The most common explanation for disagreement is also the least useful: the other side is thick, or lying, or captured. It asks nothing further of the person offering it, which is most of its appeal. And it is sometimes true. People do reason badly. Institutions do shape what their members find sayable. A picture of public argument that denied either would be the poorer for it.
If that were the whole story, though, disagreement would not look the way it does. The sharpest disputes are rarely between the informed and the ignorant. They are between people who have read the same things and drawn opposite conclusions - two economists, two historians, two neighbours who follow the news closely. Ignorance produces confusion, not conviction. The confident, opposed certainty that defines our worst arguments needs a different account.
Four things that might be happening
When two reasonable people hold firm and opposite views, one of four things is usually going on, and they are not equally interesting.
The first is cognitive: one side has made an error - a bad inference, a misread statistic, a fact simply wrong. This happens, and where it is happening the argument is easy, because evidence settles it and one party can be shown to be mistaken.
The second is social: people hold the belief their group holds, because the cost of holding another is exclusion. This is real and widespread, and it is worth naming, but it explains the belief without touching its truth. A view can be both socially enforced and correct.
The third is that the two sides weigh the same facts differently because they rank competing goods differently - liberty against security, the present against the future, the harm you can see against the harm you cannot. Nobody here is confused. They agree on what is true and disagree on what matters more.
The fourth is institutional: the information each side receives has already been shaped upstream, by what gets funded, published, and taught. This too is real. It is also the one most prone to abuse, because it can be stretched to dismiss any inconvenient claim as propaganda without ever engaging it.
Most durable disagreement - the kind that does not yield to another chart - is the third kind wearing the costume of the first. It looks like a fight about facts. It is a fight about which harm counts.
The test that does the work
The single most useful thing to ask of any heated dispute is whether it is empirical or normative. An empirical question has an answer the world could in principle supply: did the figure rise or fall, did the policy raise wages or not. A normative question asks what we should value, and no quantity of data answers it.
Where a disagreement is empirical, treating it as a matter of values is a way of avoiding being wrong - each side retreats into its priors and calls the other biased. Where a disagreement is normative, treating it as empirical is worse, because it produces the illusion that more evidence will end it, and the argument grinds forever on a question that was never going to be settled by measurement. A great deal of public life is two people demanding the other accept a value as if it were a fact.
The debt that comes with thinking for yourself
There is a moment, usually when a person feels secure enough to stop minding what a room thinks of them, where they begin to trust their own reasoning over the consensus around them. That shift is mostly a gain. It is also a loan, and the interest is a duty most people who take it out forget to pay: the moment your beliefs stop being audited by a crowd, you have to audit them harder yourself, against the evidence most likely to prove you wrong. Independence without that discipline is not clear sight. It is just a smaller, more flattering consensus of one.
And there is a trap on the same road worth marking, because it is the mirror image of the error it despises. Noticing that institutions can be captured, that funding shapes findings, that narratives get enforced - all true - does not license the conclusion that every disagreement is therefore propaganda and every disagreer a dupe. That is the second kind of explanation smuggled back in as a universal solvent, and it dissolves the thinker's own view along with everyone else's, if he ever turned it on himself.
Which leaves the useful part
Almost none of this crowns a winner, which is the point. The accurate account of why people disagree offers no villain - no side that is simply stupid, no plot that explains the others away. It offers something less satisfying and more usable: a way to tell, in any given fight, whether you are arguing about what is true, in which case evidence can end it and someone should change their mind, or about what matters more, in which case the honest move is to say so and argue the values in the open, rather than pretending the facts were ever going to do it for you.
Most arguments would be shorter, and a few would even resolve, if both sides said first which one they were having.
How this piece was made
How this piece was made. The opposing view - that disagreement is mostly stupidity, dishonesty or capture - is stated at full strength and partly conceded, because each of those does occur; the turn is that they cannot account for confident disagreement between the well-informed. The four-way distinction (cognitive, social, normative, institutional) and the empirical/normative test are standard tools in epistemology, used here as method, not authority. No live statistics are relied on, so nothing here rests on a figure a reader would need to check; the claims are structural. A critic should press hardest on the assertion that most durable disagreement is normative rather than empirical - that is a generalisation, offered as the better bet, not a proof.