Why does this sound backwards to me? If a position is hard to fill, wouldn't you want to pay the person more so that they don't leave?
Conversely, if a position is easy to fill, wouldn't the university pay less, since if the person leaves they'll be easy to replace?
How does paying someone less because their position is hard to fill make any kind of sense? What am I missing here?
I don't know if this strategy even has to pay off in actuality, in the long term, but that actual performance of the strategy isn't as important as maintaining that model of control of labor costs.
I don't think you're missing anything. Free markets (or supply and demand) are perfectly symmetric. That's why they cannot enforce any kind of morality, like meritocracy, whatever it is. Only people can do that and it will always have a cost in the completely free market.
I think it's fairly easy for these models to fail at bringing in experts because they put extreme downward pressure on the gains from expertise and aggressively enforce regression to the mean. If someone makes twice as much as someone else you have to justify it not just in terms of market demand and a personal outlook on the skillset but also in terms of the morale effects of paying someone else twice as much openly. I'm not sure that's good for many parties and I'm quite certain it's bad for at least some. I think the likely effect is fewer people in security and fewer people in AI at the companies that do these things.