TUM-Meetup Lecture Transcript
Data has rapidly become an indispensable strategic asset at most organizations. Although the use of data is increasing dramatically, data storytelling remains poorly understood. As a result, many insights fail the last mile. They do not make the necessary step from exploring to explaining. This is where data storytelling comes in. Data storytelling is much more than visualizing data. It’s about explaining insights, putting them into a narrative, into a context that is well known to the audience and therefore resonates with them.
„Humans think in stories rather than in facts.“
I very much appreciated you quoting Harari, a colleague of mine. “Humans think in stories rather than in facts, numbers, or equations, and the simpler the story, the better.” I would like to add one more quote: “Humans are a post-truth species.” So that’s in a nutshell the essence of storytelling. We are not talking about real things. We are talking about fictions.
I’m very happy to talk today to you about data storytelling. And I am especially honored to talk to data brains about this topic. So let me just send ahead: I’m no data brain at all, I know only little things about statistics and mathematics. Like Harari I used to study history. And this is where my approach of data storytelling comes from.
So how does the fictional approach of storytelling fit in with the strive for objectivity in data science? Let’s have a quick glimpse on human history and the question, what makes us, the human beings, so called rulers of the world?
The superpower
Around 40.000 years ago, our ancestors started talking about things that did not exist. The walls in an Indonesian cave display these paintings. On the left you can see funny beings – half human half animal – hunting an animal that could be a pig.
Some researchers think the scene could be the world’s oldest captured story.
So, what happened? Around their campfires humans started to talk about their life. It was the life of a hunter gatherer. So, hunting scenes were the most common topics. But their paintings did not represent a reality, they seemed at least to be some work of art with non-realistic figures and scales.
We don’t know what stories they told each other. There was no writing yet. But we think that these paintings represent the first signs of a different behavior of humans compared to all other creatures.
These stories were not just entertainment during long campfire nights. They helped to store knowledge. They helped to share common experiences, to build networks of groups with common stories and beliefs.
From an evolutionary perspective our brain is nearly the same as that of our ancestors 40.000 years ago. It works in the same way. And this means, that today stories work in the same way as they did in very early times: they activate different regions in our brain. And the more regions they activate, the better the impact of a story is. They synchronize between individuals, so all of them can share the same experience.
In order to make humans the rulers of the world, storytelling was the first and foremost ingredient. But that was not enough. The way of the human species to world domination was marked by a series of inventions in network-technology: writing, printing, broadcasting, internet, and social media. These inventions helped to enlarge the coverage of human networks. And they helped to store all the knowledge or at least that, what we understand as knowledge.
Only with these technologies we were able to connect millions and billions of people. These technologies were the precondition for the emergence of larger organisms like tribes, religious communities, states and supranational-entities.
The playground
What exactly did we create with this storytelling-superpower?
We created an inter-subjective order. What does that mean? To understand this, we need to clarify the difference between objective, subjective and inter-subjective.
- Objective is a phenomenon that exists independently of human beliefs. That’s for example gravity. Gravity exists if we believe it or not. We can mesure it, as Galilei did when throwing an apple or something else from the Leaning Tower of Pisa.
- A Subjective phenomenon is something that exists only in a single individual. It disappears when the individual changes his belief. I have two sons. When they were young, at the age of 4 or 5, they had imaginary friends. Friends that existed only in their minds. They had real names, they played funny games together. They had a good time with one another. But they disappeared when the boys got older.
- Inter-subjective is everything that exists within a communication-network. That links the individual with many other individuals within this network. If a single individual changes his or her beliefs, this has no impact on the network. Only if the majority of a network changes its mind, the whole story could change. I mentioned gravity as an example for an objective phenomenon. Gravity exists independently of human beliefs, but the theory of gravity exists as an inter-subjective story within a scientific network. And from there it connects eventually to Einsteins theory of relativity. But an inter-subjective phenomenon means much more: nearly everything in our world is built on inter-subjective agreements: a religion, a state, a nation, a legal system, money, fake news, even bitcoin. By the way: the bitcoin network is a good example for creating inter-subjective truth. If 51 percent of the bitcoin-network agrees that a transaction is valid, it will be seen as true and executed and recorded in the blockchain.
As you can see, nearly everything we are talking about today is inter-subjective. And the tool for creating and negotiating these inter-subjective things is storytelling.
So how does data come into the story?
Let me put it like this:
Storytelling is data processing for humans
Storytelling is data processing for humans.
Stories are a form of aggregated or refined data. They are stored and distributed within our network. The raw data comes as impressions through our senses.
And here data in our modern understanding fills in. Data has rapidly become an indispensable strategic asset to most organizations. We now have billions of new data sources, from servers, smartphones, machine-sensors and many more. The raw data does not come in stories, but in zeros and ones. They are mostly digital assets.
As we don’t have sensors for zeros and ones, we use computers to process this data. And we need you, the data brains, to make sense of this data. So, in my view data storytelling is augmented storytelling. What does that mean?
We now not only have our seven senses. With the industrial revolutions we have billions of more distributed senses, such as motion sensors and accelerometer in our smartphones, temperature sensors in machines, tracking sensors or “cookies” on the internet and so on. And we are now connected to all of these sensors. But how to make sense of it? How to integrate this amount of raw-data into the human world of stories?
About stories and narratives
To make things a little bit more complex, I would like to talk about stories and narratives. For most of us, story and narrative are the same, interchangeable. But I think, there are differences: Story, in its most simple form, is about a character and the things that happen to them. Often stories are described as a hero’s journey, or as Michael Lewis put it, “a story is — people and situations.”
A story has a beginning, middle, and end. The hero goes out into the world, faces adventures and returns with a learning. There we explore the desire, dilemmas, and choices that a character faces. Think of a story as an anecdote, that recounts specific moments, with a specific time and place. It provides us with entertainment, insight, or even a lesson on life. It creates a shared emotional experience that can bond us together. A customer story. A “values in action” story. A new employee on-boarding story.
The problem is: In business, we are swimming in a sea of stories.
Everybody has a story. From the boardroom to social media. Billions and trillions of stories. In such an environment how can you make sense and meaning of things? How can you get everyone aligned around a shared common story?
Here is where the narrative comes in. In contrast to story, the narrative is much bigger. It’s a way of looking at the world, of giving meaning to everything. That’s the role and power of narrative.
An overarching concept that influences thought, meaning and decision-making. A symbolic frame like the American Dream or the idea of enlightenment. It doesn’t necessarily have a clear beginning, a middle, and an end like a story. Unlike a story it is still unfolding over time. Which means: the end is not yet written. Narratives are aggregated stories. Or, to use an analogy: If stories are like a pearl, then the narrative is a string of pearls.
It’s what connects the dots at the big picture level. A good narrative, will use a range of stories to illustrate, animate, and validate its message. Narrative gives meaning to a broader vision, a view of what’s possible, and why we should head in that direction.
When I started studying history, in the 80s, it was the time of postmodernism. In “La condition postmoderne” from 1979 Jean-Francois Lyotard spoke of “a crisis of meta narratives”. And by meta narratives he meant the great narrative of the modern world, will say the idea of progress and universal history.
Now we have become used to this status quo. Our world is fragmented, is characterized of a coexistence of different narratives. Neither enlightenment nor democratic liberalism nor religions would have the power today to explain the world and to create a consistent picture. Perhaps the new narratives once will emerge from megatrends like climate change, globalization, digital transformation, mobility, urbanism and so on.
The Oracle and the Virus
How do does data-storytelling fit into this setting?
I would like to present you two patterns that explain how data storytelling works.
The Oracle and the Virus.
The oracle pattern is a very old cultural skill. It was the main cult during the time of the Greeks. The oracle of Delphi was once seen as the middle of the world. The most powerful and important people came there to learn about their future. They hoped to get valuable insights by Apollon, the god of wisdom.
The recipient of the wisdom was a medium called Pythia. She was said to be in touch with Apollo. But to translate the divine wisdom into human instructions, the oracle used the help of the priests. They translated the messages into predictions. Such as: “If you cross the halys, you will destroy a great empire.” It was Croesus, the king of Lydia, who then crossed the river and destroyed a great empire. His own. But that’s another story, one of handling predictions.
So, times are changing, but the oracle-pattern still is the same. In our western world there is just one difference: Wisdom is not derived from divine input any longer, but from science. And one of the most important sources of science is data. So, we have the similar constellation of prophet and priests, of recipients of the wisdom (Pythia) and their translators (priests). Today the job is done by scientists, and especially by data brains. They explore the data and find relevant insights to make predictions.
So how do these insights spread and change human behavior?
At this point, the second pattern comes in: the virus.
In the end of 2019, just before the outbreak of corona, Robert Shiller published his book “Narrative economics.” In this book he analyses narratives and their impact on human behavior. I strongly recommend it. He looks at the narratives of the gold-standard, the American dream, the great depression, and bitcoin. What makes this book so exciting is the analogy, that these narratives can be seen as viruses. They spread into the networks, they mutate, and they are more or less contagious. He shows that only some mutations of a narrative go viral, and some of them do not. Furthermore, he shows that going viral can be very slow – over decades as for the American dream narrative – and very fast, as for the bitcoin narrative.
Together, both patterns, the oracle and the virus, can offer a model on how data-storytelling works: exploring data, finding insights (the oracle) and spreading them into the network (the virus). The virus pattern connects the individual story to a greater narrative. That’s the mechanism of spreading ideas and making sense of these insights by connecting them to the narrative. The more contagious the virus is, the more sense making power it has.
John Snow and the cholera map
To give you an example, I want you to tell the story of John Snow and the cholera map.
London 1854: the population of the largest city in Europe suffered from a severe cholera outbreak. It was part of a worldwide cholera pandemic. At that time, London was the center of the industrialized world. And the Soho district was it’s machine room. So everything that was dirty and smelly was there: Slaughter houses and grease-boiling dens lined the streets. Animal droppings, rotting fluids and other dirt were all around. And there was no sewer system.
On the last day of August 1854 a major outbreak of cholera occurred in Soho. Within ten days, 500 people died. What did John Snow do? The doctor mapped the death by cholera. He marked every death with its exact location on the map. As you can see, there is a cluster of cholera cases in the center of the map. This is where the water pump of the Broad street was located.
There was another pump in the northern part of Soho that belonged to a brewery (you can see it on the top of the map). None of the workers contracted cholera. There people drank beer instead of water. And that saved their lives. The map showed a clear correlation between beer drinking and survivorship. But was it also causal inference?
Unfortunately not.
In drawing the map, John Snow could visualize and explore the data. By locating every death he found a cluster of deaths around the Broad Street pump. John Snow came to the following conclusion: The water in the Broad Street pump was contaminated by germs that transmitted cholera.
So, with the map we have an important insight. But how did the insight go viral?
The greater narrative
Now, let’s see, how this story could be constructed and how it could connect to a greater narrative. As mentioned earlier, a story has a beginning, a middle part, and an end. This is the structure of a story reduced to the minimum. So much has been written about stories and their structure since Aristotle. Here we have the classic three-step.
We start with the setting – with giving information about the background and the hook. From there, the story leads to one central insight. And from there we draw the conclusion.
To understand why John Snows perception was so important, we need to know more about the context of the debate, or the narrative it fits in. At his time, there were two competing narratives on the causes of cholera. The theory of Miasma and the theory of Germs. The theory of Miasma states that particles travel through the air and infect individuals, and thus cause diseases like cholera. If you imagine the stench in Soho, this theory seemed to be very likely.
The germ theory of disease holds that the principal cause of cholera was a germ cell that traveled through water. And John Snows map seemed to confirm this theory.
To make this insight go viral, he needed to bring it into the current context, the narrative about miasma and germs.
To set the context and the hook, we start with the theory of miasma and the problem, that it can’t really prevent death by cholera.
Next, we deliver the most important supporting insight: the map of the death located around the pump.
The central insight is that death does not travel through the air, but through the water.
The conclusion is to build a sewage system that does not contaminate the water supply.
With this narrative, John Snow was able to convince the scientific community and make an important contribution to establish the germe theory or narrative. Above all, he was able to convince the health authorities of London to invest a lot of money in building a sewage system. The first and most important project of its kind at the time. Based on these insights and connected to the larger narrative, he was able to move decision makers to action. In a nutshell, this was to be the last cholera epidemic in London.
How to rule the world with data storytelling?
As we have seen, Humans think in stories rather than in facts. These stories are embedded in greater narratives. Today there is no single narrative, that all people agree on. Rather, it is a complex mishmash of narratives. They sometimes interact with each other, sometimes ignore each other.
Data storytelling is successful when both patterns are in action:
- The oracle pattern as the way to new insights.
- The virus pattern as the way to spread these insights and get people to act.
To make the insight go viral, it’s necessary to find out more about the surrounding narratives. Learn as much as you can about the target group – the potential host, his or her surrounding narratives, the values behind it.
To achieve this, starting with analyzing search queries and the social media, definitely makes sense. And once the specific use case is defined, there are much more data sources coming up for sure. Build stories that connect with existing narratives. You have the toolset to find out more about the connection of stories and narratives. And you have the tools to create new stories. Maybe gpt3 is one of them.