Only Variety Can Absorb Variety

I recently read the expression “only variety can absorb variety”, but I had no idea what it meant. It took me some time, but I figured it out. It is a short version of the Law of the Requisite Variety by William Ross Ashby (on the right), a pioneer in Cybernetics: “The larger the variety of actions available to a control system, the larger the variety of perturbations it is able to compensate.” (source: Principia Cybernetica). In layman’s terms it means the more flexible a system is, the better the chances that it is effectively able to react to change. For example, the more languages I know, the better my chances are to find my way in a foreign country.

Ashby’s Law

The Law of the Requisite Variety is often referred to as Ashby’s Law. Ashby once said: “Only variety (in the regulator) can destroy variety (in the system being regulated)”, which sounds like quite the opposite of what I was talking about before. This discrepancy comes from the fact that there are two ways to deal with variety: either by change - absorbing more variety -, or regulation - controlling the input as much as possible. Ashby’s quote refers to regulation, whilst the definitions from the first paragraph are referring to change, which is more popular nowadays. Nevertheless, Ashby’s quote helps understand the context where the Law of Requisite Variety can be applied: there are at least two systems, they both have a certain level of variety, and they are communicating.

Understanding variety

I’ve been writing about variety, but never really defined what it is. According to the Principia Cybernetica: “Variety is a measure of the number of distinct states a system can be in”. Although it is a result of measurement, it is usually hard to quantify precisely, but in most of the cases an uncountable adjective does the trick. For example, the variety in our team is high.

However, in my language example above it is easy to provide the variety as a number: it is the number of the languages I speak. If I speak two different languages, then the variety of my system in the context of spoken languages is two. If my friend joins me, who speaks three different languages, one of which I don’t speak, then our system’s variety in this context is three.

Communication between systems

Charles E. Osgood talks about communication between systems: “the amount of output from a system is limited by the variety possible within the system and/or the variety of input to the system. The number of possible alternative communications between the two systems is limited by that system having the fewest output alternatives and/or the fewest input alternatives.

If my friend and I are visiting a country where at least one of the languages we speak is spoken, then most probably we’ll find our way (the number of alternatives are greater or equal to one). If we move to a part of this country where there is a strong dialect, we might have communication problems (the number of alternatives are zero).

In order to be able to continue our journey we need a new member to the group (system), who understands both that dialect and us. So, we have to increase the variety of our system. This is what the phrase “only variety can absorb variety” means: we cannot really respond to a new context or a change without increasing the variety of our system.

Absorbing variety

Absorbing variety is one way to respond to change. Let’s say I’m the leader of a team. I tell them what they have to do, and if they do something else, I’ll punish them. It is a classical command and control approach. If a change happens, according to Ashby’s Law, I have to increase variety in my organization (the system) in order to be able to respond. Since I’m keeping the whole team under my control, the only way to do this is by increasing the variety in myself. However, that is very expensive, and even if I’m a super human I cannot keep up with the pace of the change for a long time, so I need a different approach.

By giving up the command and control approach and introducing empowerment, I can open up the potential in the team and use the variety from their background, knowledge, etc. to absorb the necessary variety.

Ashby’s Law can be applied to business scenarios, too. In the next example, I’m running a hot-dog stand and the business is going well. I have more and more customers, but I reached a state where I cannot create the hot-dogs any faster. Due to the longer waiting time in the queue, certain customers stop coming back. The current number of customers is satisfactory, but it could be more, so I hire an employee to help me out.

Regulating variety

I mentioned before that there are two ways to deal with variety, and regulation is also an option. Let’s see the previous situation from a different angle: I’m satisfied how business is going. I’m very uncertain about the future, and I don’t know how the market will look like in three months. So, instead of hiring a new employee, I control the variety by continuing the business alone.

Depending on the context, regulation can be achieved by specialization as well. My customers are asking me whether I plan to sell beverages. I say no because of the previously discussed uncertainty, and keep being specialized in selling hot-dogs only, but I recommend a friend’s store where they can buy beverages.

Cost of response

It is important to know that even if the input of our system is regulated, the other system’s unmatched variety is still there, and it will be the input to another system. What if the customers, who were asking for beverages, find another hot-dog stand in the neighborhood, which sells canned sodas? Unless my hot-dog is extraordinary, they’ll buy hot-dogs from that stand and I’ll lose money.

As always, the response (absorb or regulate) depends on the cost or the risk I’m willing to take. Absorbing variety can be very expensive, but usually can lead to a better outcome. On the other hand, regulation seems to be a better option in the short term, but may have more disadvantages in the long term (note: I haven’t found any scientific evidence to back these statements, but somehow they sound obvious to me).

System overload

Until this point I was writing about the stage when the required level of variety is lower than the expected level. Robert Dilts calls this stage stagnation. When the system reaches a level of variety that matches the level of the variety of the other system, then the system is stable. When it is way above it, the system is overloaded.

I decide to sell beverages and I start big: I provide different kinds and in different kinds of packaging (I didn’t do any research and have no idea what they actually want). It is a huge investment, and what if my customers want only a can of soda (the one they can hold while eating the hot-dog)? Then my system is overloaded and this may have a huge effect on my business, because of the unnecessary investment.

Ashby’s Law in software development of the 21st century

Understanding Ashby’s Law can be very useful in software development, because it is all over the place. Scrum introduced the cross-functional team idea so that teams can adapt to change faster, eXtreme Programing emphasizes the importance of craftsmanship so that developers can provide multiple solutions to a problem, Kanban limits the work in progress to control the system, and in order to avoid system overload Systems Thinking is talking about understanding the demand.

comments powered by Disqus