# How Governments are Actually Reformed
Reform is closed not because reformers are weak but because the selection process selects against them — every route to a lever runs through people who profit from the capture, so the ones who reach it have been shaped by a career that rewarded not pulling it, which is why "elect better people" and "pass better laws" keep producing the same output from a structure that is very good at appearing to try.
So you don't defeat the incumbent, you make it irrelevant: build structures outside its permission that deliver the same service better and cheaper, publish what the extraction costs in numbers rather than grievance, attach personal liability to decisions so capture lands on individuals instead of institutions, and put sunset on everything so nothing survives by default — all boring, legal, slow, and all working by shrinking the set of paths available to the extractor, which is the same mechanism it uses on you pointed the other way.
Once you have the system already running, then you are in the position to fund a militia to shut down the corrupt system that doesn't work, and the results are (relatively) bloodless. In fact, history shows that the ones that don't even bother with a militia even at the end are the best. Citations:
## Claim 1 — selection filters reformers
- Robert Michels, Political Parties (1911) — the iron law of oligarchy. The founding citation for "every organisation develops a leadership stratum with interests separate from its members'".
- Gaetano Mosca, The Ruling Class (1896) and Vilfredo Pareto, Trattato di sociologia generale (1916) — the classical elite theory this sits inside.
- Robert Merton, "Bureaucratic Structure and Personality", Social Forces (1940) — "trained incapacity": the office shapes the person who holds it. This is the precise citation for "shaped by a career that rewarded not pulling it".
- Robert Niskanen, Bureaucracy and Representative Government (1971) — the formal budget-maximising model.
Anthony Downs, Inside Bureaucracy (1967) — conservatism as a function of organisational position.
- Mancur Olson, The Rise and Decline of Nations (1982) — distributional coalitions accumulate and ossify; the best single citation for "ballooning".
- Acemoglu & Robinson, Why Nations Fail (2012) — extractive institutions persist by design, not accident.
- LaPira & Thomas, Revolving Door Lobbying (2017) — the empirical mechanics of the door.
## Claim 2 — make the incumbent irrelevant
- Albert Hirschman, Exit, Voice, and Loyalty (1970) — the theoretical spine of the whole paragraph.
- Jadwiga Staniszkis, Poland: Self-Limiting Revolution (1984) — the concept you're reaching for: build the alternative, refuse to seize power until it exists. Read this one first.
- Timothy Garton Ash, The Polish Revolution: Solidarity (1983) and Padraic Kenney, A Carnival of Revolution: Central Europe 1989 (2002) — the parallel-institution cases.
- Rein Taagepera, Estonia: Return to Independence (1993) — the Citizens' Committees registering people outside Soviet legality.
Charles Tiebout, "A Pure Theory of Local Expenditures", JPE (1956) — exit as a disciplining mechanism, formalised.
- Barry Weingast, "The Economic Role of Political Institutions", JLEO (1995) — market-preserving federalism: jurisdictional competition as constraint.
Louis Brandeis, Other People's Money (1914) — "sunlight is the best disinfectant", the origin of the publish-the-costs argument.
- Ben Worthy, The Politics of Freedom of Information (2017) — how institutions resist transparency, which is the useful half.
- Texas Sunset Advisory Commission (1977– ) — the only large-scale empirical test of sunset clauses. Go to their reports directly; the academic evaluation literature is thinner and mostly dated.
- Charter of the International Military Tribunal, Articles 6–7 (1945) — the founding modern case for personal liability piercing institutional immunity.
## Claim 3 — militia last, relatively bloodless
- Staniszkis again, plus Ash — Poland 1989, the round table.
- Taagepera and Anatol Lieven, The Baltic Revolution (1993) — 1990–91 restoration, barricades as defence.
- Wheaton & Kavan, The Velvet Revolution (1992) — Czechoslovakia 1989.
Kenneth Maxwell, The Making of Portuguese Democracy (1995) — 1974, the closest thing to your sequence executed quickly.
- Samuel Huntington, Political Order in Changing Societies (1968) — praetorianism: what happens when the militia step arrives before the administration.
- Charles Tilly, Coercion, Capital, and European States (1990) — how armed groups become permanent states.
One caveat that matters. The bloodless cases are precisely the ones where the militia step was never taken — Poland, Estonia, Czechoslovakia. So they're evidence that a functioning parallel administration makes force unnecessary, not that the militia step is safe when used. Your paragraph asserts the latter and the citations support the former
If you want the strong version, the claim has to be "the militia is insurance whose non-use is the success condition", because there is no good historical sample of "parallel administration plus militia, bloodless". Huntington and Tilly are the citations for why that sample doesn't exist.
Reform is closed not because reformers are weak but because the selection process selects against them — every route to a lever runs through people who profit from the capture, so the ones who reach it have been shaped by a career that rewarded not pulling it, which is why "elect better people" and "pass better laws" keep producing the same output from a structure that is very good at appearing to try.
So you don't defeat the incumbent, you make it irrelevant: build structures outside its permission that deliver the same service better and cheaper, publish what the extraction costs in numbers rather than grievance, attach personal liability to decisions so capture lands on individuals instead of institutions, and put sunset on everything so nothing survives by default — all boring, legal, slow, and all working by shrinking the set of paths available to the extractor, which is the same mechanism it uses on you pointed the other way.
Once you have the system already running, then you are in the position to fund a militia to shut down the corrupt system that doesn't work, and the results are (relatively) bloodless. In fact, history shows that the ones that don't even bother with a militia even at the end are the best. Citations:
## Claim 1 — selection filters reformers
- Robert Michels, Political Parties (1911) — the iron law of oligarchy. The founding citation for "every organisation develops a leadership stratum with interests separate from its members'".
- Gaetano Mosca, The Ruling Class (1896) and Vilfredo Pareto, Trattato di sociologia generale (1916) — the classical elite theory this sits inside.
- Robert Merton, "Bureaucratic Structure and Personality", Social Forces (1940) — "trained incapacity": the office shapes the person who holds it. This is the precise citation for "shaped by a career that rewarded not pulling it".
- Robert Niskanen, Bureaucracy and Representative Government (1971) — the formal budget-maximising model.
Anthony Downs, Inside Bureaucracy (1967) — conservatism as a function of organisational position.
- Mancur Olson, The Rise and Decline of Nations (1982) — distributional coalitions accumulate and ossify; the best single citation for "ballooning".
- Acemoglu & Robinson, Why Nations Fail (2012) — extractive institutions persist by design, not accident.
- LaPira & Thomas, Revolving Door Lobbying (2017) — the empirical mechanics of the door.
## Claim 2 — make the incumbent irrelevant
- Albert Hirschman, Exit, Voice, and Loyalty (1970) — the theoretical spine of the whole paragraph.
- Jadwiga Staniszkis, Poland: Self-Limiting Revolution (1984) — the concept you're reaching for: build the alternative, refuse to seize power until it exists. Read this one first.
- Timothy Garton Ash, The Polish Revolution: Solidarity (1983) and Padraic Kenney, A Carnival of Revolution: Central Europe 1989 (2002) — the parallel-institution cases.
- Rein Taagepera, Estonia: Return to Independence (1993) — the Citizens' Committees registering people outside Soviet legality.
Charles Tiebout, "A Pure Theory of Local Expenditures", JPE (1956) — exit as a disciplining mechanism, formalised.
- Barry Weingast, "The Economic Role of Political Institutions", JLEO (1995) — market-preserving federalism: jurisdictional competition as constraint.
Louis Brandeis, Other People's Money (1914) — "sunlight is the best disinfectant", the origin of the publish-the-costs argument.
- Ben Worthy, The Politics of Freedom of Information (2017) — how institutions resist transparency, which is the useful half.
- Texas Sunset Advisory Commission (1977– ) — the only large-scale empirical test of sunset clauses. Go to their reports directly; the academic evaluation literature is thinner and mostly dated.
- Charter of the International Military Tribunal, Articles 6–7 (1945) — the founding modern case for personal liability piercing institutional immunity.
## Claim 3 — militia last, relatively bloodless
- Staniszkis again, plus Ash — Poland 1989, the round table.
- Taagepera and Anatol Lieven, The Baltic Revolution (1993) — 1990–91 restoration, barricades as defence.
- Wheaton & Kavan, The Velvet Revolution (1992) — Czechoslovakia 1989.
Kenneth Maxwell, The Making of Portuguese Democracy (1995) — 1974, the closest thing to your sequence executed quickly.
- Samuel Huntington, Political Order in Changing Societies (1968) — praetorianism: what happens when the militia step arrives before the administration.
- Charles Tilly, Coercion, Capital, and European States (1990) — how armed groups become permanent states.
One caveat that matters. The bloodless cases are precisely the ones where the militia step was never taken — Poland, Estonia, Czechoslovakia. So they're evidence that a functioning parallel administration makes force unnecessary, not that the militia step is safe when used. Your paragraph asserts the latter and the citations support the former
If you want the strong version, the claim has to be "the militia is insurance whose non-use is the success condition", because there is no good historical sample of "parallel administration plus militia, bloodless". Huntington and Tilly are the citations for why that sample doesn't exist.
5❤️1