I was surprised to hear anyone claim this was an applause light. My prediction was that many people would hate this idea, and, well, at time of writing the karma score stands at −2. Sure doesn’t seem like I’m getting that much applause :)
I think the optimal number of most bad things is zero, and it’s only not zero when there’s a tradeoff at play. I think most people will agree in the abstract that there’s a tradeoff between stopping bad actors and sometimes punishing the innocent, but they may not concretely be willing to accept some particular costs in the kind of abusive situations we’re faced with at the moment. So, were I to write a post about this, it would be trying to encourage people to more seriously engage with flawed systems of abuse prevention, to judge how their flaws compare to the flaws in doing nothing.
I post about the idea here partly to get a sense of whether this unwillingness to compromise rings true for anyone else as a problem we might have in these discussions. So far, it hasn’t got a lot of traction, but maybe I’ll come back to it if I see more compelling examples in the wild.
Assuming both false-positives and false-negatives exist at meaningful rates and the former cannot be zeroed while keeping an acceptable FN rate, this seems obviously true (at least to me) and only worthy of a full post if you’re willing to ponder what the balance should be.
ETA: An edgy but theoretically interesting argument is that we should compensate the probably-guilty for the risk of error. E.g., if you are 70 percent confident the person did it, boot them but compensate them 30 percent of the damages that would be fair if they were innocent. The theory would be that a person may be expected to individually bear a brutal cost (career ruin despite innocence), but the benefit (of not allowing people who are 70 percent likely to be guilty be running around in power) accrues to the community from which the person has been booted. So compensation for risk that the person is innocent would transfer some of the cost of providing that benefit to the community. I’m not endorsing that as a policy proposal, mind you...
The Optimal Number of Innocent People’s Careers Ruined By False Allegations Is Not Zero
(haha just kidding… unless? 🥺)
Seems like a cheap applause light unless you accompany it the equivalent stories about how the optimal number of almost any bad thing is not zero.
I was surprised to hear anyone claim this was an applause light. My prediction was that many people would hate this idea, and, well, at time of writing the karma score stands at −2. Sure doesn’t seem like I’m getting that much applause :)
I think the optimal number of most bad things is zero, and it’s only not zero when there’s a tradeoff at play. I think most people will agree in the abstract that there’s a tradeoff between stopping bad actors and sometimes punishing the innocent, but they may not concretely be willing to accept some particular costs in the kind of abusive situations we’re faced with at the moment. So, were I to write a post about this, it would be trying to encourage people to more seriously engage with flawed systems of abuse prevention, to judge how their flaws compare to the flaws in doing nothing.
I post about the idea here partly to get a sense of whether this unwillingness to compromise rings true for anyone else as a problem we might have in these discussions. So far, it hasn’t got a lot of traction, but maybe I’ll come back to it if I see more compelling examples in the wild.
I am confused by the parenthetical.
Assuming both false-positives and false-negatives exist at meaningful rates and the former cannot be zeroed while keeping an acceptable FN rate, this seems obviously true (at least to me) and only worthy of a full post if you’re willing to ponder what the balance should be.
ETA: An edgy but theoretically interesting argument is that we should compensate the probably-guilty for the risk of error. E.g., if you are 70 percent confident the person did it, boot them but compensate them 30 percent of the damages that would be fair if they were innocent. The theory would be that a person may be expected to individually bear a brutal cost (career ruin despite innocence), but the benefit (of not allowing people who are 70 percent likely to be guilty be running around in power) accrues to the community from which the person has been booted. So compensation for risk that the person is innocent would transfer some of the cost of providing that benefit to the community. I’m not endorsing that as a policy proposal, mind you...