The Case of Richard Glossip

Your questions (and link) are good, but I think you might be misinterpreting Paul's logic. He hasn't concluded that Glossip is definitely innocent (although he thinks he probably is), rather he's concluded that in the absence of physical evidence and with the majority of testimony being provided by a witness who has a clear motive to implicate someone else, that there must be "reasonable doubt".

He may be wrong, and in particular "reasonable doubt" may not have the legal definition that he thinks it does (legal terms rarely match common sense). But his argument is not "Glossip is innocent", but "there is no way we have sufficient evidence to kill this man without risking that we are killing someone who might be innocent".

Have you served on a jury, by the way? I served as the foreman on one relatively minor criminal case, and it did more to shake my faith in the rule-of-law than any other (minimal) encounters I've had with the legal system. In the end, we reached a unanimous "not guilty" in the case.

This was a difficult decision for all of us to reach, since the only way this could be the "correct" decision was if both police officers who testified were lying about significant details both in their report and on the stand. And yet this is the conclusion we eventually all agreed upon as being the most likely explanation for the way the pieces failed to fit together.

The defendant may well have done other illegal things, and possibly the world would be a better place if he was behind bars, but as a jury we decided that there was sufficient reason to doubt that he was guilty as charged. And yet there was no particular guarantee that this case would turn out as it did. On a different day, with a different jury, a different judge, or different lawyers, (but the same defendant, witnesses, and evidence) a guilty verdict seems likely.

As a limited participant in the legal system (but a "logical" computer programmer) my assessment would be that if we have the death penalty, the American jury system is practically guaranteed to occasionally sentence innocent people to death. I'd guess that this is the part that bothers Paul as well, I'd guess that he probably does have experience that leads him to believe this, and I'd guess that his opinion on the matter is not likely to be changed by a counterpoint that makes Glossip's guilt seem somewhat more likely.

This is a pretty reasonable suggestion. The jury that sentences a convict to death has to actually carry out the sentence. I suspect that the number of strong supporters of the death penalty will drop dramatically when faced with having to perform the sentence with their own hands.

Your comment intrigued me and I thought it might be fairly easy to estimate this number, but it seems that justifiable homicide numbers are under reported in the US: http://blogs.wsj.com/numbers/why-the-data-on-justifiable-hom...

It is interesting to look at the graph at the top of the article, though. Why do justifiable homicides for police and private citizens track each other almost identically? I can't think of any explanation.

According to: http://www.theguardian.com/us-news/2015/jul/01/us-police-kil... the number of police homicides have also been under reported, historically. It appears (if the Guardian's reporting is correct) that some states don't even report at all some years.

But let's assume that there are 15,000 homicides per year by non-police and that 1000 of them are deemed justifiable homicides (double the reported ones to account for under reporting). That's about 7% of homicides that are justifiable.

In the second link, it says that 19 out of 547 were killed by police after they were taken into custody. 31 were killed by tasers. 119 were killed while unarmed. What percentage of these killings were justifiable?

http://www.wsj.com/articles/police-rarely-criminally-charged... claims 41 police officers were charged for shootings in a 7 year period. That is about 5.86 per year. The original link claimed about 87% of killings were shootings, so let's assume that police are charged with killing linearly and we get about 7 officers charged per year. I think it is telling that I can't find the success rate of prosecution anywhere :-P

So if we assume all 7 officers were found guilty, that means we get 540 out of 547 killings to be justified, or 98.7%.

So the average citizen has an upper bound of somewhere around 7% of having a homicide found to be justified, while the police have a lower bound of around 98% of having a homicide to be found justified. This despite the fact that 3.5% of police homicides happen after the suspect is in custody, 5.7% happen as a result of the suspect being hit with a taser, and 21.8% of the suspects are unarmed.

While I wouldn't say this is a case for "blanket immunity", I have to say that the numbers do not seem reasonable to me. Especially if 20 people are killed by police while in custody and there is not an expectation that a charge will be laid in any of the 20 cases, that does not seem reasonable at all given that at least 93% of civilian homicides are deemed not-justifiable.

But even more worrying is the sorry state of affairs for the reporting of this data (if the newspapers I browsed while doing my back-of-the-envelope calculations are correct). If they don't have better reporting, I don't know how anyone is supposed to scrutinize it.

Anyway, I hope someone found this interesting. Apologies if I made any math mistakes along the way.

The bigger issue is that these things are hard to test and noone is willing to run the tests. In theory the US has a good setup for a state to decide to do some testing but politics make it impossible (and there's a decent ethical argument for either side that is far from simple). I mean think about the reactions if some politician would say "we need to see if our courts are as good as they can be, I propose we get rid of the jury system and replace it with X to see if X is superior"

1) That's UNAMERICAN. Courts and jury is the way god intended it to be.

2) Lol crackpot.

3) What if X turns out to be worse, sucks for the people who got trials like this during that time.

4) Lol crackpot.

Experimentation for political systems, judiciary infrastructure and the like is a really hard problem and I think de facto impossible. I applaud everyone who tries no matter how crazy it seems (seasteading) but at the end of the day I think there is no such thing as radical political entrepreneurship. I think it's unfortunate but I also can't envision how (rapid) experimentation cycles could be transfered to such systems.