3 ms·

I think this in near-inevitable in any org or federation of orgs of people with more than a few hundred people involved (wild-ass-guess, substitute your own).

I think this in near-inevitable in any org or federation of orgs of people with more than a few hundred people involved (wild-ass-guess, substitute your own). Once you have too much scale for a few people to truly know what's going on, all that's left to keep people working on the right incentives is trust, and inevitably you'll have some bad actors at some point, and after that you put in some metrics and then it gets worse cause even good actors pursuing those metrics may have gone off the rails. The only places I've seen avoid it moderately successfully have had exceptional people managers, like 1-in-100/1000 level, to minimize the appeal of playing the game instead of doing the work.

borroka 3y ago
Every system that was later changed for something better was, at one time, considered inevitable, natural, ineluctable.
Bjartr 3y ago
> too much scale for a few people to truly know what's going on Somewhat related is Dunbar's number https://en.m.wikipedia.org/wiki/Dunbar%27s_number https://en.m.wikipedia.org/wiki/Dunbar%27s_number
brnaftr361 3y ago
I think "bad actors" is a poor framing, this extends well beyond malfeasance. Even in a perfect system, at some scale people need these elementary quantities to be able to model and predict exogenous factors or justify certain other behaviors to operate. Goodhart's law does tend to come into effect - but Goodhart's law may come to dominate these systems through far more than malfeasance. Just phone game-esque decay of information from one actor to the next can cause it, moreover this occurs across generations which I would posit accelerates the decay. One "generation" may have firsthand experience of the why and how, and upon relaying it to the second a loss occurs, and the third may not ever receive it or they're given some highly heuristicised version which artlessly excises nuance. Very quickly the system is reliant on decayed information which informs garbage processes, but this is ossified as the standard operating procedure. I would posit that, especially in the temporal domain, this decay should prompt us to reconsider our tolerance for staple institutions and our willingness to facilitate their development, sustainment, and immortalization and thus limit scale while maintaining the integrity of vision and information while jointly allowing voids to open and be closed by legitimate and ostensibly better suited successors. Right now we're making models of cancer, many of which are metastatic. Nice handle, by the way.

borroka 3y ago

Every system that was later changed for something better was, at one time, considered inevitable, natural, ineluctable.

Bjartr 3y ago

> too much scale for a few people to truly know what's going on Somewhat related is Dunbar's number https://en.m.wikipedia.org/wiki/Dunbar%27s_number https://en.m.wikipedia.org/wiki/Dunbar%27s_number

brnaftr361 3y ago

I think "bad actors" is a poor framing, this extends well beyond malfeasance. Even in a perfect system, at some scale people need these elementary quantities to be able to model and predict exogenous factors or justify certain other behaviors to operate. Goodhart's law does tend to come into effect - but Goodhart's law may come to dominate these systems through far more than malfeasance. Just phone game-esque decay of information from one actor to the next can cause it, moreover this occurs across generations which I would posit accelerates the decay. One "generation" may have firsthand experience of the why and how, and upon relaying it to the second a loss occurs, and the third may not ever receive it or they're given some highly heuristicised version which artlessly excises nuance. Very quickly the system is reliant on decayed information which informs garbage processes, but this is ossified as the standard operating procedure. I would posit that, especially in the temporal domain, this decay should prompt us to reconsider our tolerance for staple institutions and our willingness to facilitate their development, sustainment, and immortalization and thus limit scale while maintaining the integrity of vision and information while jointly allowing voids to open and be closed by legitimate and ostensibly better suited successors. Right now we're making models of cancer, many of which are metastatic. Nice handle, by the way.