And, to put a charitable gloss on it—many of your clients who protest their innocence did actually have something true-positive on them. TFA cites manufacturers claiming a 4% false positive rate in lab conditions, and protesters claiming a 15-18% FPR in field conditions [*]. Offensive and intolerable in the aggregate, yes; but as far as base odds for any specific case in front of me?

Assuming 18% of my defendants truly didn’t have drugs or drug residue on them, that still means 4 out of 5 of them did—and that the extra months of sitting in jail and fighting would lead nowhere for them. What’s 6 months of diversion by comparison?

Even beyond the odds, how many of defendants’ claims are precise enough to implicate test accuracy specifically? “No, man, I swear I don’t do that stuff anymore” might be completely true and still result in a marginal-but-confirmatory lab result for the swab of the ol’ contraband satchel or whatever.

Seems to me that’s the sort of stuff that the plea-negotiation part is for: instead of hoping expensive and slow science will give you certainty (or context), it’s “hey prosecutor, even if the test is right, we’re talking residue not bricks… diversion and call it a day?”

I feel like, from the outside (and fed a steady diet of Law and Order television show), people grossly overestimate the degree of certainty embedded in a criminal conviction. I’m reminded of the classic Kalven and Zeisel study (1967) [1] finding, among other things, that judges and juries who sat through the exact same trial disagreed on guilt or innocence in up to 25% of cases.

False convictions are outrageous, and I understand our individual instincts here to say “go to the mat for the truth!” But people are complicated, context is hazy, and there’s some irreducible residual rate at which any approach to justice gets it wrong. Which adds up to a lot more individuals than we’d like to assume.

One of many reasons not to confuse the person with the bureaucratic record of the person.

[*] Or, as the CNN person writes, a “91% error rate” per an NYCDOI study… guessing because there they were over-applying the tests in situations with extremely low base rates? At first I assumed this was a typical case of journalistic innumeracy, but reading through the study [0], they’re absolutely right and worth a quick aside: 91% really was the actual false positive rate the lab found in actual field tests. Surface swabs in a jail mail interdiction program, only examining the fentanyl tests, but still. It’s a wild read in all sorts of ways: there was a known reaction between the field test and chemicals in paper, but they used it for the mail anyway; the more false positives they saw, the more they panicked that all the mail had fentanyl in it. 89% of all the initial DOC field tests came back positive, 91% of those positives turned out to be false ones. The administration interpreted this to mean that all the inmates’ families were clearly “soaking envelopes in liquid fentanyl,” then mailing them to inmates to “chew.” The misconception led them all the way to trying to ban physical mail in the system.

[0] https://www.nyc.gov/assets/doi/reports/pdf/2024/FieldTesting...

[1] https://www.commentary.org/articles/abraham-goldstein/the-am... and replicated recently https://scholarship.law.cornell.edu/facpub/343/

I might accept a 0.1% false positive rate - but 4% is unconscionable. Yes, trials are messy, but this isn’t a trial - it’s a bad test.

Tbh, your “people are complicated” stance feels like the position of someone who believes they will never be on the wrong side of situations like these. I hope you never are, but I don’t think it’s a position that is defensible if you’re including people who are in groups likely to be subjected to tests like this (e.g. prison inmates)