Chapter 1: History is about data, not dates
I myself stood in front of the classroom as a history teacher for more than 20 years. Even though political history was never really my thing, as a teacher you obviously cannot avoid it. Important historical phenomena like the First World War, or in my case the Persian and Peloponnesian wars, you simply could not avoid, given the curricula. You try to make students aware of the causes, the course and the consequences. The course involving chronological events and thereby a lot of “dates” was of secondary importance. The causes and consequences got the focus.
Take a contemporary example: the war in Ukraine. As with the Peloponnesian Wars, which involved various alliances, national interests and political tensions, the causes of the war in Ukraine are similarly multifaceted. These include geopolitical tensions, national security and international relations. These topics can be quite complex, but they are essential for understanding the current world situation.
As a lecturer at UCLL (University of Applied Sciences, Leuven), I teach domestic and foreign students about data, analytics and AI. Some look odd when I say that I am actually a historian teaching about high-tech stuff. But the differences are less than you think.
Historical causality as a data problem
As a historian, you work retroactively; you study phenomena when they have already passed. You keep away from “predictions”, because you don’t want to end up in the same slide as Nostradamus.
But nevertheless, you often hear the comment that learning history, helps us to prevent repeating the same mistakes of the past. Whether that is so, I will leave that aside.
Data problems in statistics and data analysis, and by extension AI, are essentially about the same thing. You select variables (=causes) that influence a particular result (=effect or consequence). By extension, you can say that most things that occur are essentially data problems.
A student who never comes to class and does little or no studying has little chance of succeeding. Not coming to class and studying little are the causes, the independent variables that affect whether or not you succeed, the “consequence”. So, as a student, you can almost already predict how your behaviour will affect the outcome.
You can almost already
predict how your behaviour
will affect the outcome.
Even in the case of the war in Ukraine, the actors (the agents) are constantly making such trade-offs. They look at past events and constantly try to predict how a current decision might affect the final outcome. Suppose Ukraine were to fire missiles at Moscow, how will Russia react? That is predictive analysis that affects decision-making.
Predictive analysis affects
decision-making
Of course, you don’t take only one possible cause or variable into account, but often several at once. One decision can also have one or more consequences. You don’t have to wait for the event to occur, you can already draw an inference from such an analysis.
Non-linear processes and complexity
In data science and AI, a linear problem is one in which a change in input leads to a proportional change in output. For example, consider a linear regression where an increase in study time directly correlates with higher exam scores. The selling price (dependent variable, consequence) of a house correlates with multiple indicators or independent variables (=causes) such as the size, location and insulation value of the building. The bigger the house, the higher usually the price.
An example of multivariable linear regression https://www.leerschool.be/experiment/lren/
Of course, selecting the variables here is paramount. A student may be capable or smart enough to achieve good results without taking the class. Or maybe attending the class doesn’t really add value because the lecturer or professor is just reading from the course.
If we take the house example, you can see that there is indeed a relationship between size and selling price, but several independent variables affect the final price. A small house in New York will be much more expensive than a dilapidated castle in rural Hungary. Nevertheless, you can start using multivariable linear regression to predict the price of a house quite accurately. The algorithm in question is going to weigh all indicators against each other, learning from data (=data) from the past, i.e. houses already sold, to predict the price of another house. But as mentioned, the choice of independent variables is paramount. The date of birth, gender or nationality of the current resident of the house are not indicators that will affect the selling price. If you were to select such variables, it quickly leads to bias in the results.
As you can see,
AI and data science
learn from history.
Non-linear problems in data science, such as those addressed by logistic regression (“War or not”) or neural networks (we will talk about this in a later chapter), show that changes in inputs do not always produce direct or proportional changes in output (=effect). This complexity is common when modelling human behaviour or economic markets where multiple variables interact.
In history, we see similar non-linearity. Take the fall of the Roman Empire as an example. A combination of economic problems, military pressures and political corruption led to a collapse that was not simply the result of a single cause or a linear construction. The interaction between these factors is complex and the outcome is not linearly predictable.
Complexity theory and history
Complexity theory helps us understand how such non-linear interactions can affect systems. The theory states that systems consist of many parts that interact in many ways, leading to unpredictable outcomes. These systems can organise and adapt themselves, which contributes to their robustness but also their unpredictability.
Not entirely clear? No problem, in a later chapter we will explore complexity theory and chaos theory and how they can be very useful for history research as well.
Emergence in Historical Perspective
Emergence, a key concept in complexity theory, refers to the appearance of new properties or behaviours that are not evident from the individual parts of the system.
The formation of national identities or the spread of cultural movements is a historical example of emergent phenomena, which cannot be derived directly from people’s individual actions, but from collective interactions.
In the context of complexity theory and systems science, emergence (or emergent behaviour) refers to the phenomenon where larger entities, patterns or behaviours emerge from the interactions between smaller or simpler entities that do not possess these properties by themselves.
In popular parlance, people often say:
the result is more than the sum of its parts.
This concept is crucial in many scientific disciplines, including biology, sociology, economics, and, of course, artificial intelligence and computer engineering.
Characteristics of Emergence
1. New properties: New properties that appear at the level of the whole system are not observable at the level of individual components. For example, consciousness can be considered an emergent property of the complex networks of neural activity in the brain.
2. Complex interactions: Emergence often occurs in systems that have complex interactions between their components. These interactions lead to new patterns or structures that are not directly inferable from the properties of individual parts.
3. Bottom-up modelling: Emergent behaviour is often modelled or understood from a bottom-up perspective, where the behaviour of the whole is explained from the interactions of its parts, without external guidance.
Examples of Emergence
— Biological systems: The way ants work together to perform complex tasks without central control is an example of emergent behaviour in biology.
— Economic systems: Market dynamics, such as price formation, can be seen as emergent phenomena arising from the interactions of many individual consumers and firms.
— Artificial Intelligence: In AI, especially within neural networks, emergent behaviour occurs when these networks can perform tasks or solve problems in ways that are not explicitly programmed, as a result of the interactions between individual neurons.
Emergence is a fascinating concept because it helps understand how complex systems function and evolve without explicit instructions or control, and how simple rules can lead to unexpected and often ingenious solutions to complex problems.
History has no linear timeline
History is rarely a straightforward story. It involves the interaction of a wide range of factors and variables-social, economic, cultural, technological and personal-all of which combine to form the intricate web of causes and effects. These factors are often interconnected in ways that cannot simply be described as linear.
Examples of Complex Historical Events
1. The French Revolution: While it is easy to see the revolution as the direct result of royal decadence and economic crisis, in reality it was the product of a complex interplay of ideological, social and political forces. The influences of the Enlightenment, the financial pressures of warfare, and the growing discontent of different social classes, all played a role in how the revolution eventually broke out and evolved.
2. The fall of the Berlin Wall: This event is often seen as the direct result of political speeches and policy changes in the late 1980s. However, a deeper look reveals a complex mix of economic problems within the Eastern Bloc, growing protest movements, international pressure, and a gradual shift in Soviet policy over a longer period.
The Risks of a Linear Interpretation
A linear interpretation of history can lead to a superficial understanding of important events and movements. It can create the illusion that the past is a simple series of clear cause-and-effect relationships, often resulting in a misinterpretation of motives and consequences.
A deeper understanding of history requires a recognition of its complexity and the multiplicity of forces at play simultaneously. By recognising the complexity of historical events, we can not only develop a more accurate picture of the past, but also be better prepared for the complexity of current and future challenges. It is essential that education and research embrace this approach so that future generations can develop a deep and nuanced understanding of history.
Practical Applications and Interdisciplinary Approach
Complexity theory has practical applications in historical research by helping to model the interaction of economic, social, and political factors and predict possible future trends based on these complex interactions. This understanding enables historians to make scenario analyses that can help understand the likely outcomes of certain actions in history.
In teaching, an interdisciplinary approach that integrates history and data science can equip students with the tools to analyse complex historical data using modern data techniques. This includes using statistical software to analyse historical trends and applying machine learning to develop predictive models of historical events.
By integrating these elements into your text, you can make clear how history and data science overlap and reinforce each other, and how understanding complexity and emergence is crucial for both historical understanding and contemporary applications.
In a subsequent chapter, we study how complexity theory and chaos theory can help us model historical processes.
Read the next chapter on timelines:
https://medium.com/@krismerckx/timelines-or-quantum-history-59cdb8d42d2f
From my upcoming book, “What about history?”
