Eros & Thanatos: The Fatal Arrogance of Control
Humanity has never known so much about the world, yet we seem increasingly convinced that we know enough to control almost everything. We can manipulate atoms, rewrite genetic instructions, observe galaxies billions of light-years away, and build machines capable of performing intellectual tasks once considered uniquely human. Our achievements are extraordinary, but somewhere along this remarkable journey, we seem to have acquired a peculiar confidence that understanding parts of reality gives us sufficient knowledge to organize the whole.
Armed with scientific models, statistics, expert committees, regulations, algorithms, and an apparently inexhaustible supply of good intentions, modern societies increasingly attempt to eliminate the uncertainties of human existence. Every risk demands precaution, every undesirable outcome calls for intervention, and every failure becomes an argument for additional control. We have become remarkably skilled at constructing systems intended to make the world behave according to our expectations, while remaining surprisingly reluctant to question whether those expectations reflect reality.
The ancient Greeks gave us two intriguing figures: Eros, associated with love and the creative forces of life, and Thanatos, the personification of death. Much later, Sigmund Freud developed his controversial distinction between life instincts and a death drive, a framework subsequently associated with these two mythological figures. We need not subscribe to his psychology to appreciate the metaphor. Eros represents life’s capacity to connect, create, preserve, and develop, while Thanatos evokes the opposing movement toward dissolution and destruction.
But perhaps the most interesting possibility is not that these forces oppose each other. It is that one can gradually transform into the other.
Our desire to protect life is fundamentally life-affirming. We develop medicine, establish laws, build institutions, and organize societies because we want to reduce suffering and improve the human condition. Yet the same desire, pursued without restraint, can turn into an obsession with eliminating every uncertainty, every risk, and every possibility of an undesirable outcome. What begins as the protection of life gradually becomes an attempt to protect us from life itself.
And that is where things become interesting.
When the map becomes more important than the territory
Over a long career in information technology, I encountered organizations of very different kinds, from heavily regulated environments to fiercely competitive commercial operations. What fascinated me was how differently they dealt with reality. Some encouraged people to confront problems directly, exercise judgment, and adapt. Others seemed increasingly absorbed by their own structures, procedures, and internal measures of success. Yet both tendencies could coexist within the same organization, sometimes producing remarkable achievements, sometimes equally remarkable absurdities.
Anyone who has spent enough time designing or managing complex information systems knows that making computers perform calculations is rarely the greatest challenge. Understanding the reality those calculations are supposed to represent is considerably more difficult. We construct data models, establish relationships, define procedures, and create rules describing how things should work. Inevitably, reality turns out to be messier than the specification.
Good systems learn from that discovery. Unexpected behavior exposes incomplete assumptions, users reveal overlooked possibilities, and operational experience forces us to reconsider our designs. Bad systems respond differently. Exceptions become disturbances, users become obstacles, and reality is expected to conform to the original specification. When things go wrong, another layer of controls is added, as though sufficient determination might eventually persuade the universe to cooperate.
The software industry has spent decades learning this lesson, sometimes painfully. Iterative development, continuous testing, operational monitoring, and feedback loops emerged partly because complex systems cannot be perfectly specified in advance. Of course, the industry has also demonstrated a remarkable talent for turning even Agile[ML1] into an elaborate bureaucratic ceremony, complete with consultants, certifications, and enough meetings to leave little time for actual development.
Still, the underlying principle remains valuable: when reality contradicts your model, you should probably reconsider the model.
Our institutions are not always quite so enthusiastic about that idea.
Scientific knowledge, impressive as it is, remains fragmented. We understand many components of complex systems without necessarily understanding how they behave together. Economies, ecosystems, cultures, and human societies emerge from countless interactions that cannot always be adequately captured by a manageable collection of variables. Nevertheless, governing requires simplification. We reduce education to measurable achievements, healthcare to standardized pathways, prosperity to economic indicators, and human well-being to administrative categories.
These abstractions are indispensable tools. Science and engineering routinely produce reliable results without requiring complete knowledge of every underlying mechanism. The problem begins when we forget the boundaries of what our models can tell us, and especially when we mistake the resulting measurements for the reality they describe.
A map is useful precisely because it simplifies the landscape. But mountains do not disappear because the mapmaker has omitted them, and human beings do not become more predictable because a sophisticated model assumes that they are.
The political scientist James C. Scott explored this problem in Seeing Like a State, describing how ambitious attempts to improve society can fail when centralized institutions impose simplified representations of complex social realities while neglecting practical knowledge. The issue was not that planning or administration were inherently wrong, but that the confidence of those imposing the plans could become dangerously disconnected from the realities they sought to reorganize.
And once our abstractions become embedded in institutions, another complication appears.
When knowledge becomes power
An observation becomes a theory, a theory informs a policy, a policy produces a regulation, and the regulation eventually becomes an obligation. Along this journey, the uncertainties surrounding the original knowledge can quietly disappear from view. What began as a provisional understanding of reality acquires institutional authority, sometimes extending far beyond the circumstances in which that understanding was reliable.
There is nothing particularly new about this human tendency. Religious institutions spent centuries developing elaborate systems of authority around interpretations of existence, although religious traditions also produced profound intellectual achievements and reflections on the limits of human understanding. The interesting continuity is that the human appetite for certainty survives the institutions through which it is expressed.
Science, at its best, is an organized confrontation with ignorance. Its explanations remain open to challenge, its conclusions can be revised, and its greatest discoveries often reveal new dimensions of what remains unknown. Institutional power, however, operates under rather different incentives. Questioning an established model may threaten professional authority, political credibility, budgets, or the legitimacy of an entire organizational structure.
A scientist discovering that a model fails has an opportunity to improve understanding. An institution discovering that its policies fail may face a considerably less welcome prospect.
Moreover, the institution does not necessarily need to believe its model is correct. It may simply reward everyone for behaving as though it is.
Once success becomes defined by measurable targets, people organize their behavior around achieving them. Schools can concentrate on examination results rather than the broader development of their pupils, hospitals can optimize administrative performance while overlooking aspects of patient experience, and commercial organizations can celebrate quarterly targets while quietly damaging long-term customer relationships. These are possibilities, not inevitable outcomes, but they reveal how easily measurements intended to support an objective can gradually replace it.
The organization may become increasingly efficient at achieving something that no longer corresponds very closely to its original purpose.
And because rules also serve legitimate purposes, protecting citizens against arbitrary decisions, discrimination, and abuse of power, the solution cannot simply be to replace procedures with unrestricted human discretion. The real challenge is to construct institutions that remain accountable while preserving the judgment and flexibility necessary to deal with circumstances their designers could not anticipate.
In other words, institutions need structures that allow them to learn, rather than structures designed primarily to demonstrate that they were right all along.
Our growing intolerance of uncertainty
It would be convenient to blame this entire development on politicians, bureaucrats, experts, and technocrats. Unfortunately, that would let the rest of us escape rather too easily. We are not merely victims of the modern obsession with control. We are enthusiastic customers.
Human beings have always sought safety and predictability. Uncertainty makes us uncomfortable because it exposes the limits of our knowledge and influence. We prefer recognizable patterns, reassuring explanations, and the comforting belief that somebody understands what is happening. An uncertain future is difficult to accept, even when the certainty offered in its place is largely imaginary.
Civilization has achieved remarkable things by reducing uncertainty. Reliable laws, medicine, sanitation, infrastructure, and countless other innovations have made human existence safer and opened possibilities that earlier generations could scarcely imagine. This is Eros at work, preserving life and creating the conditions for human flourishing.
But the success of these achievements creates a peculiar temptation. As we become better at controlling particular risks, the risks that remain become less acceptable. What previous generations regarded as unavoidable misfortune increasingly becomes evidence that somebody failed to take the appropriate precautions.
We begin to expect protection not merely against foreseeable dangers, but against uncertainty itself. Politicians respond with promises, institutions with regulations, experts with frameworks, and organizations with increasingly sophisticated systems of compliance. Every failure creates pressure for additional intervention, while the costs of excessive intervention are often dispersed, delayed, and considerably less visible.
A visible accident can provoke public outrage and demands for immediate action. The opportunities quietly lost because rules prevent experimentation rarely attract comparable attention. Nobody organizes a public inquiry into the inventions that were never attempted, the relationships that never developed, or the solutions nobody was allowed to try.
Yet these invisible losses matter. Creativity requires experimentation, experimentation involves failure, and meaningful freedom includes the possibility of making choices others consider unwise. Human relationships require vulnerability, societies need diversity to adapt, and innovation depends on exploring possibilities whose consequences cannot be known in advance.
A society determined to eliminate every risk would eventually have to eliminate much of what makes life interesting, creative, and human.
Here the Eros–Thanatos metaphor becomes particularly uncomfortable. The desire to preserve life, taken to its extreme, can begin to suppress life’s essential qualities. We seek security because we value living, but we may gradually become so frightened of uncertainty that we sacrifice the freedom, spontaneity, and unpredictability that give living much of its meaning.
Thanatos need not arrive announcing destruction. He can come dressed as Eros, promising to protect everything we value.
And we may welcome him enthusiastically.
The technological temptation
Artificial intelligence adds a fascinating dimension to this predicament.
For decades, information technology has expanded our capacity to collect data, recognize patterns, automate decisions, and coordinate increasingly complex activities. AI now extends these capabilities into areas once considered beyond the reach of computers. Its achievements are remarkable, and its potential to improve human life is enormous.
Yet greater computational power does not automatically produce a more complete understanding of reality. An algorithm may discover intricate relationships within available data while remaining unaware of what was never recorded. A model can produce astonishingly precise calculations around questionable assumptions, and the sophistication of its output may conceal the incompleteness of the world it represents.
Over the years, competitive businesses have often been forced to develop information systems closely connected to customers, transactions, production, logistics, and financial outcomes. Reality keeps sending them feedback, sometimes in the rather unpleasant form of disappearing customers or shrinking profits. Yet even these organizations can become prisoners of their own performance indicators, confusing targets with achievements and internal success with actual value.
Institutions with weaker or more delayed feedback face an additional problem. Their systems may become extraordinarily sophisticated at recording administrative activity while remaining surprisingly poor at understanding whether that activity improves the lives of the people they supposedly serve.
This is where an old IT lesson becomes particularly relevant: collecting data is not the same as learning.
A dashboard displaying hundreds of precisely calculated indicators is not necessarily a feedback system. Genuine feedback occurs when unexpected outcomes lead people to challenge their assumptions and change their behavior. Without that ability, increasingly sophisticated information systems may simply allow organizations to become more efficiently wrong.
AI can help institutions discover patterns they previously overlooked and improve their understanding of complex realities. But it can also reinforce existing assumptions, automate poorly designed processes, and extend institutional control into areas where human judgment once provided essential flexibility.
Give powerful technology to an organization willing to question itself, and you may accelerate learning. Give it to an institution convinced that its model is already correct, and you may merely automate its mistakes on an unprecedented scale.
The technology is not the deciding factor. The willingness to confront reality is.
The danger of protecting ourselves from life
The deepest danger may therefore be neither bureaucracy nor technology, neither political ambition nor scientific overconfidence considered separately. It is our collective difficulty accepting that the world is not something we can ever fully understand and control.
Our institutions are human creations, necessarily based on incomplete knowledge, imperfect assumptions, and objectives that may conflict with one another. We need them to organize complex societies, protect individual rights, and coordinate activities far beyond the capabilities of isolated individuals. But we also need to remember that their authority does not make their understanding complete.
A society conscious of its limitations will create rules while leaving room for exceptions, use measurements without worshipping them, and pursue desirable outcomes without pretending that all consequences can be anticipated. It will encourage experimentation, respect practical knowledge, and recognize that local judgment often reveals things centralized abstractions overlook.
Most importantly, it will accept that failure is not always evidence that more control is required. Sometimes failure is information telling us that our original understanding was wrong.
The greatest achievement of modern science may not be the extraordinary power it has given us, but its repeated demonstration that reality is stranger, more complex, and less intuitively understandable than we imagined. Each major discovery expands our knowledge while revealing further dimensions of our ignorance. Quantum physics, evolutionary biology, and the study of complex systems have repeatedly challenged the comforting assumptions through which human beings once understood their world.
That ought to encourage humility, although humility has never been particularly fashionable among those exercising power.
Eros does not offer a world without suffering, uncertainty, mistakes, or conflict. Life has never made such promises. It offers the extraordinary possibility of creating, discovering, connecting, experimenting, and adapting in ways no authority could fully anticipate. Institutions can nurture those possibilities, but they cannot manufacture them through administrative instructions.
Tragically, our determination to preserve life may gradually become a determination to remove everything unpredictable from it. What begins as protection turns into restriction, what begins as knowledge becomes dogma, and what begins as a structure intended to serve humanity eventually demands that humanity adapt to the structure.
We may eventually create systems so effective at preventing undesirable outcomes that they also prevent new possibilities no one has yet imagined.
Perhaps the greatest danger of modern civilization is not that we know too little. Ignorance has accompanied humanity throughout its history, and somehow we have managed to learn an extraordinary amount.
The real danger is that we have become powerful enough to impose our limited understanding on the world, and arrogant enough to mistake that power for wisdom.
After all, a world in which everything is predictable, every risk is controlled, every behavior is accounted for, and nothing unexpected is allowed to happen might be considered the ultimate triumph of human organization.
A perfectly safe world.
And a remarkably lifeless one.