4 ms·
What does “beyond a reasonable doubt” mean in probability terms? Most people say 90-95%, which suggests even Graham’s 4% number (which is on the high end of suc
by rayiner 6y ago
What does “beyond a reasonable doubt” mean in probability terms? Most people say 90-95%, which suggests even Graham’s 4% number (which is on the high end of such estimates, see: https://dc.law.utah.edu/cgi/viewcontent.cgi?article=1130&context=scholarship https://dc.law.utah.edu/cgi/viewcontent.cgi?article=1130&con...) is actually consistent with the system working as designed.
Also, while you’re thinking about that, what’s your estimate of the probability that Derek Chauvin’s negligence was the but for cause of George Floyd’s death (and not some other factor mentioned by the defense expert in the case such as his health or drugs). Note that’s a scientific question—but for causation is an element of the crime. It’s not enough that what Chauvin did was reckless as to human life. His recklessness had to be the cause. Are you 99% certain of that?
- sneak 6y agoThere is significant tradition in the criminal justice system for the error bias to be in the other direction (in theory, if not in practice): https://en.wikipedia.org/wiki/Blackstone%27s_ratio https://en.wikipedia.org/wiki/Blackstone%27s_ratio I believe that to be "working as designed" we would see a lot more guilty people going free than innocent people being convicted.
- rayiner 6y ago> I believe that to be "working as designed" we would see a lot more guilty people going free than innocent people being convicted How do you know that’s not the case?
- sneak 6y agoConviction and plea bargain rates. Most (>90%) people who are charged (in the US) end up with either a guilty verdict or a guilty plea. IIRC even the ones that don't plead out are ~90% conviction rate, at least at the federal level. Doesn't jive at all with what we know of competence and bias levels observed in police and prosecutors in the US.
- rayiner 6y agoMost people plead out because they’re ridiculously guilty. I’m a crunchy hippie on criminal justice, but I’ve also seen these cases first hand working for an appellate judge. Prosecutors go after cases that are open and shut. You wouldn’t believe how much evidence there is in a typical case that doesn’t go to trial. They have the guy on CCTV with stolen goods in his car and cell phone records of fencing. Innocence Projects weed out hundreds of meritless requests for help for the relatively few meritorious cases they take.
- sneak 6y ago> Most people plead out because they’re ridiculously guilty. I know many innocent people who plead out to avoid jail because they're looking at $250k in legal bills and a ~90% trial conviction rate. Juries, being amateurs, are not very good at avoiding bias, as they are completely untrained. I'm not so sure that plea bargain rates are indicative of guilt either way.
- rayiner 6y agoHow many people who plead do you think have enough money not to be eligible for a PD? And how do you know “many innocent people who plead out?”
- ncmncm 6y agoI am completely sure they are not indicative of anything, in any way, shape or form. Besides legal expenses and risk of being convicted anyway, being in jail tends to result in being fired, defaulting on your home and car loans, and leaving your family destitute.
- heavyset_go 6y ago> Most people plead out because they’re ridiculously guilty. Most people plead out because they can't afford expensive trials. It's also a rational choice when prosecutors promise that they'll throw the book at you if you go to trial, and it's often your word against the word of cops, prosecutors and the expensive expert witnesses they can afford to hire but you can't. Juries and judges tend to side with law enforcement because they're authority figures and valued members of the justice system, after all.
- thinkindie 6y agotry to tell the family of people in the 4% that the system is working as designed.
- benjohnson 6y agoIt’s not ideal. But if we don’t have a working criminal justice system, victims will take the law into their own hands and their error rate will most likely be much worse.
- siva7 6y agoWill they? Doesn’t seem to be the case in Europe. At least people there don’t equal death penalty with a working justice system
- SomeCallMeTim 6y agoThere is zero evidence that the death penalty is useful for deterrence. Zero points for using a distraction from the argument at hand which is whether the death penalty is justified.
- wahern 6y ago> For example, MHA found that there was a 66 per cent reduction in the average net weight trafficked for opium in the four-year window after the mandatory death penalty was introduced in 1990 for trafficking more than 1,200g of opium. > In the four years after the death penalty for trafficking more than 500g of cannabis was introduced in 1990, there was "a 15 to 19 percentage point reduction in the probability that traffickers would choose to traffic above the capital sentence threshold", said Mr Shanmugam. https://www.channelnewsasia.com/news/singapore/death-penalty-life-imprisonment-crime-deterrent-shanmugam-13204312 https://www.channelnewsasia.com/news/singapore/death-penalty... Granted, the deterrence value of capital punishment as applied in most nations, such as the U.S., has been shown to be negligible. Most countries, especially in the 21st century, just don't have the stomach to apply capital punishment strictly enough to make it worthwhile. And for good reasons, I might add. But if I were creating a country from scratch, before certain patterns of crime and social dynamics became endemic (that is, firmly rooted in the economic life of certain classes such that they would, as a group, be systematically grossly disproportionately punished), I would absolutely employ capital punishment. I think in the U.S. we tend to assume that poverty and violence go hand-in-hand. And so to punish violence is to in some sense punish the poor. But if you look at poverty around the world, violence is not a necessary consequence of poverty. There are many places where even desperately poor communities experience much less violence than some of the richest towns in the U.S. Ultimately it's a function of the wider culture and its predisposition to violence. If you can use capital punishment to forestall societal changes (e.g. drugs trafficking, use of guns in crimes) that invite pervasive violence, you can achieve a more just outcome for everybody. But in practice you probably can't use capital punishment to put the genie back in the bottle; certainly not alone.
- Aunche 6y ago> Are you 99% certain of that? Chauvin's case was unique because the narrative of him being a murderer had already been broadcast across the entire country. While I'm not personally 99% sure of his guilt to second-degree murder, the jurors most likely were.
- jjeaff 6y agoHe was convicted of 2nd degree unintentional murder. Which, in Minnesota, one of the definitions is that you killed someone while committing a felony. So even if fully an accident, if you kill someone while robbing a store, that's going to be 2nd degree murder. And in the case of this trial, I assume the jury was simply convinced that kneeling on someone's neck for nearly 10 minutes while they beg for mercy and then continuing to apply pressure for minutes after they pass out is a crime in and of itself, even if he had not died.
- YeGoblynQueenne 6y agoReasonable doubt means doubts that a person is guilty. In probability terms it can be interpreted as a state of belief in the guilt (or innocence) of the accused. PG's 4% number is offered as the frequency of wrongful convictions. The two concepts are not related. For example, if a body of jurors say that they are 95% certain that John Smith murdered Jane Brown, that does not mean that there is a 5% probability that John Smith did not commit the murder, or a 5% probability that he will be wrongfully convicted. The 95% number quantifies the belief of the jurors that John Smith is the murderer. 5% quantifies their belief that John Smith is not the murderer. Nothing in those numbers tells us anything about how often jurors are wrong- it only tells us how strongly they believe they are right. To know the frequency of wrongful convictions we must look at the frequency of exonerations. Whatever is the state of belief of jurors or judges etc during the trial where someone was convicted of a crime can't really inform us about the probability of their future exoneration.
- nojs 6y agoI think most people interpret the task of assigning a percentage to their belief of something like this as equivalent to estimating the percentage of the time they are likely to be wrong. So to say the two are not related is not true.
- solveit 6y agoI think this attributes overly precise motives to people who have none. It may be accurate for the HN crowd (although I doubt it), but in general, people assigning percentages to their beliefs are performing cargo-cult numeracy rather than actually calibrating beliefs. To illustrate: I am pretty sure that people saying they're 99% sure are saying "I am pretty sure" and not "I estimate I will be wrong once out of a hundred times in situations of comparable certainty".
- MaulingMonkey 6y ago99% is a common number for cargo-cult numeracy, much like "give me a minute" probably doesn't mean a literal minute. On the other hand, 95%, 90%, "give me three minutes", etc. are much less cargo-cult, and much less likely to be cargo-cult numeracy. On the third tentacle, people are bad at a lot of estimates, HN crowd included.
- impendia 6y ago> What does “beyond a reasonable doubt” mean in probability terms? When serving as a juror in a criminal trial, I asked precisely this question of the judge. Her response was that the law does not prescribe a percentage.
- 6gvONxR4sf7o 6y agoThis is actually a really tough problem, because it's even unclear which rate to talk about. If you look at "trial by peers" as a test, what error rates are we worried about/trying to make unreasonably rare? Sensitivity? Specificity? FDR? Posterior probability under what prior? One of the many many others used in stats? I might argue that FDR [0] is the right number, given that you want to look at the set of convicted people and say it's unreasonable to doubt that a randomly picked one of them is guilty. [0] https://en.wikipedia.org/wiki/False_discovery_rate https://en.wikipedia.org/wiki/False_discovery_rate
- jjeaff 6y agoActually, Chauvin's actions did not have to be the only cause of death. Reckless actions only had to contribute in order for him to be guilty. The fact that there were drugs involved or other medical conditions merely establish that he may have been easier to kill than a healthier individual. You only need to be very sure that what Chauvin did was reckless and unreasonable and also very sure that had Chauvin not knelt on his neck for nearly 10 minutes, that Floyd wouldn't have died at that same moment anyway.
- hnick 6y agoOften cited as the 'Eggshell skull' doctrine. https://en.wikipedia.org/wiki/Eggshell_skull https://en.wikipedia.org/wiki/Eggshell_skull
- sjg007 6y agoMaybe you should read the actual charges and Minnesota law; it covers all of your hypotheticals.
- 6gvONxR4sf7o 6y ago"Beyond reasonable doubt" means it's unreasonable to doubt it. If you bet your friend (who is incentivized to say that they rolled above a 1) rolls a ten sided die and says they rolled above a 1, is it unreasonable to doubt them? No way, 10% chance of an event happening means it's very reasonable that it'll happen. Same for a d20. At what point is it unreasonable to be concerned about rolling a 1? > Most people say 90-95% Most people believe that the 538's 2016 presidential forecast was incompetent because it said trump only had a 1/3 chance of winning and he won. Other papers put it at 10%. People said this was a huge failure because people generally are terrible at intuitively reasoning about probabilities, which is well established. I'd put reasonable doubt at closer to 1 in a few hundred, maybe even rarer. Is the link you included talking about the probability of convicting a person, given that they're innocent, or the probability that a convicted person is innocent? It brings too mind the classic stats issues of testing positive for a disease via a highly accurate test, with it still being probable that you don't have the disease. Even a highly accurate test can be entirely unreasonable as evidence of the rare case. We should probably care about making the false discovery rate the unreasonable one. Of the convicted people, is it unreasonable to doubt that a random one is guilty.