer be necessary, and can be safely jettisoned; don’t hold onto it simply because it was useful once.

If the answer is so simple – to cull complexity – then why does complication seem to abound in the world? Unfortunately, it seems humans are predisposed to consider additive solutions over subtractive ones; they’re biased, in other words, against trimming. A study earlier this year by the researchers Tom Meyvis and Heeyoung Yoon found that, in a whole host of problem-solving scenarios, people reach for solutions that add elements – and therefore complexity – rather than take them away. They suggest several reasons why this might be the case:

“…the bias towards additive solutions might be further compounded by the fact that subtractive solutions are… less likely to be appreciated. People might expect to receive less credit for subtractive solutions than for additive ones. A proposal to get rid of something might feel less creative than would coming up with something new to add, and it could also have negative social or political consequences – suggesting that an academic department be disbanded might not be appreciated by those who work in it, for instance. Moreover, people could assume that existing features are there for a reason, and so looking for additions would be more effective. Finally, sunk-cost bias (a tendency to continue an endeavour once an investment in money, effort or time has been made) and waste aversion could lead people to shy away from removing existing features.”

That’s why we so often find ourselves in a pickle, wrestling with overly complex strategies and models that collapse under their own weight. We have an instinct to add, an instinct we must fight; only when we’ve trimmed all there is to trim can we be satisfied that our work is done.

Roblog