If Men Were Angels: Separation of Powers

The Sentence Madison Wrote About You

Lesson 8.2 ended with a question: why design a system on the assumption that the person in the chair cannot be trusted? You may have felt the sting in it. Every leader reading this believes he is basically honest, and most of you are right. The answer is in one paragraph published in the New York Packet on Friday, 8 February 1788, in the essay we know as Federalist 51.

'If men were angels, no government would be necessary. If angels were to govern men, neither external nor internal controls on government would be necessary. In framing a government which is to be administered by men over men, the great difficulty lies in this: you must first enable the government to control the governed; and in the next place oblige it to control itself.'

Read it twice, because almost everyone misquotes the point. Madison is making two moves, not one. The first sentence justifies government at all — if people were angels you would not need it. The second sentence is the one people skip, and it is the harder half: if angels were governing, you would not need controls on the government either. So the argument cuts in both directions simultaneously. Human nature makes government necessary. The same human nature makes constraints on government necessary. You cannot use the first half to expand power and then decline the second half.

Then comes the mechanism, four words that have organized American institutional design for two centuries: 'Ambition must be made to counteract ambition.' Notice what Madison is not saying. He is not saying select better people, though selection matters. He is not saying write stronger rules, though rules matter. He is saying build a structure in which the self-interest of one officeholder is the operating constraint on the self-interest of another — so the system does not depend on virtue that may not arrive. Federalist 10 runs the same logic on faction: you will not eliminate the causes of faction without destroying liberty, so you control its effects by enlarging the republic until no single faction can easily assemble a majority.

Federalist 51 also names the second axis, and it is the one Module 8 turns to next: 'In the compound republic of America, the power surrendered by the people is first divided between two distinct governments, and then the portion allotted to each subdivided among distinct and separate departments. Hence a double security arises to the rights of the people.' Two directions of division — across branches and across levels. A New York City mayor lives inside both grids simultaneously, checked horizontally by a Council and a Comptroller and vertically by Albany and Washington, and a candidate who understands only one of the two axes will spend his term surprised.

Montesquieu had supplied the architecture forty years earlier. In The Spirit of the Laws he argues that constant experience shows every man invested with power is apt to abuse it, and that to prevent abuse, power must be arranged so that power checks power. The founders' innovation was less the idea than the engineering — separated branches with overlapping tools, so that each department has both the means and the motive to resist encroachment by the others.

So here is the question this lesson forces on you, and I want you to answer it about yourself rather than about a politician you dislike. Do you believe Madison's premise applies to you? Not in theory. Operationally. If you believe it, there will be evidence in your calendar and your org chart: somebody with real standing to stop you, a decision you cannot make alone, a number you cannot revise by yourself. If there is no such evidence, then whatever you say you believe, you are running your institution on the assumption that you are the exception.

The Angel Assumption

Call the failure The Angel Assumption. It is not the belief that people are good. It is the narrower, more seductive belief that the general rule about human self-interest applies broadly but not here — not to me, not to my team, not to this administration, because our motives are clean and the situation is urgent. It never announces itself. It arrives as an argument about efficiency, or about trust, or about how the previous administration abused the process and we would never.

Scripture's contribution to political theory at this point is not a policy program; it is an anthropology, and it is the most empirically confirmable doctrine Christianity holds. Jeremiah 17:9 states it as a diagnosis: 'The heart is more deceitful than all else and is desperately sick; who can understand it?' Two Hebrew words carry the weight. The word behind 'deceitful,' aqob, shares a root with the name Jacob and with the heel — the sense is crooked, given to supplanting, a thing that trips you from behind. The word behind 'desperately sick,' anush, is used elsewhere of a wound past healing. Put them together and the claim is not that people behave badly sometimes. The claim is that the faculty by which you assess your own motives is itself compromised, which is exactly why the final clause is a question: who can understand it?

That is the doctrine of the fall functioning as political realism, and I want to be precise about what it does and does not license. It does not license contempt for people; Module 1 established that every person bears God's image, and that holds whether or not they behave well. It does not license despair about institutions. What it does is remove a specific move from the board: the move where a leader argues that his own good intentions are sufficient security for the people he governs. Under this anthropology, your good intentions are not evidence, because self-assessment is the very instrument the doctrine says is unreliable.

Now, be honest about the argument's dependencies, because a hostile reader will test them. You do not need the doctrine to reach the design conclusion. David Hume made the same argument in 1742 on entirely secular grounds: in contriving any system of government and fixing the checks of the constitution, every man ought to be supposed a knave and to have no other end in all his actions than private interest. Hume immediately notes that this is odd, since most men are in fact not knaves. That is the point. It is a design rule about tail risk, not a description of the average case. You build the bridge for the heaviest truck that will ever cross it, not the average truck.

This is where separation of powers stops being a civics diagram and becomes institutional humility. A leader who accepts The Angel Assumption's falsity will do something unusual: he will build checks on himself while he still has the power to prevent them, knowing that the version of himself who most needs the check is the version who will most want it gone. That is the whole test, and it is failed almost universally at the moment of maximum popularity — which is precisely when it is cheapest to pass.

What the Objection Gets Right

This academy does not evaluate a position before stating the opposition's case in a form its own advocates would sign, so here is the strongest case against separated powers, and it is stronger than defenders usually admit.

Separated powers make government slow, and slowness is not a neutral cost — it is paid by real people in real time. A shelter that opens in four years instead of one is three years of people sleeping outside. A subway line that takes decades costs riders their hours and the city its growth. Woodrow Wilson argued in Congressional Government in 1885 that the American system diffuses responsibility so thoroughly that no one can be held accountable for outcomes, because everyone can plausibly blame someone else. That critique has real force. Juan Linz argued in 1990 that presidential systems with separated powers and fixed terms are prone to deadlock between two branches each claiming a democratic mandate, with no constitutional way to break the tie — an argument that the parliamentary alternative is not merely faster but more stable.

And New York supplies the sharpest version of the objection in a single career. The reason Robert Moses could build at a scale no one has matched since is precisely that he was insulated from the checks. You can hold two things at once: the insulation was illegitimate, and the output was real. Anyone who tells you the constrained system would have produced the same parkways and bridges on the same timeline is not arguing, he is wishing. The honest defender of checks concedes the cost.

Concede it, then answer it, and the answer has three parts. First, the comparison is rigged if you count only the projects a constrained system failed to build and not the projects an unconstrained system built that should never have been built — the neighborhoods cleared, the communities cut in half by a highway, the decisions that no one could stop because no one had standing. Second, speed and reversibility trade against each other. A system that can act instantly can also act instantly against you, and the same absence of friction that lets a good executive move lets a bad successor move. You are not choosing the speed of your administration; you are choosing the speed of every administration that follows, including the ones you would be terrified of. Third, much of the delay people blame on separated powers is not constitutional at all. It is procedural accretion — review requirements, environmental process, litigation exposure, contracting rules — which can be reformed without touching the constitutional structure. Distinguishing those two is one of the most useful things a serious executive can do, because it lets you fight for speed without arguing for unaccountable power.

Hold the concession and the answer together. The efficiency objection establishes that checks have real costs and that the costs fall on identifiable people. It does not establish that the costs exceed the benefits, and it certainly does not establish what its most enthusiastic users want it to — that this particular administration should be exempt.

What the Research Can and Cannot Show

You will be tempted to reach for evidence here, and the evidence is thinner and more interesting than the talking points suggest.

Start with the identification problem, because it governs everything else. Countries do not receive their constitutions at random. A nation with strong judicial independence also tends to have higher income, older institutions, a particular colonial history, and a dozen other correlated features. So a cross-national regression showing that separation of powers predicts lower corruption cannot tell you which of those things is doing the work. Compounding it, most widely used corruption indicators are perception indices assembled from expert and business surveys rather than counts of corrupt acts, and perception and experience measures do not always move together. Daniel Treisman's survey of a decade of this literature is candid about how much of what looked robust depended on which measure was used.

Now the studies with real identification, which is where you should actually spend your confidence. Benjamin Olken's field experiment in Indonesian villages, published in the Journal of Political Economy in 2007, randomly assigned road projects to a higher probability of government audit and measured missing expenditures by digging into the roads and comparing what was built against what was billed. Announced audits reduced missing expenditures. That is a real causal estimate with a physical outcome measure, and it is about as clean as this field gets. Claudio Ferraz and Frederico Finan, working with Brazil's randomized municipal audits, found that releasing audit results before an election affected the reelection prospects of mayors found to have misappropriated funds, with the effect larger where local radio could carry the information. Independent work has associated press freedom with lower corruption.

State carefully what that body of work supports. It supports the proposition that monitoring changes the behavior of officials, and that the effect depends on the information reaching people who can act on it. It does not establish that any particular constitutional architecture reduces corruption, because none of those studies randomized a constitution. If you use them, use them for the narrow claim. Overclaiming from good studies is how leaders lose the credibility that good studies were supposed to buy them.

There is a second literature worth knowing and worth handling gently: work on veto players, associated with George Tsebelis, which formalizes the intuition that the more actors whose agreement is required, the harder policy change becomes in either direction. That framework is genuinely useful for prediction and it is deliberately silent on whether the resulting stability is good. Stability protects good policy and bad policy identically. Anyone who cites veto-player theory as an argument for their own side has not read it carefully.

Who Actually Stops a Mayor

Now the practical map, because a candidate who cannot name his own constraints will promise things he cannot deliver and then blame the people who stopped him.

Inside the city, the Council is the first real check. Charter section 21 vests the legislative power of the city in it. Section 28 gives it the power to adopt local laws. Section 37 gives the mayor a veto and the Council a two-thirds override. On money, section 249 puts the executive budget in the mayor's hands and section 254 lets the Council increase, decrease, add, or omit units of appropriation; section 255 gives the mayor five days to disapprove increases and the Council a two-thirds override. Section 29 gives the Council investigatory power over matters within its jurisdiction and requires standing committees to review agency activity on a continuous basis. Section 31 conditions certain appointments on advice and consent after a public hearing. That is a legislature with teeth, and any mayor who treats it as an obstacle rather than a counterpart will spend four years losing slowly.

The Comptroller is the second, and independently elected. Charter section 93 carries the audit powers, including a duty to establish an auditing cycle reaching each agency's programs at regular intervals, and the power to take testimony. The registration function is the underrated one: under Charter sections 328 and 375, contracts are not implemented until the Comptroller registers them, which places an independently elected official in the path of the city's procurement. The Public Advocate, under section 24, is a member of the Council without a vote, receives and investigates complaints about city services, may obtain agency records necessary to those investigations, and reports publicly. Borough presidents under section 82 consult on the budget and review land use in their boroughs; community boards under section 2800 — up to fifty members appointed by the borough president, at least half from council member nominees — hold the initial review and public hearing on land use applications before they reach the City Planning Commission.

Then the integrity institutions, and here is the honest complication. Chapter 68 establishes the Conflicts of Interest Board: five members appointed by the mayor with Council approval, serving six-year terms, administering the prohibited-conduct rules in section 2604. The Department of Investigation, under Charter sections 801 and 803, is empowered to make any study or investigation the commissioner believes is in the best interests of the city, including investigations of the affairs, functions, accounts, methods, personnel, or efficiency of any agency. Now notice the structural tension: the DOI commissioner is a mayoral appointee, and section 801 gives the mayor his own removal power over that specific office — narrower than it looks, since section 6 itself excludes 'officers for whose removal other provision is made by law,' and section 801 is exactly such other provision: the mayor must file his reasons and allow the commissioner an opportunity to make a public explanation before removing him. The city's chief investigator investigates the administration that can dismiss him, even on a record the administration has to create in public. Whether that check holds has historically depended less on the Charter than on whether removing the investigator would cost the mayor more than the investigation would.

Beyond the city there are checks a mayor cannot touch at all. District attorneys are independently elected and federal prosecutors answer to no one in City Hall. Courts constrain the city continuously, and the discipline for a serious candidate is to read the opinions rather than the coverage — the liability opinion in Floyd v. City of New York, decided in the Southern District of New York in 2013, is a case in point on police practice. The State sets the frame: home rule powers flow from Article IX of the New York Constitution and the Municipal Home Rule Law, the MTA is a state authority, school governance depends on Albany's renewal, and most tax changes require the Legislature. And in extremity the State has taken the city's finances out of the mayor's hands entirely, as it did with the control board structures created during the 1975 fiscal crisis.

Here is the disposition I want you to leave with. Do not learn this map so you can complain about it. Learn it so that when you promise something you can say who else must agree, and so that when you are stopped you can tell the difference between an institution doing its job and an institution being misused. Lesson 8.4 takes up the boundary that is hardest for a leader of deep conviction, and the one this city will examine you on most closely: the line between the church and the state.