Hacker News new | past | comments | ask | show | jobs | submit login

> I think people usually go to prison for the legally correct reason

Or not, if most of the incriminating information was eyewitness testimony, the #1 cause of wrongful convictions:

https://www.innocenceproject.org/

MY opinion of the US criminal justice system is that https://en.wikipedia.org/wiki/Blackstone%27s_formulation is almost definitely not being followed, with the bias far into the false-positive conviction camp.

Come to think of it, why is the criminal justice system not "tested"? In other words, set up a mock situation where you know in advance who is innocent and who is not (because it's a setup), and run a mock trial by all the rules using actual jurors, and compare the outcome of the trial to the actual innocence or guilt of the actor?




The vast majority of convictions arise from plea bargaining.

Prosecutors and law enforcement stack charges so that it appears to be a 'good deal' to just accept a plea deal rather than risk trial and spending decades behind bars on the slight chance you will be falsely convicted.

IMO this is the single biggest problem with our system.

Plea bargaining and prosecutorial discretion makes a mockery of due process.

Overloaded public defenders and the lack of competent and affordable defense attorneys shifts most of this problem to the lower income brackets.


There is not a lot wrong with plea bargaining if an innocent suspect can except a fair trail, it trades a more lenient sentence for criminals for a more efficient, speedier and cheaper system.

The problem in the US is that poor innocent suspects have heavily stacked odds in a trail: multiple biases in judges and juries and no good lawyer with enough time for their defense. So an innocent must take the plea bargain because trail isn't a reasonable option.


So, there's a lot wrong about plea bargaining.


No...I think the parent was arguing that the problems are around plea-bargaining, not plea-bargaining itself. I personally think the biggest problem is the inbalance between prosecutor's ability to threatening absurdly large sentences. That distorts the expected value of the risk analysis in the favor of prosecution. But if the worst-case penalties were more reasonable, then plea-bargaing could be a reasonable way to reduce risk.


> Come to think of it, why is the criminal justice system not "tested"? In other words, set up a mock situation where you know in advance who is innocent and who is not (because it's a setup), and run a mock trial by all the rules using actual jurors, and compare the outcome of the trial to the actual innocence or guilt of the actor?

Probably for the same reason most things are or aren't done - some form of status-quo bias. I mean, someone has to think of the idea, push it, administer it, etc. It's a lot of work. And no one really has an incentive to do that work right now.

Btw, your statement isn't really accurate: >> I think people usually go to prison for the legally correct reason >Or not, if most of the incriminating information was eyewitness testimony, the #1 cause of wrongful convictions:

The fact that eyewitness testimony is the #1 cause of wrongful convictions doesn't tell us anything about the rate at which wrong convictions occur. It's possible for eyewitness testimony to be the #1 cause of wrongful convictions, but for there to be only e.g. 3 wrong convictions a year out of 100k trials. Which would probably be pretty "acceptable".

(I say this because it sounded like you were making the common probability mistake of flipping two things: knowledge of how often a wrongful conviction is caused by eyewitness testimony, and knowledge of how often eyewitness testimony causes a wrongful conviction. Knowing the former doesn't mean we know the latter. Not sure if you're actually claiming this, but wanted to clarify what it sounded like to me).


>It's possible for eyewitness testimony to be the #1 cause of wrongful convictions, but for there to be only e.g. 3 wrong convictions a year out of 100k trials. Which would probably be pretty "acceptable".

"It is better 100 guilty Persons should escape than that one innocent Person should suffer." Three innocent people being punished is a travesty and in no way acceptable.


https://en.wikipedia.org/wiki/Blackstone%27s_formulation actually says that it is better that 10, not 100, guilty go free.

And sadly, in a world of imperfect information, it is arguable that 3 wrongly convicted innocents out of 100,000 trials is "acceptable," because you're never going to get to 100% precision/accuracy.


Benjamin Franklin's paraphrase of the principle says 100, which is my direct quote ;D

I don't disagree that it's unrealistic to achieve 100% accuracy, but it should never be accepted - we should never, ever stop striving for improvement in this arena, and to err on the side of letting people go free if it can't be conclusively and impartially proven they are guilty.


Let's take this to an extreme. Let's get rid of incarceration and legal consequence entirely. Then you're to never wrongly convict someone, right? All other ways could and likely will end up with false positives.


Also, just to answer your actual hypothetical of "testing" the criminal justice system:

I'm not sure how you would test it. You can run a "mock" trial, although I'm not sure it's so easy - to be realistic, I think you need to create a scenario which can plausibly go in either direction (innocent or guilty). But what would that tell you? I mean, "beyond a reasonable doubt" is a pretty high standard and all, but if you make it seem pretty clear that someone is guilty when they're "really" innocent, that doesn't mean much, because this is fiction - you made up a tricky case, and people got it wrong. So what? The question is how often does real life cause a tricky scenario to manifest.

You can go the other way and set up a situation that's pretty clear, but I'm not sure that will be very insightful.

That's not to say we can't think of some way to test the system which does work - I'm sure smarter and more knowledgeable people than me can think of something. But that brings us to problem number 2 - now what? Let's say people got it wrong. What do you do? It's not like you can just change the way trials work - you need laws for that.

The most you can reasonably do is use trial tests as a basis for potentially changing the law, but that's before any politics gets involved. Realistically, I'm not sure you can do much more than gets done now - once in a while an article gets written about a specific bad practice, it causes a public stir, and the law maybe gets changed. I'm not sure systematizing any of the testing steps really changes anything else in this picture.

And we might as well ask about easier things, why don't we test them. I.e. why not test economic policies by building "test cities" which have different economic policies, and seeing which ones work? The States kind-of does this with different laws in different States. So there is kind of a natural experiment of the effect of different laws on economics, on justice, etc. This does affect some things, but I'm not sure the effect is very large (but then again, I'm not an American, so I'm not that up to date on the politics there. It's certainly an idea that a lot of lip-service is paid to, but maybe it's more than that).


> I.e. why not test economic policies by building "test cities" which have different economic policies, and seeing which ones work?

I would be exactly behind this. See: the old definition of https://en.wikipedia.org/wiki/Technocracy


Part of the point of a jury is to have multiple disinterested individuals ruling on the defendant's guilt. That in itself forms a partial test of the process.

You could run a trial with multiple independent juries, or record a trial and play it back to a set of juries and judges to see if they come up with the same verdict and sentence. This would help determine whether a given jury has a particularly forceful and prejudiced member, or if a hanging judge is handing out unfair sentences.

I'd be surprised if that sort of thing doesn't already happen occasionally. However, all this does is indicate where a problem might be, rather than validate a potential solution.

A/B testing of potential enhancements would be very difficult and limited. There are too many confounding factors. Judge and jury would have to be different each time. Witness testimony may seem unconvincingly rehearsed, or more convincingly thorough, the nth time they give it. Defendants will be able to avoid incriminating traps the prosecuting barrister set in previous iterations of the trial.


Given the intese scrutiny the litigating side give to jury selection, I'd say that it is well understood that trials may go this way or that based on a particular jury and regardless of the facts


"3 wrong convictions a year out of 100k trials. Which would probably be pretty "acceptable"."

Not if you're one of the 3.


Another test which I think has been done on rare occasions is measuring the inter-rater reliability between different juries (which could be done by having several juries hear the same trial and then deliberating independently). (It's harder to do this with judges if the judges are also overseeing the trial and ruling on evidentiary and procedural matters.)


Wrongful convictions are a serious issue, but at the same time, I don't think that we could say, even if it were solved tomorrow, that everything else about the correctional system is fine.


> Come to think of it, why is the criminal justice system not "tested"?

I can think of a few.

1. That costs time and money. Even if it would help, I see it being very difficult to justify, essentially, "secret shoppers" for the court system. I think it also kind of underestimates how much time, money, and effort goes into trials. It's not like a quick batch of unit tests. It involves lawyers, judges, jury selection (if it's a jury trial), etc. And juries themselves are made up of people who have to take time off work, miss important events, etc. You can't just (somehow while avoiding detection) fake a trial, and then go "surprise, this was just a test!" At the very least it would just piss a lot of people off.

That's not to say that there's no way to make sure our legal system is functioning properly, it's just I don't think mock trials are the right approach. Currently, I believe, the rest of the court system is relied upon to police itself, with the help of the legislative branch. Different levels of courts, ostensibly, keep each other in check (e.g. if you disagree with a lower court ruling, you can appeal to a higher court, which then makes a decision, and if you don't like that, you can appeal to the Supreme Court, etc.) Congress also has the power to do things like create federal judgeships, so presumably they have other powers to regulate and manage the federal judiciary (state judiciaries are, I imagine, left to the individual states to oversee.)

2. Cynically, there are certain vested interests that want the legal system to function the way it does (e.g. people who own private prisons and people who profit off of minor drug offenders being put away for life). There's also good old fashioned racism and classism, helping keep the overwhelmingly poor, black prison population stuck where they are.

3. I don't think it would actually tell you much; there are too many variables: different lawyers, different judges, different jurors, different facts, etc. Even if you ran the same trial with the exact same people and circumstances, but on a different day or something, you could get wildly different results. The court system is not deterministic, not by a long shot.

The court system also doesn't deal in absolutes like that. It's not "is this person definitely guilty of a crime?", it's "are we sure, beyond a reasonable doubt, that this person committed a crime (based on all of the evidence)?" Because it's impossible to know the entire truth. Sometimes it's impossible to know much of anything at all, at least with certainty. Did OJ commit those murders? We'll never /know/ know in the sense that when I flip a coin I can see that it lands on heads, or in the sense that I know the earth is a spheroid (i.e. it's something that happened in the past and we're trying to determine exactly the series of events given limited information).


#1 is not important. Everything costs money, including appeals and cleanups when long-term corruption is uncovered. Adding 1% or 0.1% for integrity testing is trivially justifiable.

#2 and #3 are saying the same thing: The powers that be know the system is broken, and don't want to fix it, and therefore want to suppress evidence that shows the brokenness.


#1 is absolutely important. You can't just take valuable time and money from people to do "testing" that, in my opinion, wouldn't tell you anything anyway. It's insulting to all of the professionals involved, especially when they have a backlog of /real/ cases to get to.

But let's entertain the idea and say that you get the trial to happen, and your innocent person was found guilty, or vice versa. What do you do with this information? It doesn't help you because you know the outcome already, but in real trials, you don't know, a priori, exactly what happened (unless you were a witness to something, but even then). So a court messed up. What do you do? Real cases don't operate on some outside observer knowing, in advance, what's going to happen. It's why trials are, at best, a good faith attempt at figuring out what happened and dispensing justice appropriately. Everyone just tries, to the best of their ability (ostensibly; I'm ignoring things like biases and such), to piece together the truth, and render an appropriate judgement.


> 1. That costs time and money.

False convictions have a gargantuan cost in time, money, and human suffering.

Justice is incredibly important. We need to get it right.


I think almost everyone agrees with your generic statement above. I'm not sure this means anyone will agree on where the actual line is.

(Not that I think that money is the reason this doesn't get done).


> #1 is absolutely important. You can't just take valuable time and money from people to do "testing" that, in my opinion, wouldn't tell you anything anyway.

If you used this argument against something like TDD (in which a 15% extra upfront cost is more than paid for by a 60-90% reduction in produced bugs, according to the Nagappan paper), it would fall flat.

If you do it enough times you will get a picture of precision and recall, which IMHO would be an important determinant of effectiveness. See: https://en.wikipedia.org/wiki/Precision_and_recall


You want to do double blind, randomized, strictly controlled testing?


> elipsoid earth theory

Do you actually know this is true? Have you tested this?


You don’t have to bite the donut to know it’s sweet.


I agree. That being said, the reason I tried to choose my words carefully is that, in this instance, I think your point mainly functions as a distraction. I agree that things like witness testimony are sub-par as numerous studies have shown, but that's also an issue wholly separate from the point I was trying to make and I felt like mentioning it would be more of a distraction than a boon with regard to highlighting that there is a difference between being guilty and deserving to be imprisoned and purposely making life needlessly hellish for the inmates.




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: