When two experts disagree, the disagreement is the information

Disagreement isn’t a scandal

A field where every specialist agrees is either dealing with a settled question or not looking very hard. On live questions, disagreement among competent people is the normal condition, and the public expectation that expertise should produce a single answer misunderstands what expertise is for. It narrows the range of defensible positions; it rarely collapses it to one.

The useful response is not to average the experts, and not to pick the one who sounds most certain. It is to work out what they are disagreeing about, because the reasons fall into a few recognisable categories and each has different implications for what you should do.

They are looking at different evidence

The most common and most tractable source of disagreement is that two specialists have read different literatures, or the same literature at different times, or one has access to unpublished or practical experience the other does not. This is boring and fixable, and it is worth asking about first, because the resolution is simply to find out what each has seen.

A related version is that the evidence base is thin enough that reasonable readings diverge. When studies are few, small or conflicting, two careful people can weigh them differently without either being careless. In that situation the disagreement is telling you the truthful thing: the answer is not known, and any confident public position is doing something other than reporting the evidence.

They are answering different questions

A surprising share of expert disagreement dissolves once the question is stated precisely. Whether a treatment "works" can mean whether it changes a measurement, whether it changes how someone feels, whether it does so on average, or whether it does so for a particular group at a particular stage. Two specialists can hold entirely compatible views and appear to be in open conflict.

This is especially common across disciplinary boundaries, where the same word carries different technical content. Economists, engineers and clinicians use "risk", "efficient" and "significant" in ways that do not translate directly, and a debate conducted in those terms can run for years without the participants noticing that they are describing different quantities.

They differ on values, not facts

Many disputes presented as technical are disputes about what to optimise. How to trade a small chance of a large harm against a large chance of a small one, whose costs count, how much weight the future gets — these are not settled by data, and no amount of additional research resolves them.

It is worth separating the two components explicitly whenever you can, because the factual part is usually narrower than the argument implies and the value part is usually where the real disagreement lives. An expert has no special authority on the second, and pretending otherwise is one of the reliable ways for expertise to lose its credibility.

They have different incentives and different exposure

People are influenced by what they work on, who funds it, and what a change of position would cost them professionally. This is worth weighing without descending into the assumption that every position is bought, which is both usually false and conveniently unfalsifiable.

The more informative version of the question is about exposure to consequences. Somebody who will personally deal with the results of being wrong tends to hold a different kind of opinion from somebody who will not, and in an unsettled field that difference is often a better guide than credentials. It isn’t a guarantee — practitioners have their own systematic biases, drawn from the cases they happen to see.

What a non-specialist can actually do

You will not adjudicate a technical dispute from outside it, and attempting to is how people end up confidently attached to whichever account they read most recently. What is available is more modest and more useful: identify which of the categories above the disagreement belongs to, look for what both sides concede, and notice whether either side can state the other’s position accurately.

That last test is worth more than any credential. Someone who can articulate the strongest version of the opposing case, and say specifically what would change their mind, is doing the thing expertise is supposed to consist of. Someone who can only characterise the other side as ignorant or motivated is telling you that the disagreement has stopped being about evidence, whatever it started as.