Some government friction is there on purpose
The U.S. federal government makes consequential decisions through a system that frequently prevents any one part of government from acting alone.
Congress can pass legislation, but the president can veto it. Congress can override that veto, but only with a two-thirds vote in both houses. The president nominates Supreme Court justices and other federal officers, but many appointments require Senate confirmation. Congress controls legislative powers while the executive is charged with faithfully executing the laws. Federal judicial power sits in a separate branch.
We usually describe this arrangement through separation of powers and checks and balances. The National Archives describes the system of checks and balances as preventing any one of the separate powers from becoming dominant.
I've been wondering what happens if we examine the same architecture as a coordination system.
From that perspective, the federal government contains something we often try to remove from organizations: friction. Authority is distributed. Important actions cross boundaries. Some decisions require multiple actors. Other actors can constrain or reverse an action. Changing the rules themselves can be intentionally difficult.
That suggests a useful distinction. A coordination system doesn't necessarily need to minimize friction. Some friction may exist because the system is trying to make certain actions difficult.
The branches create boundaries around authority
The first three articles of the Constitution establish legislative, executive, and judicial power in different parts of the federal government.
Article I vests the legislative powers granted by the Constitution in Congress. Article II vests executive power in the president. Article III vests federal judicial power in the Supreme Court and the lower federal courts Congress establishes.
Those boundaries don't make the branches independent of one another.
A bill must pass Congress before being presented to the president. The president can return it with objections, and Congress can enact it over that objection with sufficient support. The president nominates Supreme Court justices, but appointment requires the advice and consent of the Senate. The House has the sole power of impeachment, while the Senate has the sole power to try impeachments.
These are dependencies between parts of government that retain distinct authority.
In an organization, we might look at a consequential decision that requires several functions, multiple approvals, and the possibility of escalation and ask whether the process has become unnecessarily complicated.
Sometimes it has.
But the constitutional example makes the opposite possibility unusually visible: the dependency may be part of the design.
Coordination efficiency isn't the only property that matters
Imagine evaluating the federal government only by how quickly a preference can become action.
Requiring legislation to pass two chambers creates delay. Giving the president a veto creates another potential interruption. Requiring Senate participation in appointments prevents the president from making certain appointments unilaterally. Amending the Constitution requires considerably more agreement than passing ordinary legislation.
Those mechanisms would look inefficient if throughput were the only thing we valued.
It isn't.
The constitutional architecture distributes governmental power and creates mechanisms through which one part of government can constrain another. The National Archives describes the purpose of checks and balances in precisely those terms: preventing one of the separate powers from becoming dominant.
Seen as a coordination system, this suggests that coordination quality can't be understood simply as the absence of friction.
The more useful question may be whether the system makes necessary action possible while appropriately constraining actions that should be difficult.
Constraints can be part of what makes the system work
Organizations frequently treat constraints as obstacles to execution.
An approval slows the project down. Security blocks a deployment. Finance won't release the money. Legal won't approve the contract. Another team has to participate before a decision can move forward.
Sometimes those are genuine coordination failures. Authority may be unclear. A dependency may have no owner. Information may not move across a boundary. A process may have accumulated approvals that no longer protect anything important.
But removing friction isn't automatically an improvement.
A production deployment may require additional review because the consequences of failure are large. A financial transaction may require separation of duties because concentrating authority creates unacceptable risk. A major acquisition may need approval from multiple parties because the organization intentionally wants a higher threshold for an irreversible decision.
The government example makes this easier to see because many of its constraints are explicit.
The constraint isn't necessarily interfering with the system.
The constraint may be part of the system.
The Constitution also specifies how coordination crosses boundaries
There is another feature of this architecture that I find interesting.
The Constitution doesn't only distribute authority among institutions. In several places, it specifies how those institutions must interact.
The president nominates; the Senate provides advice and consent.
Congress passes a bill; the president signs it or returns it with objections; Congress can reconsider it under a higher voting threshold.
The House impeaches; the Senate conducts the trial.
Congress proposes constitutional amendments under one route; states determine whether those amendments are ratified.
These aren't merely descriptions of organizational responsibilities. They establish conditions under which authority crosses a boundary between otherwise distinct parts of the system.
I've been exploring a possible concept in Recursive Coordination Theory that I currently call a Coordination Contract: the structural specification governing how coordination occurs when autonomous coordination systems depend on one another.
I'm not yet claiming that constitutional mechanisms establish or validate that construct. But they provide an unusually explicit place to investigate it.
The interesting questions become: What authority belongs on each side of a boundary? What requires participation from both? What information has to cross? Which constraints govern the interaction? What happens when the parties disagree? What mechanisms exist when one part of the system exceeds its authority?
Organizations have versions of these questions too. Their answers are simply much less likely to have been written into a constitution.
Even adaptation is constrained
The system can change, but changing different parts of it requires different levels of coordination.
Ordinary legislation can alter federal law through the legislative process. Elections change many of the people exercising governmental authority. Appointments change others.
Changing the Constitution itself requires a substantially higher threshold.
Article V provides two routes for proposing amendments and requires ratification by three-fourths of the states. That makes constitutional adaptation possible while deliberately making it harder than ordinary lawmaking.
This is another useful coordination-system property.
Adaptation capacity doesn't necessarily mean that a system should be easy to change. A system may need some state to change quickly and other state to remain stable unless unusually broad agreement exists.
Organizations face the same design problem at a different scale. Some decisions should be reversible by a team. Others should require executive approval. Still others may change commitments, ownership, capital structure, security posture, or customer obligations deeply enough that a higher threshold is appropriate.
The design question is not simply whether the organization can adapt.
It is whether the cost of adaptation is appropriate to what is being changed.
Leaders should examine what their friction protects
This creates a different way to investigate organizational friction.
Take an approval, handoff, dependency, review, or escalation that regularly slows work down and ask:
What failure is this constraint intended to prevent?
If nobody can answer, the friction may be coordination debt. The organization may be paying an ongoing coordination cost for a constraint whose original purpose has disappeared.
If the answer is clear and the protected consequence still matters, removing the friction may make the organization faster while making the system worse.
The next question is whether the constraint is proportional to the failure it contains. Requiring three executives to approve a reversible $500 decision probably doesn't become good system design merely because someone can explain why the approval exists.
The objective is not maximum constraint.
It is intentional constraint.
Good coordination isn't necessarily frictionless coordination
The U.S. federal government is obviously much more complicated than three branches interacting through a constitutional document. Each branch contains many coordination systems of its own. Federal agencies interact with Congress, courts, the White House, states, regulated organizations, and citizens. Federalism adds another set of boundaries between federal and state authority.
That complexity makes government an interesting environment for studying coordination rather than a simple analogy for organizational design.
But the basic signal is useful.
Some of the characteristics that look dysfunctional when we encounter them inside an organization can serve a legitimate structural purpose. Distributed authority can prevent unilateral action. Dependencies can force participation by another interest. Higher decision thresholds can protect consequential state. Slow adaptation can preserve stability.
That doesn't mean organizational bureaucracy should be defended as "checks and balances." It means friction needs to be diagnosed before it is removed.
A coordination system has to make action possible.
It may also need to make some actions appropriately difficult.
