alexham said in #5516 5h ago:
What is meritocracy? Why is it cherished and self-justifying in some contexts, and an object of ridicule in others? I have been reflecting on these questions, and I think I figured it out how to think about it.
War is a verifiable domain, and provides dense and low bias feedback. By low bias I mean the platonic ‘true’ objective of the institution aligns well with what you can measure, e.g. how much progress this battalion achieved, how many casualties did they suffer vs inflict etc. While not of perfect predictive validity, these measures tend to be relatively less game-able and low bias. By dense I mean events happen frequently. Thus the meritocracy of an army at war is one that is focused on promoting leaders that show initiative and courage, buying more of the weapons that work in the battlefield and so forth. I should note that this understanding of war is applicable to attritional and territorial conflicts and not to counterinsurgencies.
Contrast this with say academia, a relatively non-verifiable domain particularly in the humanities. Attempts at introducing legibility here mean that one’s worth as an academic is measured through the number of ‘prestigious’ publications, awards and grants. But in non-verifiable domains notions of objective measurement are necessarily highly biased, particularly as shared cultural commitment to notions of truth or other fundamental principles fail to scale or be transmitted.
This then brings me to my favorite exercise: the creation of a 2x2 matrix. Suppose organizations can be either responsive or non-responsive to ‘performance’, and the measure of performance can be high vs low bias.
Government agencies for example tend to be non-responsive to ‘performance’ and as such tend not to be viewed as meritocracies. A sclerotic big company that has a near monopoly position otoh, like academia, can be highly responsive to ‘performance’, but the performance measures for hiring / promotion can be high bias. These orgs are perceived as ‘political’ and meritocracy in these contexts is oft ridiculed.
It could thus be very fruitful to explicitly think about the bias of performance measures in meritocracies. For example, in STEM academia, practices like pre-registration of hypotheses and replication challenges can partially reduce the bias of the performance measures. Furthermore, this should be a regular exercise as the bias of a performance measure is not static due to Goodhart's law.
This then is the answer I arrived at. Organizations that try to measure performance, and are highly responsive to it, are meritocracies. This is a good thing when the performance measures are low bias, and a bad thing otherwise.
War is a verifiable domain, and provides dense and low bias feedback. By low bias I mean the platonic ‘true’ objective of the institution aligns well with what you can measure, e.g. how much progress this battalion achieved, how many casualties did they suffer vs inflict etc. While not of perfect predictive validity, these measures tend to be relatively less game-able and low bias. By dense I mean events happen frequently. Thus the meritocracy of an army at war is one that is focused on promoting leaders that show initiative and courage, buying more of the weapons that work in the battlefield and so forth. I should note that this understanding of war is applicable to attritional and territorial conflicts and not to counterinsurgencies.
Contrast this with say academia, a relatively non-verifiable domain particularly in the humanities. Attempts at introducing legibility here mean that one’s worth as an academic is measured through the number of ‘prestigious’ publications, awards and grants. But in non-verifiable domains notions of objective measurement are necessarily highly biased, particularly as shared cultural commitment to notions of truth or other fundamental principles fail to scale or be transmitted.
This then brings me to my favorite exercise: the creation of a 2x2 matrix. Suppose organizations can be either responsive or non-responsive to ‘performance’, and the measure of performance can be high vs low bias.
Government agencies for example tend to be non-responsive to ‘performance’ and as such tend not to be viewed as meritocracies. A sclerotic big company that has a near monopoly position otoh, like academia, can be highly responsive to ‘performance’, but the performance measures for hiring / promotion can be high bias. These orgs are perceived as ‘political’ and meritocracy in these contexts is oft ridiculed.
It could thus be very fruitful to explicitly think about the bias of performance measures in meritocracies. For example, in STEM academia, practices like pre-registration of hypotheses and replication challenges can partially reduce the bias of the performance measures. Furthermore, this should be a regular exercise as the bias of a performance measure is not static due to Goodhart's law.
This then is the answer I arrived at. Organizations that try to measure performance, and are highly responsive to it, are meritocracies. This is a good thing when the performance measures are low bias, and a bad thing otherwise.
What is meritocracy?