The theme of this post is bad comparisons. We start with Alex Tabarrok in what has been a continuing examination by him of Britain’s insane (I don’t use that word lightly) attempt through its Equality Act 2010 to second guess the labor market. He writes [emphasis added],
Suppose that apples sell for more than oranges and Parliament in it’s wisdom decides that, at last, apples and oranges must be compared. Not by shoppers — shoppers are biased, they merely reveal what they are willing to pay — but by a tribunal, which will determine whether apples and oranges are of truly equal value and thus must sell at the same price.
What would the tribunal need to know?
. . .
An employer who defrauds his workers, an employee who steals from the till, a product sold under false pretenses — condemn them, take them to court. But the pattern of prices that emerges from millions of voluntary trades is nobody’s conduct. No one chose it, no one designed it, no one can be guilty of it. The constellation of prices is, in Ferguson’s phrase, the result of human action but not of human design. Demanding that prices be just is a category error, like suing the weather. Prices don’t grade our merit; they guide our actions. Ask them to do the first and they can no longer do the second.
And of course there are unintended (but not unforeseen) consequences:
Here’s the problem, which contributed to Birmingham going bankrupt in 2023. The council employs roughly 400 binmen and something like 6,000 women in comparable graded roles. Every extra pound paid to a binman therefore implied about fifteen pounds owed to the cooks and cleaners. The council simply didn’t have the money to pay everyone binman wages so they cut the binmen’s wages. But the market wage for collecting rubbish is what it is, so when Birmingham finally deleted the premium in January 2025, the binmen went on strike — and they are still on strike, nearly eighteen months later. The rubbish piled up, 17,000 tonnes of it, and the city declared a major incident.
Here’s the most amazing part. The council hired an outside contractor to take over its rubbish collection and it now pays roughly triple its pre-strike outsourcing bill–more than it was paying its own employees. So much for sexism. Apparently the premium wasn’t a favor to men; it was the price of the job. If you want your rubbish picked up and your graves dug, you must pay the market wage. This was pure regulatory arbitrage, of course. Because the new binmen were contractors they legally had a different employer than the Council’s female caregivers and the Act’s comparator rules stop at the employer’s nexus.
I suffer no fools when it comes to people who want to second guess market prices in this manner and to these ends. I’m sorry, but it is so stupid that I cannot be charitable here. Reasonable minds cannot disagree. To say this is child-like is an insult to children who themselves wouldn’t often make such mistakes.
More forgivable yet still a somewhat elementary mistake is what David Friedman points out in criticizing CNBC’s quality of life rankings for states and the actual behavior of people moving to the low-rated states. He writes,
The CNBC authors were confusing characteristics of a state with characteristics of its population, implicitly assuming that the reason A’s residents were rich was that they lived in A instead of that the reason they lived in A was that they were rich, and similarly for the residents of B. Of the ten states that CNBC rated with the lowest quality of life, five were, measured by average income, among the ten poorest states.
There is actually a lot going in in CNBC’s confusion. Part of it is overlaying particular biases (CNBC’s preference bundle) with that of others. In economics this is a confusion of normative versus positive analysis. In layman’s terms the evaluation of “what should be” (normative) versus “what is” positive.
Additionally, there is a confusion of income effects and substitution effects. It is not or should not be a mystery as to why the entire U.S. steakhouse segment has total revenues less than half of what McDonald’s (a single fast food firm) generates in the U.S.—about $20B versus about $55B. Don’t people know those steakhouses are higher quality than MickeyD’s?1
A more subtle, nuanced confusion is discussed by Arnold Kling in his post comparing the Web of the 1990s and AI today. The implicit bad comparison is how people make one a strong analog for the other. He writes,
Until now, human progress has come from our collective brain. Individually, nobody knows how to make a pencil. But when we can cooperate, whether in small groups, large organizations, markets, or computer networks, we can expand knowledge in all sorts of directions.
AI as it operates today truncates this learning process. Claude and I can have a chat in which I propose a theory of the Great Depression that is outside of the consensus that Claude usually works with. And during the conversation Claude may concede the merits of my approach. But when it leaves the conversation, Claude will have learned nothing from me. When the next person asks it about the Great Depression, it will revert to its previous stance.
So the AI labs are trying to do away with the what Joseph Henrich called The Secret of our Success. Implicitly, they are assuming that we no longer need to share ideas with one another and to contest ideas with one another. They just have to throw more data and computer resources at problems.
I find this distinction insightful. AI has a lot of promise and makes greater promises still. Those hopes are largely around improvements to personal (individual or firm) efficiencies. There is a natural siloing at play here. Part of it is a disconnection from social exchange (will we someday long for the “toxicity” of today’s social media?), and part of it is a disconnection in additions to social knowledge (the AIs may learn an accumulation, but one would have to know what to ask them to extract what was added by another user).
Substacks mentioned:
@
@
Yes, astute reader, this example contains both normative-positive confusion as well as the income-substitution effect confusion.