Requisite Variety for Decision Factories!

In today’s post, we continue the Mistakes of Mainstream Management (MMM) series and discuss the perils of ignoring Ross Ashby’s Law of Requisite Variety. Before we discuss the law, I’d like to first drive home its importance.

Stafford Beer, the father of Management Cybernetics (which he described as the science of effective organization) said it well:

Ross Ashby was a pioneer in cybernetics and the author of the book, ‘An Introduction to Cybernetics’. In the book, he introduced the concept of variety and used that to formulate his ‘Law of Requisite Variety’, which is simple, yet profound:

Variety, in this context, means the number of possible states or behaviors of a system or a situation. This means that the regulator of the system should have enough variety to absorb rapid changes in its environment (e.g. pandemic triggered chip shortage). External variety is always larger than internal variety. So the regulator must have the means to attenuate unwanted external variety and at the same time to amplify internal variety in order to stay viable.

Note that variety here depends on the ability of the observer - this is a key implication from second order cybernetics. These ideas were developed later by Stafford Beer with regards to organizational control and viability.

Table of Contents

• Mainstream management has let us down

• Attempts to amplify variety

  • Real-world example

  • Organizing for Complexity

Mainstream management has let us down

Technology was initially heralded as a means to boost productivity and grant us more leisure time. Yet, the paradox we face today is that tech. companies themselves are grappling with their own productivity challenges. Their employees often endure long hours, experience burnout, and struggle to maintain a healthy “work-life balance”.

There is a fundamental mismatch between the nature of modern knowledge work and the organization of it. That's because our current management principles, methods and practices are not able to cope with the complexity of our environment and our own internal organization.

In businesses run by executives (vs. founders), it is VERY easy to forget that the point of the game is to stay in the game - long-term viability. Not to optimize for a specific output variable like quarterly revenue or profit margin. They end up installing attenuators and amplifiers at the wrong ends which then cripples innovation and results in the corporation’s slow march to irrelevance in the long-term, while the executives collect cash and stock bonuses in the short-term.

Attempts to amplify variety

There have been multiple attempts to solve for this in the past. Amazon's ‘separable single-threaded’ teams approach is a popular example. A simple flip in perspective here is to organize around work (e.g. projects) and not job families.

Work happens cross-functionally anyways. Small cross-functional individuals (who are empowered to make decisions) who are neither regularly interrupted by external teams to get their work done nor juggling multiple “priorities” can obviously accomplish a lot. This is why Jeff Bezos famously said

Here's the excerpt/story from the book, 'The Everything Store':

I’ll now present a real-world example of how a publicly listed company organizes for complexity - in a unique way, and discuss it from the lens of Ashby’s Law.

Real-world example

While many pure-play software companies struggle with agility, Tesla, as a hardware+software company introduces on average 60 new vehicle parts each day and all the cars are yet certified road legal (homologation). But, many of its customers don't even know about the sheer variety of parts in their cars that may be different when the purchase date is different by just a day or two.

This is despite the fact that car manufacturing is a highly regulated industry - it deals with the lives of people and not just numbers in a database. This level of change might take traditional car companies well over 2 years to pull off. How can a car manufacturer iterate so rapidly under strict regulations?

Let's focus on how their employees interact.

• Tesla has a clear and compelling vision, which one might call as the aspirational “1000 years goal”. This goal is to ensure the survival and prosperity of humanity for the next 1000 years, by providing clean and renewable energy, transportation, and technology. This goal provides a direction for the company’s strategy and innovation.

• Tesla leverages Open Space Technology to form self-organizing teams around bold and specific goals, such as improving the energy efficiency of a heat pump, or reducing the cost of a battery. These teams are cross-functional, blending software & hardware skills, and have full decision-making authority, as long as they respect the 15 core rules of the company. These rules include safety, quality, customer satisfaction, and alignment with the vision.

• Tesla practices radical experimentation, testing multiple versions of its products in production at the same time, and collecting data and feedback from the customers and the environment. Experimentation is driven by thorough testing; testing is a central activity in every team and helps pass regulatory rigor with automation. Experimentation allows the company to learn quickly and to iterate and improve its products based on real-world evidence. Tesla also uses simulation and modeling to test and optimize its designs before building them.

• Tesla teams self-organize as mobs based on top projects that they can pick daily via a mobile app. No managers assigning tasks! Internal AI stacks use rich data collected over several years to help make economic decisions when it comes to project prioritization. This enables real-time budgets instead of annual budgets. AI not for AI's sake, but it is used to achieve rapid innovation rather than summarizing a long-email thread with an escalation!

• Elon Musk, when on the factory floor is typically working on the critical path of the top project - actively problem solving and learning as an equal to any employee. At given point in time there is one 'driver' for a mob and others around him/her are collaborators. The driver rotates every few minutes and when its say Elon's turn, he has to 'drive' and install doors or draw CAD diagrams or whatever the task that needs to be performed. Hardly any work is done alone by any employee. The multiplicity of perspectives towards problem solving enhances variety and idea generation.

Tesla employees naturally learn cross-functional skills (killing specialization) as they get exposed to various projects over their tenure. One of the goals is to make sure nobody at the company has a "daily" task. Any repeatable task that has been standardized is aggressively automated so that humans can focus on creative problem solving and decision making. More importantly, there are no performance ratings for any employee.

Tesla’s "operating system" (which also continuously improves) enables the company to generate and manage a high level of variety, both internally and externally. Tesla can respond to the complexity and uncertainty of the environment, and to the needs and preferences of the customers - that’s how it was able to increase production despite the chip shortages during the pandemic while its competitors were shutting down their factories. Tesla is also able to cope with the complexity and diversity of its own system, by empowering and engaging its employees, and by fostering a culture of experimentation and learning.

Not just at Tesla! Neuralink makes new hardware every 45min and SpaceX made 1,000 hardware changes to the world’s largest rocket in 69 days. Same organizing principles and same rapid innovation.

Organizing for Complexity

An insight from complexity science is that how things interact with each other matters far more than what they are. Two powerful factors influence productivity - the way the work is structured and the ability to learn lessons.

Navigating complexity by installing the amplifiers and attenuators at the right places is vital. It takes experimentation over top-down knowledge or "best practices"! People on the ground are the ones directly interacting with & servicing the customer or solving a problem - that's where the most learning and decisioning must happen.

Instead of pushing authority to where information is (people on the ground) many times our management systems push information up to "authority". Hierarchy based decision making introduces delays. Complex interdependencies between teams create further churn and impede decision making. As Douglas McGregor put it,

I'll end today’s post with the wise words of Stafford Beer:

If you are an executive, how many decisions are you making personally in any given day? Are you in meetings all day in a conference room with the help of a handful of people whose only job is to coordinate your meetings and take notes on your behalf (layers of abstraction)? Or are you busy walking the proverbial Gemba? Are you doing do “quarterly business reviews” or actively working on the “factory floor” along with your people, on the most important problems? Is real-time data being used to fund new innovative ideas or do people have to go through dozens of approvals at snail’s pace or worse, wait for next year’s annual “financial planning” cycle?

That’s it for this week. Go for a walk and think about how decisions are being made in your decision factory.