The Selfish Meme: How Organisational Memes Define Culture

What is "The Selfish Meme"?

My work as a consultant first for ThoughtWorks and latterly for Codurance has placed me within many complex organisations doing what we used to call Agile Transformations. That phrase has become a bit toxified now, so I’m not sure if ThoughtWorks still uses it. At Codurance we prefer to talk about Software Modernisation. Frequently, we have to battle against the organisation itself which seems to have some collective Borg-like intelligence that resists change. This is often referred to as the "Corporate Immune System". Sometimes we managed to "win" the battles and the war and effect positive and lasting change. Other times we seemed to fail in our transformation goals.

After reading the Selfish Gene and The Beginning of Infinity I wondered if I could draw on the themes in those books to better understand how company cultures can evolve and ideas and processes can persist, even long after the originators have moved on. This led me to define a framework for a mental model of an organisational culture in terms of memes that exist within the culture and started me on a journey to explore whether this mental model could be useful in helping us drive positive change. These ideas, their retro-fitting onto some earlier work and their application on some of our later work, are the basis of the Selfish Meme.

The Selfish Gene

The Selfish Gene is a book written by Richard Dawkins, first published in 1976. Dawkins later became probably more widely known (and certainly more controversial) for "The God Delusion" which was published in 2006.

In the Selfish Gene, Dawkins, a biologist, proposes that organisms (plants and animals) are not the thing that reproduces, rather the genes that they carry are the "replicators". Genes survive through successive generations in a way that the organisms do not. Genes even appear across several different species. So a "successful" gene is one that alters the characteristics of its host, which Dawkins calls a "survival machine", in a way that makes that gene more likely to survive into subsequent generations.

This book is the place where the word "meme" was coined. Dawkins’ definition of a meme is any kind of information that self-replicates, in the manner of genes, through some kind of vector (humans). The book was written in the pre-internet age, but his examples are still highly relevant in today’s society, such as popular songs, poems, beliefs, superstition, mnemonics etc.

The memes carry some value to the host, such as an easy-to-remember template for an action or perhaps just fun, and it is that value that causes the meme to be propagated from person to person, perhaps through generations.

Memes can mutate, an example being the well-known song "Auld Lang Syne" which is widely sung every year on New Years Day and apparently is usually sung wrong as it has mutated through generations.

Over time the value delivered can change or diminish or even go away altogether, but the perceived value to the host somehow remains. It is those types of memes, within organisations, that interest me.

Organisational Memes and the Selfish Meme

Back in 2017, I read a book called "The Beginning of Infinity" by David Deutsch. This was during the time I was studying quantum computers for ThoughtWorks and I’d hoped that this book (I didn’t read the blurb) would be similarly themed to Deutsch’s book "The Fabric of Reality". It turned out it wasn’t, but happily it is one of the best books I’ve ever read.

The theme of the book is how knowledge is passed along and allowed to develop and grow into new knowledge. Deutsch proposes that societies through the ages had (or have) memes that define the culture of that society. These memes are pieces of knowledge that find a way to pass through the generations.

There is a chapter in the book that contrasts the static culture of ancient Sparta with the learning culture of ancient Athens and talks about how the prevailing memes of those societies shaped their present and future by permitting or not permitting certain innovations.

The key point in this comparison is that the Spartan memes allowed the society to attain a good stability which for a long time ensured its survival. At the same time though, such memes inhibited innovation because they placed such a high value on conformity and stability. This part resonated very strongly with me and the notion of Risk Management Theatre that we see in many organisations. I wondered if there were memes floating around within organisations that support the status quo, work against useful innovation and keep alive obsolete and even harmful processes.

Understanding Organisational Culture through Memes

One of the earliest things I learnt working at ThoughtWorks was to ask, "What is the outcome that this process supports?" to understand why things are done the way they are. This works up to a point, but can lead to a dead end with "it’s just the way we do it here". I found that frustrating many times. At some point I started to ask, "Why is it that way?" This allowed me to form ideas about the history that led to the present. But that doesn’t explain why people still do the things they do.

Eventually I formed a conjecture that the fulfillment of a process can give some kind of value to the individual alongside, but possibly entirely independent of, the value (or supposed value) that the process returns to the organisation. So I started to wonder if there were organisational memes responsible for keeping alive processes and parts of culture that could be reasonably regarded as suboptimal or even harmful. So I think if we can identify the memes that confer value to the hosts and understand that value, this helps us to understand how the culture evolved the way it did.

Cultural Memes v. Organisational Memes

As we saw from the discussion on Sparta and Athens, it is possible that over time memes can mutate and evolve such that they are no longer useful or even become harmful. They still persist though, which means they must be delivering at least some perceived or real value to the vectors of the meme. So can they be modified or removed? Should they be modified or removed?

In an organisation, this evolution of memes exists but is much accelerated. For example, a meme could develop in an engineering group around change approval in pre-Devops days. The organisation got genuine value from this process, and the vectors (people) gained confidence around the success of their release as well as comfort that they had done their due diligence.

After the DevOps revolution, the meme still existed and was still passed along from person to person. The process that the meme endorses, on the other hand, has become obsolete and probably counterproductive. The meme that drives the process still gives comfort to the vectors, in the sense that they gain confidence that they can’t be blamed if the release fails, so the process lives on supported by the memetic knowledge.

Driving Positive Change in Organisational Culture

Just to be clear here, what I am presenting is a mental model of an organisation. Any mental model is by definition a simplification of the whole that picks out only the things that are relevant to help you understand some aspect of the behaviour of the system. So using the kind of mental model that we have come up with can help to understand what some of the levers could be to try to encourage positive change.

Clearly it takes a long time to change organisational culture in any meaningful way. Sometimes attempting to make big changes, even changes that most people would consider to be entirely reasonable in isolation, can cause so much upheaval and move so many goalposts, that they can start to harm the organisation’s existing capabilities. It isn’t too farfetched to suggest that people leave organisations because a blame culture was replaced (or an attempt was made to replace it) with a learning culture and many people feel uncomfortable with the kind of "touchy feely" rituals that this could involve.

The first thing I did was to retrospectively look at some of the work I had been involved in at ThoughtWorks to see if it felt like my original conjecture could have any merit within my own frame of reference. Clearly, if it didn’t stand up to my own scrutiny and my own inevitable confirmation bias, then it would be entirely without merit. I also asked some colleagues with whom I’d worked to give their opinions on the merit of what I’d done. Happily (and otherwise I wouldn’t be writing this), we collectively decided that there was enough here to justify carrying the idea forward. This helped us to put together a loose playbook which I have tried on subsequent engagements.

It had always been my approach to first understand the history of a situation before I try to change the status quo. I have always felt that it was necessary to understand the reason why something was there, what value it was perceived to be delivering, in order to reasonably suggest alternatives. It is always easy, perhaps lazy, as a consultant to suggest that "everybody should pair program" (itself a meme that pervades many organisations), but this can only succeed against a mantra of code reviews if it serves the same need to the vector (the person in the organisation) in which it is living.

I realised that I had to suggest alternatives in opposition to the existing meme that still satisfied the value to the vector but which delivered the real value to the organisation. In memetic and genetic terms, I needed to create an appropriate meme allele. So to apply the framework of "how to use memes to influence change in an organisation’s culture" the first stage was to identify extant memes that were not useful. Having done that, we set up a suitable alternative, an allele, in opposition to it. It is important to understand what value the vectors get from propagating the existing meme and make sure you can explain the alternative in those terms.

There are a few fairly obvious meme allele pairs, such as "Code Review" v. "Pairing", "Branching" v. "Trunk Based Development (TBD)" or "Very small, frequent releases tested as they are developed" v. "Large, infrequent releases tested after a code freeze". My key insight was that which was explained above. Whilst most people would probably describe a code review as something that guarantees quality or perhaps helps to share context, the real value to the vector is probably a degree of comfort in following what is believed to be the correct process and thus rendering oneself immune from blame for later failure.

Clearly, our goal in transformation is to reduce, or eliminate, the prevalence and therefore the (malign) influence of the existing meme and replace it with influence from our competing meme. Ideally, for example, we would like all of the technology organisation to be using trunk based development, and no longer be using a workflow based on branching and merging. I have had success and failure in the past at persuading people of the technology-focused benefits of using TDB over branching and merging. When I started to attempt to understand the personal motivations behind certain behaviours, I had more success by speaking to how those needs could be satisfied in a different way.

In the Selfish Gene, there is a description of a model of how two competing genes can exist in a population. This is based on differing evolutionary payoffs associated with the behaviours that the genes influence. The example used is how genes for always fighting rivals (hawk) can compete with genes for always running away (dove). If you always fight, you will win more fights and thus more resources, but will risk injury. If you always run away when challenged, you’ll "win" some confrontations against other doves and you’ll always flee the more aggressive hawks, but you’ll never risk serious injury. Clearly those outcomes have a value, or a penalty, associated with them.

It is possible to create conditions in which the payoffs and penalties, together with the initial populations, will cause the two genes to co-exist at some equilibrium point. Depending on how you vary the initial conditions and the payoffs and penalties, it is possible to cause the model population to be dominated by one or other of the competing gene alleles.

So when we combine the idea of competing meme alleles with some basic model of what the payoffs and penalties could be, we have a framework based on this mental model of how we can proceed in trying to get our new meme to dominate the existing (presumably to some degree malign) meme.

We can get two key insights from the modelling of gene pairs. Firstly, varying the initial weights within the population of the starting memes shows us that there can be a critical mass associated with one or both genes that will simply never allow the other to emerge significantly. Secondly, if we vary the payoffs or penalties associated with behaviours, it is possible to create the conditions under which one or other gene will always dominate.

These insights enabled myself and colleagues to form a model of why certain things had been successful in the past and certain other things less so. More than that, it gave us a template to look at new work which helped us to give ourselves a better chance of setting things up for success in the future.

Our first lever is that of the initial weights of the competing alleles within the population. We know that it is hard for a message (meme) carried by only, say, five people, to penetrate a technology organisation of, say, 60. Our modelling supports this. So in these circumstances we can see that we cannot immediately increase the prevalence of our meme throughout the whole population, but we can reduce the size of the population into which we introduce competition. So what we do is we form a new team consisting of our five people carrying the new meme together with a small number, say three, of the client’s employees who are presumably carrying the extant organisational meme. This then creates relative weights that allow our new meme to eventually dominate and eliminate the old one. At that point, the team can be split and two newer teams formed to repeat the process.

The other lever we have at our disposal is to change the values of the memes within the population. There are a few ways to do this. By reasoning out and being very deliberate about what we are doing and understanding the personal value conferred on the vectors, we can use our mental model to tailor our messaging to the wider organisation to speak to the perceived value to the individuals and teams within it.

Another approach is to understand how our technology solutions may help the vectors to propagate our new meme. This is analogous to increasing the payoff associated with our good meme. If we use Trunk Based Development v. GitFlow as an example here, we can show that the vectors’ daily pain is greatly reduced by not having to endure painful merge cycles at regular intervals. Provided that we also make sure that we service any other personal value needs, such as the comfort of avoiding blame, we have a powerful way of driving adoption of modern practices.

So in conclusion, what we have created is a mental model that allows us to build a framework around some tools and techniques that might help us to guide positive cultural change within an organisation. I can’t guarantee that this approach will always work, indeed I am far from having enough data points to even claim that it will work better than a straightforward prescriptive approach to change. I do, however, feel that I have seen enough to at least say that thinking in these terms and using this model has helped with some of the transformation work in which I’ve been involved.

About the Author

James Birnie has worked in software since the late 1990s when Agile and Lean were words you used to describe gymnasts, Pipelines were for carrying oil and TDD was something you studied but never did. Birnie worked for a startup for 9 years where he learnt the hard way about XP practices, transformations, splitting the monolith and continuous delivery. In 2015, Birnie started a new life as a transformation consultant which opened him up to a whole new world of learning and experience. Now, he works for Codurance where he attempts to drive positive change through Agile values, a DevOps mindset and a relentless drive to improve.