Two percent of Americans say they trust the federal government to do what is right just about always. Another 15% say most of the time.1 Gallup's average confidence across America's major institutions has now sat below 30% for five consecutive years, with trust in media at 28%.2
Those are the numbers everyone quotes. The reading attached to them is usually some version of cynicism: people have grown suspicious, polarization did it, social media did it, the scandals did it.
The direction is right and the mechanism is wrong. Trust did not fall because Americans became cynical. It fell because they became accurate, and then they adapted — and the adaptation is what costs money.
Honesty Was Load-Bearing, Not Admirable
Integrity was never widespread. It was structural. Courts, contracts, journalism, partnerships, elections and ordinary conversation all ran on a working assumption that truth-telling was more common than deception — not because people were good, but because the systems could not price every interaction individually.
That assumption was the cheapest input in the economy, and it has been drained slowly enough to look like weather.
The General Social Survey has asked Americans the same question since 1972: can most people be trusted? Forty-six percent said yes then. Thirty-four percent said yes by 2018, and an identical 34% told Pew the same thing in 2023–24.3 Averaged by decade, the series fell in every single decade — 44.8% in the 1970s to 28.7% in the 2020s, a slide of roughly a third of a percentage point every year for fifty years.3
Nothing in that curve looks like a crisis. That is the problem with it.
Honest People Convert Into Auditors
When honesty is exploited often enough, offering it freely stops being virtuous and starts being negligent. But the adaptation is not what the cynicism reading predicts.
Honest people rarely convert into liars. They convert into auditors.
They qualify their statements, withhold information, speak in safe abstractions rather than precise language, and begin treating bad faith as a default variable rather than an exception. That is not moral decay. It is Bayesian updating, and each individual instance of it is correct.
The asymmetry is what makes it expensive.
The dishonest actor prices friction into their strategy from the beginning. The honest actor absorbs it as overhead.
A virtue that only survives when convenient is not a virtue. It is a preference, and preferences yield under pressure. Once optimization becomes the dominant moral framework, everything is negotiable, including truth — and the people who kept their side of the bargain are the ones who pay for the verification layer that replaces it.8
The Overhead Is on the Books
The verification layer is not a metaphor. It has line items.
The U.S. tort system consumed roughly $529 billion in 2022 — about 2.1% of GDP, or $4,207 per household — and only about half of that money reached the plaintiffs alleging injury. The rest went to administration and fees. Costs have been compounding at an average 7.1% a year since 2016.4 Legal services run around 1.3% of GDP on their own.5 Tax compliance costs the economy something on the order of $546 billion, close to another 2%.6
Not all of that is distrust. Some is genuine complexity, and the tort figure comes from an organization that argues for tort reform, which is worth saying before someone else says it.4 But the direction is the part that matters: a system in which cooperation were cheap would not need an overhead line growing at 7% a year.
The economics here is not speculative either.
Knack and Keefer's 1997 study of 29 market economies found trust and civic norms associated with measurably stronger economic performance, working through exactly this channel — social trust lowers transaction costs.7 Zak and Knack later modeled the size of it: growth rises by close to a full percentage point for every 15-point increase in trust.8
Warning
Run that backwards against a 16-point fall and the implied drag is not rhetorical.
What Collapses First Is Conversation
Markets do not go first. Morality does not go first.
The honest founder stops building in public. The thoughtful citizen stops speaking plainly. The insider shares the useful version only inside a trusted circle. None of that appears in any index, because you cannot count the ideas that never circulated or the trust that never compounded.
The winners of extractive systems count what they captured. They do not count what retreated.
High-trust societies compound because voluntary cooperation is cheap. Low-trust societies litigate. Every interaction gets heavier, every agreement longer, every transaction slower, and the system thickens with defensive layers that are mistaken for sophistication.
It is decline disguised as maturity.
The Objection
The United States has become dramatically richer across the same fifty years in which trust fell by sixteen points. If trust were load-bearing in the way this argument claims, that should not be possible.
But two things make it possible, and only one of them is comforting.
-
The trust literature is about growth rates, not levels. A drag of a fraction of a point a year disappears inside technology gains an order of magnitude larger — which means a real cost can be fully paid and never once show up as a bad year. The compliance and litigation curves are where it surfaces, and those are rising faster than output.
-
The second is more awkward. Knack and Keefer found that trust is strongest in countries whose formal institutions reliably protect property and contract rights.7 That points the causal arrow the other way: functioning institutions may produce trust rather than depend on it. If so, the fix is not cultural exhortation about honesty. It is enforcement — which is the one variable that was quietly abandoned.910
We never formally repealed integrity. We stopped treating the expectation of it as load-bearing, tolerated the absence, normalized the absence, and then optimized around it.
The Wall
Societies do not collapse because lies exist. They erode because truth becomes privately rationed — available inside the circle, priced outside it.
That is what the 2% is measuring. Not an opinion about politicians, but a population that has finished updating and is now acting on the update: hedging participation, engaging because it must rather than because it trusts, and paying the verification premium as a cost of doing business in its own country.
These are not isolated scandals. They are correlated stress fractures in the same load-bearing wall.
And the wall does not fail all at once. It fails when enough people decide it is safer not to lean on it.