4 ms·

An interesting thing to ask about the false positive rate: What is the false positive rate on determining that there was a false positive? That is, we may decid

by burfog 8y ago

An interesting thing to ask about the false positive rate: What is the false positive rate on determining that there was a false positive? That is, we may decide that someone was wrongly convicted, but we may be wrong in that determination. It is also interesting to note that a "wrong" conviction seldom means that the person was a fine upstanding citizen. Often it means something like "convicted of the killing, but only provided the weapon" or "convicted of the killing, but was only trying to help a fellow gang member by disposing of the body". This sort of false positive isn't anywhere near as bad as what we commonly imagine to be the case, with a 100% innocent person being wrongly convicted.

lb1lf 8y ago
-IMHO even a tiny minority of actual, true, false positives are enough to warrant getting rid of capital punishment (for that reason alone; there are other arguments, too!) To phrase my question differently - if we take your concerns about false positives into account, and accept that trying to help dispose of the body or providing a weapon makes you fair game, capital punishment-wise - how large a remaining false positive rate is acceptable to maintain that fine institution, capital punishment? 1 in 100 fine, upstanding citizens being wrongfully convicted and executed? 1 in 1,000? At which point is the collateral damage, if you like, acceptable? The matter was discussed here on HN recently, in the case of Kevin Cooper[0]. That's but one instance of what seems like a case where there's definitely reasonable doubt - to put it mildly - regarding the convict's guilt. Whether the person was a 'fine, upstanding citizen' IMHO is largely irrelevant, unless one wants to set the bar for being sentenced to death lower - like, for being an accessory to murder, for instance. [0] https://news.ycombinator.com/item?id=17098059 https://news.ycombinator.com/item?id=17098059
- burfog 8y ago
  Probability for people on trial, or total lifetime probability? It's very different. I think the total lifetime probability matters more. We consider it acceptable to drive cars. This kills 1% to 2% of the population. This seems like the baseline for acceptability. Not even 1% of the population gets put on trial with the possibility of the death penalty, so we could just go straight to execution without a trial and still be more acceptable than letting people drive cars. Or shall we ban cars? Only 1 in 119,012 die from execution. Another baseline for acceptability is drowning. Homes all across America have pools. A person is about 100x more likely to die of drowning than an execution, yet pools are still legal.
  - lb1lf 8y ago
    ...but surely it is reasonable to maintain there's a distinction between having people die from what we can call accidents or mishaps - falling into a pool, being run over by a truck, getting electrocuted or choking on a peanut &c on one side... ...and on the other being locked up for years by the system we - as a collective - have created to protect us from just that kind of arbitrary violence, before that same system murders you -ahem- for the Greater Good(tm)? The fact that the murder of an innocent is highly unlikely doesn't (IMHO) make it any more acceptable. Government is supposed to protect you; not kill you in the very textbook definition of premeditated murder. (Besides, cars and swimming pools have verifiable upsides. The 'benefits' -if any- of the death penalty are highly debatable.)

by burfog 8y ago

lb1lf 8y ago

-IMHO even a tiny minority of actual, true, false positives are enough to warrant getting rid of capital punishment (for that reason alone; there are other arguments, too!) To phrase my question differently - if we take your concerns about false positives into account, and accept that trying to help dispose of the body or providing a weapon makes you fair game, capital punishment-wise - how large a remaining false positive rate is acceptable to maintain that fine institution, capital punishment? 1 in 100 fine, upstanding citizens being wrongfully convicted and executed? 1 in 1,000? At which point is the collateral damage, if you like, acceptable? The matter was discussed here on HN recently, in the case of Kevin Cooper[0]. That's but one instance of what seems like a case where there's definitely reasonable doubt - to put it mildly - regarding the convict's guilt. Whether the person was a 'fine, upstanding citizen' IMHO is largely irrelevant, unless one wants to set the bar for being sentenced to death lower - like, for being an accessory to murder, for instance. [0] https://news.ycombinator.com/item?id=17098059 https://news.ycombinator.com/item?id=17098059

burfog 8y ago
Probability for people on trial, or total lifetime probability? It's very different. I think the total lifetime probability matters more. We consider it acceptable to drive cars. This kills 1% to 2% of the population. This seems like the baseline for acceptability. Not even 1% of the population gets put on trial with the possibility of the death penalty, so we could just go straight to execution without a trial and still be more acceptable than letting people drive cars. Or shall we ban cars? Only 1 in 119,012 die from execution. Another baseline for acceptability is drowning. Homes all across America have pools. A person is about 100x more likely to die of drowning than an execution, yet pools are still legal.
- lb1lf 8y ago
  ...but surely it is reasonable to maintain there's a distinction between having people die from what we can call accidents or mishaps - falling into a pool, being run over by a truck, getting electrocuted or choking on a peanut &c on one side... ...and on the other being locked up for years by the system we - as a collective - have created to protect us from just that kind of arbitrary violence, before that same system murders you -ahem- for the Greater Good(tm)? The fact that the murder of an innocent is highly unlikely doesn't (IMHO) make it any more acceptable. Government is supposed to protect you; not kill you in the very textbook definition of premeditated murder. (Besides, cars and swimming pools have verifiable upsides. The 'benefits' -if any- of the death penalty are highly debatable.)