The Machinery of Legitimation II

Institution by Institution, What Actually Breaks

Ask yourself this question. What is the legitimate basis of political and economic authority, and what happens when that authority is systematically used against the people it claims to represent?

No Villains Required

This is not a checklist for spotting bad men. Our previous essay on [how states form and legitimate their authority] rejected great-man history — a founding lawgiver bringing order out of chaos — and the same rejection applies here, doubled. What follows is not a story about a tyrant’s personality, an evil ideology, or a population that simply chose badly. Blaming an individual explains nothing about the dozens of times this pattern has repeated across completely different personalities, eras and continents. Blaming a system or an ideology alone — fascism, communism, whatever label fits the reader’s own priors — explains nothing about why the same mechanism shows up inside systems that call themselves by opposite names. And “the people voted for it,” history’s most comfortable line, mistakes a symptom for a cause: a population’s vote is itself downstream of the pressures we are about to trace, not an independent moral failing sitting outside them.

What actually explains it is the same three forces that have shaped every transition already traced: pressure, structure, and opportunity. A state under genuine or invented threat has a pressure it can use. An institution with a specific structural weak point — a legislature that can be bypassed, a judiciary that can be reshaped, a press that can be bought rather than banned — has an opportunity waiting inside its own design. And the change happens through mechanism: identifiable, repeatable steps, not a single evil decision. Nor does it happen without cover. Every shift traced so far — kinship giving way to a leader, a feudal lord’s authority hardening into a fiscal-military state, empire’s extraction machinery re-legitimising itself as capital — needed a story that made the change acceptable to the people living through it. This one is no different. What follows is that story’s mechanics, one institution at a time.

Winning Without Cheating — How Elections Get Tilted, Not Stolen

A functioning opposition is not democratic decoration. It is closer to plumbing — a pressure-release valve built into the state’s own design, there to let “the government is getting this wrong” go somewhere real: to a rival with an actual chance of taking over, rather than nowhere at all. Squeeze that valve and the elections built to include it become hollow rather than real. V-Dem’s own methodology — the same framework used in Part One to separate electoral democracies from liberal ones — treats a broken opposition as disqualifying in itself, not as a complaint layered on top of an otherwise-working system.

An election can be lost before a single vote is cast. Political scientists Steven Levitsky and Lucan Way call the resulting system “competitive authoritarianism” — regimes that hold real elections, with real opposition parties and real ballots, but where the incumbent has spent years making sure the contest cannot go against them. No stuffed ballot boxes are required. The rules of the game are simply written by one of the players.

The first tool is the map itself. Gerrymandering — drawing district boundaries to lock in a result rather than reflect a population — takes its name from Elbridge Gerry, the Massachusetts governor whose 1812 state senate map produced a district so contorted a newspaper cartoonist drew it as a salamander. The technique is two centuries old and still current: Hungary’s Fidesz party redrew parliamentary boundaries after taking power in 2010, converting a popular-vote plurality into a supermajority large enough to rewrite the constitution itself. The vote is free. The map decides the outcome before the vote happens.

The second tool is the voter roll. Deciding who is allowed to vote is as powerful as deciding how the votes are counted. The post-Reconstruction American South built an entire architecture for this — literacy tests, poll taxes, grandfather clauses — engineered to disenfranchise Black voters while remaining, on paper, racially neutral. Modern equivalents rarely announce themselves so directly: strict voter-ID requirements calibrated against which citizens are less likely to hold the required documents, or purges of the electoral roll that fall disproportionately on renters, students, and the poor. The mechanism is the same one used in 1890s Mississippi. Only the paperwork has changed.

The third tool is fear. An opposition can be legally permitted to run and still be too frightened to campaign. The same post-Civil War South that built the legal disenfranchisement architecture also relied on lynching and organized terror to make sure Black citizens who cleared the legal hurdles never reached the ballot box regardless. Belarus offered a live version in 2020: opposition candidates jailed or forced into exile before election day, and a security apparatus turned loose on the protesters who contested the result afterward. Intimidation does the work a rigged ballot would otherwise have to do.

The fourth tool is the charge sheet. A state moving the wrong way down that table rarely bans an opposition outright — that is a loud, legible act, and legible acts draw exactly the attention a state trying to shift quietly wants to avoid. The quieter version prosecutes an opposition figure on a charge that happens, always, to be about something else: tax irregularities, “extremism,” a decades-old case suddenly revived. The opposition still legally exists. It just can’t field the person who was winning.

The fifth tool is the count. If the map, the roll, the fear, and the charge sheet all fail to produce the right result, the institution that tallies the votes can simply report a different one. Venezuela’s National Electoral Council did exactly this in 2024, declaring Nicolás Maduro the winner while withholding the polling-station-level data that Venezuelan law requires it to publish — data that independent tallies, compiled from those same polling stations, showed pointing the other way.

None of these five tools requires a single fraudulent ballot. That is the point of naming them together. A country can hold an election with a real count, a real opposition, and a real ballot, and still fail the test of a genuinely free contest completely.

Flooding the Zone — The Press When Everyone Can Publish and Nobody Owns the Whole Pipe

A free press is the other half of the same plumbing. Where a functioning opposition gives “the government is getting this wrong” somewhere to go politically, a free press gives it somewhere to go publicly: a scandal becomes a story, argued over in daylight, before it can curdle into a grievance nobody in power ever has to answer for. Squeeze that valve the traditional way and the mechanism is familiar from a century of press history — ownership consolidating into fewer, friendlier hands; libel and “national security” law stretched until investigative journalism becomes personally costly rather than formally illegal; press credentials and official access granted selectively, so that criticism still technically exists, it simply can’t reach anyone who matters. None of that required banning a single newspaper.

That mechanism now runs into a genuinely new problem: it was built for a world with one pipe. For most of the twentieth century, controlling the press meant controlling a small number of owners, printing presses, and broadcast licenses — a chokepoint a state or an oligarch could squeeze directly. Digital publishing broke that chokepoint from the content side. Anyone with a phone can publish; a government can no longer simply revoke a printing license and expect the story to vanish. But the infrastructure carrying that content — the platforms, the algorithms, the ad networks deciding what gets seen — has concentrated into fewer hands than the old newspaper industry ever managed. Content became free. The pipe did not.

Capital’s newest move is to stop trying to buy the pipe from outside and simply become it. Elon Musk’s 2022 acquisition of Twitter, relaunched as X, is the clearest case: not a media mogul buying influence over a platform he doesn’t control, but capital owning the infrastructure directly, able to amplify, throttle, or reinstate any given voice by decision rather than by pressure. The old model bought newspapers. The new model buys the pipe the newspapers now depend on.

Where the old censorship model removed information, the new one drowns it. Steve Bannon described the strategy to journalist Michael Lewis in 2018 in four blunt words: flood the zone with shit. The goal isn’t to stop a true story from being told — on a platform anyone can publish to, that’s no longer achievable. The goal is to bury it under such a volume of competing claims, conspiracy, and noise that no single story can hold public attention long enough to matter. Suppression becomes unnecessary once distraction works just as well.

Hungary supplied the live test case, and the result cuts against the mechanism’s own confidence. Viktor Orbán’s government spent a decade building KESMA, a consolidated pro-government media conglomerate absorbing hundreds of previously independent outlets under one ownership structure — the old chokepoint strategy, updated. It still lost the 2026 election, with the defeat credited in part to opposition organising that ran through the digital-native channels KESMA was never built to control. KESMA itself then collapsed financially, a state-built media empire that had captured the old pipe just as the water started running through a different one.

What the same broken chokepoint produces, politically, depends entirely on who reaches through it first. One harvest looks like platform ownership fused directly with political intent — the Dark Enlightenment strand of thought that treats concentrated platform control as a feature rather than a risk, and figures like Nigel Farage building an “authentic,” algorithm-native political style that older media gatekeeping would never have permitted. The other harvest arrived later and from the opposite direction: a younger generation’s digital-native organising, built on exactly the same open-publishing layer, arriving too late to stop Hungary’s KESMA project but not too late to help defeat it.

Underneath both harvests sits a contradiction capital has not resolved, and this essay does not resolve for it. X’s own commercial position has deteriorated sharply since Musk’s acquisition — advertising revenue falling from roughly $4.4 billion to around $2.5 billion — even as its owner’s political reach arguably grew. A platform can be losing money as a business while gaining power as an instrument. Whether that combination is stable, or whether one side of it eventually forces the other, is a live question rather than a settled one.

The Concept Outlives Its Capture — Law, Courts, and What Holds When Power Is Tested

Law is not one thing, and confusing its two halves is where most arguments about it go wrong. There is law as a concept — the idea that power should be constrained by rules applied consistently, regardless of who holds it — and there is law as captured content, the specific statutes and rulings a given legislature or court happens to produce at a given moment. The concept can survive terrible content. What it cannot survive is content produced by a process no longer bound by the concept at all.

E. P. Thompson made the sharpest version of this case from the political left. In Whigs and Hunters, writing about eighteenth-century English law used overwhelmingly to protect landed property against the poor, Thompson still concluded that the rule of law was “an unqualified human good” — not because the law of his period was fair, but because a state bound to its own rules, however imperfect those rules were, is a fundamentally different and less dangerous thing than a state that answers to nothing. The content can be captured by class interest. The concept of being bound at all remains worth defending regardless.

Legal philosopher Lon Fuller gave that concept a working checklist rather than leaving it abstract. Laws should be public, prospective rather than retroactive, reasonably clear, non-contradictory, possible to comply with, stable enough to plan around, and — the item that matters most here — actually enforced as written rather than applied selectively depending on who stands in front of the court. A legal system can satisfy every item on Fuller’s list in its text and fail the entire test in its practice, the moment enforcement starts depending on who the defendant is.

This is exactly the line V-Dem’s own framework draws between an electoral democracy and a liberal one. An electoral democracy can hold clean votes while its judiciary enforces the law unevenly. A liberal democracy requires the courts to hold regardless of who is in office — which is precisely why judicial independence functions as its own separate test, not a formality that free elections automatically produce.

Weimar Germany shows what happens when the exception built into the law becomes the rule. Article 48 of the Weimar constitution allowed the president to rule by emergency decree, bypassing the legislature, during a genuine crisis — a reasonable provision for an actual emergency. It was invoked so routinely in the Republic’s final years that emergency rule became the normal mode of governing rather than the exception to it, and by the time it was used to install a dictatorship the mechanism had already been worn smooth by years of ordinary use.

Capture itself comes in degrees, and the mildest form is worth naming precisely because it fails. Franklin Roosevelt’s 1937 attempt to expand the US Supreme Court from nine justices to as many as fifteen — a direct response to a hostile Court striking down New Deal legislation — met refusal from Congress, including members of Roosevelt’s own party. The attempt is the point: a capture mechanism was tried against a functioning system, in daylight, and the system’s other branches held.

Poland’s judicial overhaul in the 2010s shows the middle depth, where the mechanism works but leaves a paper trail. The Law and Justice government restructured the National Council of the Judiciary and lowered the retirement age for Supreme Court judges, moves that pushed out sitting judges and installed replacements more sympathetic to the ruling party — all done through ordinary legislation, none of it requiring a single law to be broken.

Turkey’s post-2016 purges show the deepest form, where the mechanism no longer needs cover at all. Following the failed coup attempt that year, tens of thousands of judges, prosecutors, and civil servants were removed from their positions, officially for alleged links to the coup, in a purge whose scale and speed went far beyond anything a genuine security investigation could have individually verified in the time available.

None of this means judicial independence is a fortress that either stands whole or falls whole — it never was one to begin with. Judges are appointed through political processes everywhere, including in states with genuinely independent courts; the difference is not the absence of political influence but whether that influence stops at the appointment or continues to reach into the ruling. And politics is not the only thing capable of capturing a court — a legal system priced out of reach of everyone but the wealthy protects the concept in name while failing it in practice just as thoroughly as a packed bench does. Protecting the concept was never about building a safe space above the fight. It means treating independence and equal access as the front line of the fight itself.

Whose Side Is Watching You On — Security Services, and the Population They’re Actually Protecting

Every state has an internal security apparatus, the question is which direction it faces? Does it protect the population under the law from genuine threats — crime, terrorism, foreign espionage — or does it protect whoever currently holds power from the population itself? Both versions can look identical on an organizational chart: officers, files, informants, budgets. The direction only shows up in who ends up in the files.

East Germany’s Stasi described its own function with unusual honesty, if the wrong kind. Officially the Ministry for State Security called itself the “shield and sword of the Party” — not of the state, not of the population, but of the ruling party specifically. By 1989 it held files on roughly six million East Germans, more than a third of the country’s entire population, maintained by around 100,000 staff and a network of citizen informants estimated at anywhere from 500,000 to 2 million. The apparatus was enormous, effective, and entirely unambiguous about which direction it faced.

China’s domestic security spending offers a more recent, less openly named version of the same choice. Weiwen — “stability maintenance” — spending overtook the country’s formal defense budget around 2010 and reached roughly $210 billion by 2020, with per-capita spending in Xinjiang and Tibet running far above the national average. The framing is stability, not repression, but the geographic distribution of the money tells its own story about which populations are being watched hardest.

The UK’s public-order spending offers a genuine contrast, not because Britain is uniquely virtuous but because the money is visible. At roughly 1.8% of GDP, UK public-order and safety spending is itemized, subject to Freedom of Information requests, and debated in Parliament as a line item like any other. That does not make British policing beyond criticism. It means the spending itself is an accountable fact rather than an opaque one — the difference this section is actually tracking.

McCarthyism shows the mechanism can work without ever appearing on a budget line at all. The early-1950s American anti-communist purges ran almost entirely through informal pressure operating inside formal institutions — congressional hearings, blacklists, loyalty oaths, reputational destruction — rather than through any dedicated new security agency built for the purpose. No new ministry was required. Existing institutions simply began asking a different question of the people who passed through them.

Sometimes neither the population nor a scheming government is driving the expansion — the apparatus expands itself, and nobody outside it is asked. The UK’s undercover policing scandal, still working through a public inquiry as of 2026, showed officers from a unit originally created to monitor genuine public-order threats spending years inside political and environmental campaign groups that posed none, several fathering children under false identities before disappearing from those families’ lives entirely. No minister ordered this. No law was passed expanding police remit to cover grieving families’ campaign groups or climate activists. The unit simply kept operating past any mandate that could plausibly justify it, for decades, because nobody with the authority to stop it was watching closely enough to notice.

Consistency is the test here too, the same one any legal system worth the name has to pass. A security apparatus enforced evenly against genuine threats, regardless of who those threats implicate politically — and kept inside whatever mandate authorized it in the first place — is doing something different in kind from one that discovers threats disproportionately among a government’s critics, or quietly operates beyond any mandate at all. The organizational chart cannot tell you which one you’re looking at. Only the distribution of who ends up in the file, and who authorized the apparatus to be there, can.

The Institution That Decides Last — Armies, Loyalty, and Manufactured Emergencies

“The military stays above politics” is itself a political claim, not a description. When an army refuses an order from its own government, that refusal is a political act — a decision about which authority the institution actually answers to. When it complies, that is equally a political act, just one that goes unnamed because it produces no visible rupture. There is no neutral third option where the army simply isn’t involved. Political scientist Samuel Huntington’s influential theory of “objective control” — professional, apolitical militaries that simply obey civilian authority by design — describes an aspiration more consistently than it describes what militaries actually do when the test arrives.

Chile in 1973 is the baseline case for compliance with the wrong side. The military under Augusto Pinochet overthrew Salvador Allende’s elected government in a coup that killed the president and installed a dictatorship lasting seventeen years — professional, well-trained, entirely willing to act against the constitutional order it was nominally built to serve, the moment its own institutional interests aligned against it.

Turkey shows the same institution captured, purged, and then captured again by whoever did the purging. For much of the twentieth century Turkey’s military saw itself as the guardian of Kemal Atatürk’s secular state, staging coups in 1960, 1971, and 1980 whenever it judged civilian government had drifted too far from that founding settlement. President Erdoğan then broke that guardianship in a different direction: an alliance with the Gülen movement drove the Ergenekon and Sledgehammer trials through the 2000s and into the early 2010s, purging the Kemalist officer corps on conspiracy charges of variable credibility. The Gülenist network that had helped engineer that purge then attempted its own coup against Erdoğan in 2016 — and failed. This is not a fixed story of secular army versus Islamist government. It is the same institution captured twice, by two different factions, in two different directions.

Where the army held instead, the pattern runs the other way. Portugal’s Carnation Revolution in 1974 saw military officers themselves overthrow a decades-old dictatorship, disillusioned by an unwinnable colonial war their political leadership refused to end. East Germany’s and Romania’s security and military apparatus in 1989 either declined to fire on protesters or, in Romania’s case, turned against the regime outright once the order came. The Philippines in 1986 saw military defection tip a popular uprising against Ferdinand Marcos into a peaceful transfer of power. Egypt in 2011 initially followed the same pattern — the army declining to save Hosni Mubarak — only for the same institution to reverse itself in 2013, backing the coup that removed his elected successor. The pattern predicts resistance. It does not guarantee it holds.

One structural factor predicts which way an army tips more reliably than any stated ideology. An army drawn from the general population through conscription tends to break from a government more readily than a small, professional, deliberately separated force — the same bond that once holds together a group of kin, working in reverse. A conscript army is still substantially made of the population it might be ordered to fire on. A professional army, recruited, trained, and housed apart from that population for a career, has had exactly that connection engineered out of it.

Composition is not the only thing separation changes — training embeds a worldview, not just a demographic gap. An army does not simply follow orders; it is built, through doctrine, institutional memory, and the specific ethic of service instilled in training, to hold a view of what it is for — defending the nation, the constitution, the people, the regime, depending on how that training has been framed. That worldview is genuinely held, not cynically performed, and it is exactly what makes an army’s decision in a crisis so hard to predict from the outside: officers do not usually experience a coup, or a refusal to fire, as betraying their institution. They experience it as fulfilling the purpose the institution itself taught them to believe in. Chile’s officer corps in 1973 believed it was saving the nation from communism. Portugal’s in 1974 believed it was saving the nation from an unwinnable war and a discredited regime. Both were acting on a genuinely acculturated sense of duty. Only one produced a dictatorship.

None of this makes an army a neutral instrument waiting for instructions. It is an institution with its own interests, its own factions, and its own capacity to decide — correctly or not — what the constitutional order actually requires of it. Built to serve that order rather than whoever currently commands it is the standard the institution is measured against, and its own acculturated sense of duty is exactly what decides whether it meets that standard or believes, wrongly, that it already has.

An army also needs a reason to act, and reasons can be manufactured as easily as they can be genuine. The Reichstag fire in February 1933 — an actual act of arson whose true origin remains debated to this day — gave the Nazi government the pretext for an emergency decree suspending civil liberties. That decree was never given a real expiry. It simply became the permanent legal basis for the rest of the regime. Germany’s invasion of Poland in 1939 opened with the Gleiwitz incident: German operatives staged a fake Polish attack on a German radio station, using prisoners dressed in Polish uniforms and killed for the purpose, to manufacture the outward-facing justification for war the same way the Reichstag fire had manufactured the inward-facing one.

Neither fire required a uniquely evil individual to set it. No lone tyrant’s personality explains why this pattern repeats — they required an institution already primed by pressure, structure, and opportunity to treat a single event, real or staged, as grounds for a permanent expansion of power. The emergency is rarely the cause. It is the occasion.

No Playbook — How Capital and the State Negotiate the Turn

Capital is not a fixed ally of the state, nor a fixed enemy of it — it is a negotiating partner whose position shifts with the terms on offer. When a state moves toward the mechanisms already covered in this part, capital does not respond with one behaviour. It responds with whichever of four available moves currently serves its interests best.

The first move is exit. Capital that can leave, leaves. Argentina’s recurring capital controls — the cepo cambiario restricting currency conversion and outflow — exist precisely because governments have learned that capital will flee at the first sign of instability if nothing stops it. Venezuela’s collapse accelerated the same way: capital and capital-holders departed faster than any policy could replace them, hollowing out the economy the state was trying to control.

The second move is active midwifery — capital doesn’t wait for a regime change, it helps produce one. ITT’s documented involvement in the run-up to Chile’s 1973 coup, and Italian industrialists funding the early Fascist squads in the 1920s before Mussolini ever held office, show capital treating a friendlier political order as an investment worth making directly, not merely benefiting from afterward.

The third move is accommodation for profit — capital doesn’t fight the new order, it works inside it. Krupp, IG Farben, and Thyssen didn’t need to be coerced into supplying Nazi Germany’s war economy; each found the arrangement profitable and organized itself accordingly. Accommodation requires no ideological sympathy. It requires only that the numbers work.

The fourth move isn’t capital’s choice at all — it’s the state turning on the capital that built it. Nuremberg’s industrialist trials came after the regime industrial capital had accommodated collapsed entirely. Khodorkovsky’s prosecution and the dismantling of Yukos showed Putin’s Russia reversing the oligarch bargain of the 1990s once it no longer served the state. Jack Ma’s disappearance from public view and the regulatory dismantling of Ant Group’s IPO showed the same reversal inside a very different system. Musk’s relationship with the American state he now sits partly inside remains a live, undecided version of the same question — whether the fourth move is coming, or whether this time capital has captured the state instead of the reverse.

The state has its own four dimensions it can pull to shape which move capital makes:

Ownership concentration. A handful of oligarchs is a far easier negotiation to squeeze than a dispersed shareholder base.
Market size and leverage. Exit is a real option against a small economy and a much harder one against a market too large for capital to afford to abandon.
Deliberate autarky. A state can remove the exit option altogether by design — Nazi Germany’s 1936 Four Year Plan explicitly aimed at self-sufficiency precisely so that international capital’s threat of withdrawal would carry no weight.
Capital type. Industrial capital is comparatively immobile, financial capital highly mobile, and a newer, more dangerous category has emerged in tech capital that doesn’t just operate inside state infrastructure but becomes it: SenseTime, Hikvision, and Dahua, all sanctioned by Western governments, built the surveillance architecture Xinjiang’s security apparatus runs on directly, collapsing the line between private supplier and state instrument entirely.

Underneath all of this sits the dimension least argued over and most decisive: how the state itself is financed. Political scientist Terry Lynn Karl’s rentier-state theory — developed from her own work on Venezuela’s oil economy and generalized from there — holds that a government funded by resource rents needs its own population’s consent far less than one that has to tax a working economy broadly, because rent flows regardless of whether citizens are satisfied. Russia’s oil and gas revenue insulates the state from its own population in exactly this way. India’s sheer scale substitutes for rent without eliminating the need to tax broadly, leaving it a genuinely different case. Hungary runs on conditional EU rent, which gives Brussels real donor leverage the fully rentier states never have to answer to. Turkey has no comparable cushion at all, which is a large part of why it cycles through recurring currency crises rather than settling into stability. And Africa’s gatekeeper states run on a rent of a different kind again: geopolitical competition rent, extracted from great powers competing for influence rather than from the state’s own resource base or its own citizens.

None of this adds up to a single formula, and that absence is the argument’s actual point. Capital’s four moves and the state’s four dimensions combine differently in every case, because the specific mix of ownership structure, market leverage, capital type, and financing base is different every time. There is no universal script that tells a state how capital will respond to pressure, or tells capital how a state will respond to accommodation. There is only the negotiation, worked out case by case, with the outcome never guaranteed in advance.

Endurance Is the Norm — What It Actually Takes to Bring a State Down

The natural state of an authoritarian government is not collapse. It is endurance. The Soviet Union lasted 74 years. Franco’s Spain lasted 36. North Korea has now outlasted both, at over 75 years and counting. Every mechanism this part has traced — tilted elections, flooded information, captured courts, watchful security services, a loyal army, a negotiated peace with capital — exists to produce exactly this outcome: not stability as an accident, but stability as the achieved result of continuous, deliberate work.

But endurance was never a one-sided fact, and treating it as one misses what’s actually happening underneath it. For every year a regime endures, the resistance to it endures too — not as a single uprising waiting for its moment, but as a continuous condition: smuggled information, underground organizing, quiet non-cooperation, individuals who never stop wanting out from under it even when getting out looks impossible. The Soviet Union’s 74 years were not 74 years of settled consent. They were 74 years of a state working continuously to manufacture consent it never fully secured, against a population that never fully stopped withholding it. Orwell’s Winston Smith, ground down until he loves Big Brother, is the literature’s most famous image of the state winning that contest completely — which is exactly why he stands out. Most who live inside an enduring state do not end there. Some genuinely comply. Many simply wait.

Syria’s Assad regime showed how fast that achieved stability can still unravel once its supports genuinely fail together. The regime had lasted 53 years, through father and son, when it collapsed in eleven days in December 2024. The speed only looks inexplicable if the mechanisms are considered one at a time and the resistance is assumed to have started only when the collapse did. Taken together, the crack becomes legible: an army no longer willing to fight, for a state whose financing had been strained past its limit by over a decade of civil war and international sanctions — loyalty and money failing at once, against a population that had never stopped resisting even through the years the regime looked most secure.

Political scientists Erica Chenoweth and Maria Stephan’s research on civil resistance gives the clearest empirical account of what that continuous resistance actually achieves, and how often. Their study of resistance movements found nonviolent campaigns succeeded roughly 53% of the time against violent campaigns’ 26% — and identified a rough threshold, since nicknamed the “3.5% rule,” at which no campaign that achieved active participation from that share of a population’s people had failed to produce change. Estonia’s Singing Revolution, Georgia’s Rose Revolution, and Czechoslovakia’s Velvet Revolution all sit inside this pattern: mass, sustained, largely nonviolent participation succeeding where narrower or more violent resistance had not — participation built up over years of the quieter resistance described above, not assembled overnight.

The pattern is real, and it is not a guarantee — both facts matter equally. Bahrain’s 2011 uprising and Brunei’s pro-democracy movement both reached participation levels near or above the 3.5% threshold and still failed to produce lasting change. More troublingly, Chenoweth’s own more recent data shows civil resistance’s overall success rate declining sharply: from roughly 65% in the 1990s to below 34% since 2010. The most credible explanation is that authoritarian states have been learning. Security services have studied earlier collapses and adapted specifically to prevent the one mechanism that ends them fastest: mass defection by the people meant to enforce the government’s will.

None of this offers comfort, and it was never going to — but it isn’t a story about the state’s endurance alone either. Endurance is the norm because every institution this part has examined is built, in normal operation, to produce endurance. Resistance enduring alongside it, mostly unseen and mostly unrewarded, is just as normal, and just as continuous. Collapse is not the system malfunctioning. It is the rare moment when enough of the state’s machinery fails at once for the resistance that was there the whole time to finally have somewhere to go.

The Oldest Border Problem, at a New Scale — Capital, Labour, and Who Still Counts as In

Authority was never founded. It accreted — kinship, bureaucracy, monopoly of force, manufactured consent, built and rebuilt across millennia. And when that authority turns against the people it claims to represent, the result is not a villain’s decision but a mechanism: pressure meeting structure meeting opportunity, institution by institution, with no single fix available. Neither answer resolves into comfort. Both were only ever meant to resolve into clarity.

Capital’s own contradiction is really one contradiction wearing several disguises. It exits when a state threatens it, and it midwifes new states into being when the opportunity is worth the risk. It accommodates regimes it has no ideological sympathy for, and it is occasionally devoured by the very state it helped build. There is no playbook governing which move it makes when — but at the scale of the empires that actually matter now, not nineteenth-century territorial ones, capital is simultaneously opposed to state power and fused with it: opposed through arbitrage and the permanent threat of flight, fused through direct infrastructural capture, a platform owner who is also the platform, a surveillance contractor whose technology is the state’s surveillance apparatus. Capital is not on any side. It is the mechanism by which every side gets tested.

The oldest border problem in human political life is being run again now, at a scale and speed nothing in its earlier history anticipated. Every human group that has ever organized itself has had to decide who counts as in and who counts as out — the foundational question underneath every state that came after. Migration is that same question, run at planetary scale, under planetary pressure: climate displacement, conflict, and information moving people and the knowledge of where else to go faster than any border regime built for a slower world was designed to process.

Harder borders raise the cost of the crossing. They do not touch what is driving it. A wall, a patrol boat, a detention facility changes the price of movement, not the climate pattern, the conflict, or the economic collapse pushing someone to pay it. States spend enormously on the former because the latter is harder, slower, and less politically satisfying to address — and because the machinery of a state is, in the end, built to manage populations, not to solve the conditions that move them.

Capital, for its part, is largely indifferent to the human stakes of any of this. It wants labour where labour is cheap and capital mobile where capital needs to be, and it has historically found ways to get both regardless of which states are trying to restrict migration and which are trying to encourage it. The people crossing borders are, to capital, a labour supply question before they are anything else. The state manages the border. Capital manages what crosses it, and on what terms, largely on its own timetable.

The question underneath all of it has never changed: who counts as one of us, and what happens to everyone else. A kinship group asked it first, with no bureaucracy, no state, and no capital yet in existence to answer it. Every institution examined since was, in one way or another, built to keep answering the same question at a larger scale. That is not a question this piece was ever going to resolve. It is the world its reader is actually standing in.

Methodology note: this piece was written collaboratively between a human and an AI — the human providing the instincts, provocations, editorial judgement and voice; the AI providing research synthesis, intellectual scaffolding and drafting. Full method at [athomehefeelslikeatourist.blog].