Executive Summary

This essay proposes a practical theory of power, particularly in international relations.¹ The basic points are:

• The form politics takes is primarily determined by the tension levels between the States involved — friends cooperate, rivals compete, and enemies fight.²
• These tension levels determine the means of power/threat States use and the logic of their interaction. E.g. friendly neighbours can’t use military threats against each other and stay friends, just like military enemies are past trade disputes and sanctions.
• The escalation of tension levels happens in predictable, concrete steps and is fairly well researched. I’ll illustrate and conceptualise this using Friedrich Glasl’s nine-step-model of conflict escalation.
• In uninstitutionalised interactions, power is a successful threat of sorts. Unsuccessful uses of power — unsuccessful threats — are the primary way conflicts escalate.
• Institutions and treaties are designed to withstand a certain maximum level of tension. If tensions escalate beyond their dispute settlement mechanisms, these structures falter or break down.

I believe the theory laid out here is superior to current theories in several respects:

• It is more accurate (and empirically falsifiable to begin with).
• It is useful to practitioners who need a strong framework of analysis.
• It leads researchers to ask better research questions.

What is politics?

As a starting point, it’s difficult to disagree with Andrew Heywood’s introductory definition that politics is “the activity through which people make, preserve and amend the general rules under which they live.”³ I’ll use “politics” and “make the general rules” as a shorthand to refer to this idea.

Heywood’s introduction goes on to differentiate four categories of definitions: Politics understood as (1) the art of government (think Bismarck, Machiavelli, Easton, and arguably Clausewitz) and politics as (2) public affairs (Aristotle, Mill, Rousseau, Arendt, the liberal tradition more generally); politics as (3) compromise and consensus (Habermas, Crick), and politics as (4) power and the distribution of resources (Leftwich, Lasswell, realism, often feminism and Marxism).

Politics without the State?

The first two categories differ mainly on whether or not politics happens when the State isn’t involved. To define politics as the art of government is an institutional definition. It’s similar to Clausewitz’ definition of warfare as ‘everything that is related to the armed forces; their creation, maintenance and use.’⁴ So, labour relations, gender, religious obligations, or beauty standards; are these issues ‘political’ even when they don’t involve the State? Can irregular guerrilla forces conduct warfare?

I don’t think this question should be predefined for all societies throughout history. Instead, let’s extrapolate from Heywood’s initial definition: Whether or not politics is that which pertains to the State depends on whether States are in fact the institutions that set the ‘general rules’ for a given issue at a given place and time. If it’s other institutions (e.g. the workplace, religious institutions, tribal clan structures, international organisations, et cetera), then the process by which those general rules are set is political, too, and our theory should yield insights for it.

Politics without power?

Heywood’s latter two groups disagree mostly about whether politics also happens when nobody imposes their will on somebody else. When everybody freely agrees on a reasonable general rule, is that still politics?

As we shall see, power is often a key ingredient in politics. And it is also true that using power is anathema to friendly relations. But to then say politics is defined as that which involves power and rivalry is effectively to assume that setting general rules cooperatively – as happens in households, parliaments, schools, workplaces around the globe on a daily basis – is humanly impossible, or too boring to think about. Setting those general rules without power struggles is evidently possible. As Sir Bernard Crick had it: ‘Why call … a struggle for power ‘politics’ when it is simply a struggle for power?’⁵

What is power?

In political science, power is often defined thus: A has power over B if A can make B do what A wants.⁶ Hans J. Morgenthau elaborated: “Power may comprise anything that establishes and maintains the power of man over man. Thus power covers all social relationships which serve that end, from physical violence to the most subtle psychological ties by which one mind controls another.”⁷

The weakness of this definition is that it is exclusively bilateral. Max Weber’s terminology evades this limitation and that’s why I use his definition of power (Macht) as: “Power is any chance to impose one’s own will within a social relationship, even against resistance, regardless of what this chance is based on.”⁸ This covers bilateral uses of power, but also the case of imposiing one’s own will vis-a-vis a collective (e.g. a ruler imposing their will on a population), or two collectives fighting for supremacy (e.g. two nations at war).

Power looking backward or forward in time

An interesting property of this definition is that one can only say whether A has power after A has successfully used that power. In terms of international relations: The definition is totally agnostic about what makes a State powerful. It defines the effect you’ll see when you look back in time at its interactions. The means by which A achieved this power are not part of the definition. In fact, even in retrospect, it’s not always clear what made the others give in. The means of power that achieved this aren’t predefined – if anything, they’re apparent ex ante.

A simple example to illustrate: My martial arts skills do not give me power over you in and of themselves, only once I successfully used them to threaten you into giving me your ice cream can we potentially say that they gave me power over you. (I don’t have any martial arts skills. Relax.) To know beforehand whether or not you’ll submit to the threat is somewhere between an art and a science. The reason is that it is you who chooses to submit or not. In the moment right after I issued my threat, it’s your reaction that gives me power, not mine. So it may be that you’re a judo master and confident that you could easily match my threat. But even then, objectively having superior means of power, maybe you still yield because the hassle just isn’t worth an ice cream. To predefine the means of power that make a State powerful in all situations is, in my estimation, futile and will lead to dogmatism.

And if you choose not to submit and violence breaks out? Well, as Morgenthau (and many, many others) knew: “When violence becomes an actuality, it signifies the abdication of political power in favour of military or pseudo-military power.”⁷ If you consider that I essentially try to control your behaviour using a threat of violence, you immediately realise that if I have to actually use my means of threat that signals that I don’t have power over you after all. Sure, I might ultimately win the fight and get your ice cream, but I haven’t made you do what I wanted.

Let’s apply this to a recent example in international relations: In summer of 2024, Israel and Hezbollah were in a two-decade-long stalemate. Had one looked at Hezbollah’s military capabilities, one might have concluded that Hezbollah can make an Israeli invasion of Southern Lebanon prohibitively costly. No definition of power that predefines the means of power would have helped foresee the possibility that Israel would rig explosives to pagers and walkie-talkies to kill and injure Hezbollah personell and civilians. Would anybody have counted pagers and walkie-talkies as a means of power? I think not. But in a power struggle, those involved constantly try to outsmart each other (in the logic of their already degraded interaction).

The basic point stands that the relevant means of power are revealed in retrospect to the use of power. Using power is always a bit of a gamble. The interesting thing a theory of power should do is to try and predict the means of threat/power that States will marshal before the fact but without dogmatic assumptions. And this theory tries to do just that.

What this theory is and isn’t

Repeated and institutionalised interactions diminish the uncertainty laid out above by potentially transforming power (Macht) into Weberian domination (Herrschaft). This transition occurs when power is not only routinised but also grounded in a belief by the dominated in the legitimacy of the dominant’s power (Legitimitätsglauben). Power is thus met with the cooperation of the dominated in the form of obedience (Gehorsam), which stabilises power and establishes domination.⁹ Jean-Jacques Rousseau expressed the same idea rather more beautifully:

And so when interactions are becoming institutionalised, there grows an ever thicker layer of confounding variables: Powerful institutions might arise by the sheer need for functional cooperation (e.g., trade regimes); control of these institutions might be obtained through means other than power, yet provide the occupant with considerable means of power/threat (e.g., bureaucracies, arguably democracy); history might lend formerly powerful nations advantages through path dependencies (e.g., colonial legacies).

Let’s then distinguish two ideal-typical poles of social interaction: uninstitutionalised and institutionalised.

• At one extreme, uninstitutionalised interactions are episodic, ad-hoc, and lack stabilising institutions.
• At the other extreme, institutionalised interactions are embedded in rules, roles, and routines, where power is mediated by legitimacy, repetition, and shared expectations.

In what follows we’ll focus focuses on uninstitutionalised interactions first to isolate the all too human roots of many power struggles in our psychology. By stripping away the effects of institutionalisation at first, we can examine how power functions in its raw and most uncertain form.

Predicting the means of power/threat

What we endeavour now is can be thought of as an update to Morgenthau’s view of human nature. It’s based on the insight from conflict research that the tension level between parties foreshadows before-the-fact which means of threat they will likely use if they choose to use threats against each other. I’ll argue that attempts to use power are usually a threat of conflict escalation to the next higher tension level.

That’s why it’s utterly unbecoming to a friendship to attempt to use power — you’re essentially threatening your friends with conflict escalation. Likewise, it’s irrelevant if you threaten an enemy with an angry letter — because the relationship is long past that point. The means I can reasonably be expected to use to attempt to control you depend on whether you are my friend, my rival, or my enemy. For any individual act, we can ask: Given the current relationship and its tension level, is this act escalatory, de-escalatory, or confirming the status quo?

This same way we also determine how extreme an act is. If the French president Emmanuel Macron had bought an ice-cream for Angela Merkel when she was still the Chancellor of Germany, that’d have been a friendly gesture confirming their friendship. Now with Olaf Scholz as Chancellor, it might be a bit more awkward given their colder relationship, but still basically just confirming their friendship. But consider if Vladimir Putin were to offer Wolodimir Zelenskyy an ice-cream “as a gesture of goodwill” — that’d be supremely awkward and Zelenskyy would quite rightly have it tested for toxins.

The logic of interaction is determined by the relationship and its tension level. Little details that are irrelevant among relaxed friends can become very important status symbols between tense enemies – like when Yasser Arafat and Ehud Barak couldn’t decide who enters the door first at Camp David.

In conflict research and social psychology, a great deal of research has been conducted into the nature of conflict escalation and de-escalation — a field I propose political analysts should tap into. How tensions escalate and friends become enemies is actually quite well understood, predictable and known in considerable detail. I shall now give a fairly detailed overview.

Basic logic of conflict escalation

This account uses a popular model of conflict escalation from the German conflict research literature to present the evidence – Friedrich Glasl’s Konfliktmanagement.¹⁰ He proposes a detailed nine step model of conflict degeneration (or escalation) in which the relationship of two or more parties slowly disintegrates from a mild hardening of positions to a total war of mutual obliteration.

On each step of the ladder, the parties intuitively know which kind of behaviour is adequate and what constitutes a taboo. But because conflicts are very frustrating situations for us humans to be in, there is a tendency in each party on each step to try to dissolve the distressing ambiguity of the situation. And, one way we humans ironically choose to do this is breaking the taboo-barrier to the next the next step! We often perceive this as a clarifying moment, when we finally say: “Enough!”

Using these distinct escalation steps, we can deduce what means of power can be expected to be marshalled on each step. Indeed, as we shall see, this perspective yields rich insights into the nature of power, postulates many empirical hypotheses that are eminently falsifiable,

There is no universal state of nature

For the present purposes it shall suffice to collate the nine steps of Glasl’s staircase into three spheres (following Glasl himself):

• On steps one to three, the parties’ perception is still issue-oriented and they share a win-win mentality, they are friends.
• On steps four to six their perception is person-oriented and they have a win-lose mentality, they are rivals.
• And on steps seven to nine their perception is destruction-oriented and they have a lose-lose mentality, they are enemies.

These three spheres roughly correspond to the way Alexander Wendt read Martin Wight’s ‘three cultures of anarchy’ in his Social Theory of International Relations– Kantian, Lockean and Hobbesian. ‘[A]t the core of each kind of anarchy is just one subject position: In Hobbesian cultures it is “enemy,” in Lockean “rival,” and Kantian “friend.”’¹² But Glasl’s work, as we shall see, is much better rooted in research and adds considerable depth and detail. Also, crucially, it explains the dynamics of change.

The brief sketch that follows paints a picture of an ideal-type conflict degeneration without any conflict resolving forces in order to illustrate it. Nothing here suggests that the ‘normal’ state of human affairs is conflict-free or peaceful, or that human societies teleologically gravitate towards cooperation.¹³ Indeed, I agree with Glasl: While conflict escalation is driven to a great degree by primitive, partly unconscious drives, conflict de-escalation is usually a conscious, laborious, and rational effort.¹⁴ There simply is no state of nature; we possess the capacity for cooperation, rivalry, or violence.

1-3 — Friends: ‘We have a problem’

Through the first three steps (the Kantian sphere of cooperation), the parties involved think in terms of issues. The sovereignty and physical integrity of the other participants is not only guaranteed, rather, it is so deeply internalised that the parties don’t even think about the possibility of violating it. For example, Switzerland ‘accidentally invaded’ neighbouring Liechtenstein in 2007 (170 infantry soldiers unknowingly crossed the unmarked border).¹⁵ Nobody doubted this was an accident and a simple apology sufficed to make amends. The same scenario is impossible on, e.g., the Lebanese-Israeli border.

Nevertheless, absent an apology or other conflict resolving forces, the conflict will begin to polarise the parties’ thinking, feeling or willing.¹⁶

1. An initial hardening of positions and behavioural patterns (step 1) gives way to debates and polemics¹⁷ (step 2) when a party deliberately and consistently uses (polemic or behavioural) tactics to win the argument rather than to discuss the issue. That’s the first taboo-barrier from step one to two (henceforth abbreviated like this: ‘taboo-barrier 1>2’).
2. Then, when the parties start using unilateral actions to confront the other party with a fait accompli (taboo-barrier 2>3), the conflict escalates to the third step, where the parties believe that ‘actions speak louder than words!’¹⁸ (step 3).
3. The last taboo-barrier to be broken upon becoming rivals and entering the Lockean sphere is the profound shift from an issue- to a person-orientated perception, when one party deliberately and consistently attacks the other ad hominem (taboo-barrier 3>4). The parties’ conviction that they are friends with an issue to solve is obliterated. ‘Thomas Gordon called this: From act (a disturbing action) to actor (a disturbing person)!’¹⁹

4-6 — Rivals: ‘You are a problem’

The increasing importance of self- and other-image already anticipates the primacy of personal identity throughout Lockean power politics. 4) Initially (step 4), the parties worry about their reputation and honour.²⁰ The next taboo-barrier is broken when a party wilfully provokes the public loss of face of the rival (4>5). 5) In their struggle for their reputation (their ‘face’), the parties seek to convert bystanders of their other-images of the rival (step 5).²¹ The parties might now start to use punitive violence in a discernibly restricted manner – often in ways that can later be repudiated. Their frustration leads the parties to paradoxically seek to assert their independence by using (step 6) strategies involving threat and coercion (‘taboo-barrier 5>6’)²²: We’ve seen that unilateral actions have been used from step three onwards and grown more and more extreme, up to the punitive violence from step five. What now occurs is the following: The previous unilateral acts were past-oriented (‘You did X, therefore I now do Y!’), but now the parties being to use clear, irrevocable threats which in contrast are future-oriented (‘If you do X, I will do Y!’). However, the (quite logical) effect of such threats is the exact opposite of independence, because the parties essentially surrender their agency to their rival. Also, ‘[t]hese threat-strategies work only so long as it is clear that they are intended to avert a greater escalation of violence.’²³ 6) But the parties sooner or later begin to threaten with the violation of the other’s physical sovereignty which fundamentally shakes the parties’ sense of physical security. Enacting such a threat and violating the other party’s physical sovereignty (‘taboo-barrier 6>7’) makes Lockean rivals become Hobbesian enemies and the conflict escalates into the Hobbesian sphere of war.

Conflict escalation through the Lockean sphere defines the alliances which will fight against each other in the subsequent Hobbesian struggle by totalising the self- and other image across all aspects of life and making their division absolute. Interestingly, ‘to a striking degree, the misperceptions of those in conflict are mutual’²⁴ – an effect which is called negative mirror-image perceptions.²⁵ This makes sense in the following way: Both parties need to justify to themselves the same frustrating state of affairs and – according to Glasl and the theory outlined here – they both create mental models with clear self- and other-images that absolve them of their part of responsibility in creating the status quo. The primacy of these identities creates considerable in-group cohesion²⁶ and lets members’ individual identities become more and more irrelevant.²⁷ They also increase the profound suspicion (step 4) to an autistic enmity (step 5 and 6) and finally profound hatred (step 7). Under these psychological pressures, the parties now seek to find loopholes in established norms that allow them to hurt the other party without openly violating the norm.²⁸

7-9 — Enemies: ‘You threaten my existence’

The importance of physical security and threats towards the end of the Lockean rivalry already anticipates the primacy of the means of violence in the Hobbesian sphere. The enemy’s loss is now considered a direct personal gain in and of itself (nicely captured in the German Schadenfreude). For example, ‘[in the nuclear-war planning of the cold-war era] the calculation went, if each side was likely to suffer 50% or even 75% casualties among its entire population, the war would be ‘won’ in ensuing years by the side that could more quickly recover, rebuild (and repopulate).’²⁹ In other words, the logic of the situation is no longer zero-sum but negative-sum.

7. The parties now typically attempt to permanently destroy the capability with which the other party threatened them on previous steps. They attempt a restricted final blow,³⁰ aimed at disabling the threat the other party poses without destroying the other party completely. But even if this is the initial aim of destructive violence, it is not always possible to disable the threat decisively enough.
8. Thus, eventually, the application of violence broadens in the next step when the parties begin to target the enemy’s central nervous system.³¹ In war this would be the moment a party begins to target the civilian infrastructure. They shoot to kill the other party as such.
9. If neither party destroys the other in this total war, the one last barrier that can be crossed is the kamikaze mentality: ‘[The parties] lash about senselessly because they experience slithering towards an abyss from which there is no escape. Their only solace is to drag their enemy down along with them so that the enemy has to perish too.’³²

The shift from Locke to Hobbes coincides with a shift from rivalry to war – again with Morgenthau: ‘When violence becomes an actuality, it signifies the abdication of political power in favour of military or pseudo-military power.’³³ Delivering on the promises made at the end of the Lockean sphere in the form of threats (step 6), in the Hobbesian sphere both parties aim, from the beginning, at the lasting destruction of the enemy’s capability that is threatening them. As one security analyst put it in the context of the Iran-Israel crisis:

But the parties do not leap into an all-out war right away, and powerful tacit agreements underpin the Hobbesian stage too (step 7).³⁵ For example, Kane and Lonsdale argue that the common Clausewitzian assertion that ‘war is the continuation of political intercourse with the admixture of other means’³⁶ rests on the consequentialist ethics that ‘military power is only valid if it serves the interests of the community…’³⁷ – which is maybe a little too optimistic a reading of Clausewitz, but serves to make the point.

Morgenthau too argues against Waltz’ claim that States in anarchy cannot afford to be moral – Kane and Lonsdale summarise his position thus:

The particular insight of Glasl is that the shift from step seven to eight obliterates these basic moral principles. The very category ‘civilian’ is obliterated and population centres become fair game.³⁹

Summary

So, if one looks back at Glasl’s conflict degeneration ladder, one can summarise that

Part of this degeneration is the corrosion of established norms.

I’d add that this corrosion of established norms is often accompanied by the emergence of a primitive tribal morality.

Observe that of these three spheres, only fully-degenerated Hobbesian enemies lash out at each other according to the logic of total struggle. As such ‘[they] cannot count on each other for help or even to observe basic self-restraint’⁴² – ultimately, as we’ve seen, not even on the self-restraint induced by self-preservation. Wendt is quite right that such ‘lack of inhibition and self-restraint in Hobbesian cultures suggests that balances of power there will be difficult to sustain’⁴³ because States would attack at every single opportunity.

Offensive realism is thus actually not quite as Hobbesian as it might portray itself to be. ‘Sovereignty is the ground rule of inter-State relations in that it identifies the territorial entities [which] are eligible to participate in the game.’⁴⁴ If States acted towards one another the way Hobbes claimed individuals act in the state of nature, or the way offensive realists claim States behave, they would never respect, let alone establish, the modern ground rule of international politics: territorial sovereignty.

Indeed, it is important to stress that the behavioural logic at such high tension levels is often unrestrained even by elementary tactical thinking. If you agree with Clausewitz that defence is a stronger form of war than attack,⁴⁵ well, that alone would undermine offensive realism. But it doesn’t: Because the insanity of offensive realism actually exists at high tension levels. There, parties are quite happy to die to win and are generally unreasonable.

A new understanding of uninstitutionalised power

I said before that uninstitutionalised power is a successful threat using means from higher tension levels. Now we have pretty solid idea of what that means and can attempt a before-the-fact approximation of the means of power/threat parties will use to try to control somebody – the after-the-fact definition of power we usually use in political science.

Say tensions are already on step four, where the parties worry primarily about their reputation and honour. If somebody wanted to use power against somebody else in this context, the additional means of power/threat they’d now most likely employ would be threatening to release some dirty secrets about their rival to cause her or his loss of face. This while continuing to apply all previous means of power/threat already used.

This view of power alone, I think, already provides us with a straightforward research agenda for political science in general and IR in particular. We should study and catalogue the specific means of power/threat parties (for IR mostly States) tend to use at which tension levels. If catalogued, this would additionally provide us with data for network analyses: If actors A and B have an escalating conflict, when and how do they involve actor C? (The guess is from stage five onwards.) How do the tension levels between A and C, or B and C, influence this? Whereas Kenneth Waltz’ crude notion of material capability didn’t lead to knowledge accumulation (and in my opinion positively obscured it), this could provide a more realistic backbone for research.

There is little reason to constrain such research to political science only: We’ve seen the main insights in this part came not from our own field at all, but from conflict research and psychology. Indeed, while the groundwork for this theory might already touch on many aspects of political science, it itself still consists mostly of claims about human psychology. So far, it really is just an update of Morgenthau’s view of human nature.

I invoke ‘human nature’ not to unduly and dogmatically root a competitive urge in humans and thus explain politics (a very common misinterpretation of Morgenthau). There exists no universal proactive competitive or violent urge in humans. There exists a universal competitive and violent potential in humans which can be provoked – along the steps outlined above.