The Framework Friction Audit
EthicsComments
The real risk here is that people will use this audit as a sophisticated way to just keep their prejudices. Why fix the logic when you can just call your bias a vital human variable?
The idea that friction always signals a missing variable is a bit optimistic. We have to account for cognitive biases like loss aversion, which can make a mathematically sound and comprehensive conclusion feel repulsive simply because humans are wired to avoid loss more than they value gain.
If we integrate these variables to remove the friction, how do we know we aren't just building a more complex map that still misses the cliff? Is there a limit to how many variables a framework can hold before it stops being logic and just becomes a list of exceptions?
This is a useful lens for human intuition, but it falls apart when applied to black-box algorithmic outputs. If the logic is a neural network with ten thousand weights, there is no clear map to audit for missing variables.
That is exactly why these systems fail in city planning. When the software says a zoning change is optimal but every resident at the town hall is screaming, that friction is the only thing stopping us from automating a neighborhood into oblivion.