The votes for Trump and Brexit have highlighted a division between "elites" and the "people". For me, though, this is the wrong dichotomy. The question instead is: under what conditions are the people right, and under what conditions are elites right?

Both can sometimes go wrong. Experts are prone not just to professional deformation (pdf) - the tendency for perceptions to be warped by their training - but also groupthink. For example, the replication crisis - which is by no means confined to psychology - suggests that peer review fails to weed out poor academic research and might even enhance groupthink.

What's more, pretences to expertise can often merely be a desire for wealth and power, as Alasdair MacIntyre wrote:

The realm of managerial expertise is one in which what purport to be objectively-grounded claims function in fact as expressions of arbitrary, but disguised, will and preference. (After Virtue, p107)

But on the other hand, we know that the people are often wrong about basic facts, are terrible at understanding connections between economic phenomena, are misinformed by a biased media, and prone to countless cognitive biases.

Rather than always side with the people or always with elites, we should abandon grandiose generalizations and ask in each specific context: who is most likely to be right here - the elite or the people? Here are some tests I'd apply.

First, are the conditions in place for the wisdom of crowds to operate? Namely, are individuals' judgments diverse, uncorrelated and decentralized?

In economic forecasting, I think this is the case. Each individual's decision on how much to spend is based upon specific private information about his personal future. Aggregate data (pdf) on consumption-wealth ratios thus do a good job of gathering together dispersed fragmentary information about our future prospects. I'd rather trust the crowd, therefore, than economic forecasters.

This isn't to say the crowd is perfect: spending decisions can be influenced by peers. But nothing's perfect. The question is one of the relative size of errors.

Secondly, are beliefs motivated by nasty or self-interested preferences?

This can be true of elites. Bosses' hostility to worker democracy and fund managers' claims to be able to beat the market owe more to self-interest than fact. George Monbiot says the same is true of oil companies who finance climate change deniers. Equally, though, it can also be true of the public: antipathy to immigration might in some cases be founded upon a dislike of foreigners rather than purer motives.

Thirdly, are there obvious errors and biases at work here?

Everybody - expert or layman - is prone to cognitive bias. I prefer to look for errors and then ask: what biases might lead people into error?

I suspect this is the case with the public's belief that immigration has significantly pushed down wages and job prospects. It is due in part to the post hoc ergo propter hoc fallacy ("some immigrants came last year and now I can't find a decent job"); to the failure to see that immigration can raise natives' wages by allowing them to do more skilled jobs; and to an ignorance of the myriad other forces depressing pay.

This isn't to say the public need always be wrong on immigration. If people in areas of high immigration offered factual evidence about (say) how migration is eroding social capital, I would take that as evidence that locals possessed a "ground truth" that elites had overlooked.

Relatedly, I'd ask how much effort people have taken to slough off cognitive biases. In David Davis's case, the answer seems to be: none.

Fourthly, is this a statement of fact or opinion?

It's possible in theory that the EU referendum might have gathered thousands of fragmentary facts. People might have figured "the EU stops me doing x, y, or z" (or facilitates them) and voted accordingly. But this isn't, for the most part, what we got. Instead, we had a poisonous mixture of racism, lies and hyperbole.

One reason why I favour worker democracy more than plebiscites is that the former is a way of gathering dispersed information - of aggregating marginal gains about corporate performance - whereas the latter are not.

Now, I'm not saying these tests are perfect. Nothing is. And we might disagree from context to context on how to apply them. Sadly, however, these tests seem not to be applied in politics. Which is one reason why we see the worst of public opinion being expressed rather than the best.