Ciberia

I came across the Conant-Ashbey Theorem recently. It states:

Every good regulator of a system must be a model of that system.

It got stuck in my head longer than it logically should – I guessed that there must be something to it that I’m just not seeing. Then we worked on a gamification project, and it started making sense. In plain english it could mean: To tune a system you need a good analytics model of that system.¹

Why is this important?

During the design phase of a project we rarely think about designing control mechanisms that allow us to fine tune our designs when they are out in the world. Instead, we unleash variables in the world that run wild and soon disappear from sight.

Design with an analytics model upfront

Analytics not only tell us if our designs are doing their jobs – designing with a good analytics model in mind gives us levers to pull when they are not.

Good consultants spend more time thinking about their clients’ challenges than the clients themselves – and making models is a big part of this.

It follows that at the start of each project we spend a lot of time making models. It’s the best way to get under the skin of a project. Models are great at focusing a group. We do it to understand situations better, and it helps our teams visualise and think through a design challenge from all angles. Model making sits at the core of product discovery.

The models that I’m referring to here aren’t deliverables, but they could be if we polish them up, but in most cases this is a waste of valuable time. I see models as necessary discardables, their purpose is to move you along, without them you’re not going anywhere.

Systems Approaches to Managing Change highlights two interesting things about models that are worth considering.

Models are subjective

As with any model, viewpoints are inevitably partial in the sense of being both incomplete and of being viewed from a particular or partisan perspective necessarily based on it’s particular purpose.

Subjectivity is good here, it gives you a place to start when faced with a blank canvas. The sole reason for making models is to change them. Good models change shape as soon as you start showing them to people because the partisan perspectives start breaking down when more minds start looking at a problem.

Models are wrong

Constructing a model is a practical way of visualising the key elements of a problem. Statistician George Box said, ‘Essentially, all models are wrong, but some are useful.’ Models are always wrong in that they don’t serve as detailed illustrations of the problem. This is also why they’re right. The simpler you can make the model, the easier it is to understand a problem.

Models lose their power when they grow too complex, and when they are ‘finished’ they no longer draw ideas out of us.

Making better models

It turns out that the two weaknesses of models are also their strengths. But it takes courage to take the first step and risk being subjective and wrong. But if you think about it, when you start exploring, you can’t start out any other way.

Jared Spool’s Beans and Noses caused quite a stir when it re-surfaced and worked it’s way around our office. It’s the perfect metaphor to describe the consultant’s dilemma when working with big clients:

The idea is blindingly simple, actually. Every so often, you’ll run into someone with beans who has, for no good reason, decided to put them up their own nose. Way up there. In a place where beans should not go.

What do you do?

Lean Enterprise: How High Performance Organizations Innovate at Scale takes a more diplomatic approach framing it as the planning fallacy:

Due to a cognitive bias known as the planning fallacy, executives tend to “make decisions based on delusional optimism rather than on a rational weighing of gains, losses, and probabilities. They overestimate benefits and underestimate costs. They spin scenarios of success while overlooking the potential for mistakes and miscalculations. As a result, they pursue initiatives that are unlikely to come in on budget or on time or to deliver the expected returns — or even to be completed.”

The planning fallacy is a means of managing uncertainty by spinning ‘scenarios of success’ at the outset of a project. It’s easy to fall for this delusion because we don’t like thinking about failure, or admitting that we may be acting before having explored different options sufficiently. What can we do about this? The Lean Startup process can help:

The Lean Startup process being relatively cheap, in an enterprise context we can pursue multiple possible business models simultaneously using the Principle of Optionality.

The Principle of Optionality simply means that by investing limited amounts of time and money in small experiments we can investigate more ideas simultaneously:

…the principles of constraining time and resources, thus limiting downside, and building a minimum viable product to test your value hypothesis as soon as possible with real customers should be applied at the start of every endeavor.

The Principle of Optionality: building and testing minimum viable products simultaneously. Most will fail, but the probability of a big win increases. ¹

All roads lead to a minimum viable product, and testing a minimum viable product with real customers is a rational means of weighing up the gains, losses, and probabilities when things are uncertain. It is the antidote to delusional optimism, and will increase the likelihood that beans end up in the ground.

 ↩