A lot of the discussion around this seems to focus on football, but that's just the easily-measurable proxy for a bad mood the authors found. Presumably an argument with a spouse, a poker game loss, a parking ticket, or a bad interaction with a coffee clerk would have the same effect. It's disheartening that justice is so capricious, but setting sentences through the legislature seems even more fraught. I don't have a good solution.
Maybe simple cases could some day be automatically handled by a computer, in which case the judgement will be completely impartial. Initially they could be trained in parallel to real judges to see if the judgements more or less match and to detect errors.
You don't want "impartial" judgements this is how martial law works, the legal system has to take circumstances into account unless you want to the days where we used to send people to prison for stealing a loaf of bread.
You can't teach computers empathy and empathy is part of our legal system for better or for worse.
My experiences have been pretty positive in that I was able to avoid jail time or anything going on my record that I would have to explain to a potential employer.
For the minor stuff I've been to court on that I handled myself, I've always gone in a suit and tie and been very respectful to the DA. Both times, he basically put me on probation and then told me to get out of his court.
The other times I've been to court and required an attorney, I took the financial hit in order to avoid having to go through the system. Again, every time I was able to get out unscathed. I had decent attorneys that were able to make deals with the DA's for reduced sentencing, longer probation, etc. Everything was usually done before I actually had to appear in court.
So shy of being incarcerated and having to go through the system (arrest, posting bail, court appearance, and having to serve jail time) I've had fairly positive experiences.
> unless you want to the days where we used to send people to prison for stealing a loaf of bread.
I found out this actually happens recently (if you have a record in Canada). A recent client of my lawyer partner had a past record ( domestic abuse and one more minor crime) and went to prison for stealing about $15-20 of food because he was hungry.
Regardless I know what you're saying. But there may still be the potential for a computer in the future which can analyze the context of the past thousands and thousands of human judge cases with machine learning algorithms. So it's not just a literal interpretation of law.
The challenge is establishing parameters such as previous record, age, grievousness of crime, etc. The goal is to provide a balance to human emotion. Maybe the solution here is providing that to judges, prosecutors, and defense attorneys through technology.
That's why I wrote "simple cases" and circumstances could be taken into account as long as the software has access to the person's criminal record, family situation, job, etc.
Empathy in the end comes from knowing about the person's background and, assuming it's possible to provide this information to a software program, it could make a decision as good, and maybe better, than a judge (if anything, that will allow the judge to spend time on more complicated cases). Also it's well known that, for instance, less attractive people get harsher sentences, and machines will be completely immune to this kind of bias.
We can't get a machine to recommend a movie accurately atm you want it to case judgement?
And there are solutions for less attractive people getting harsher sentences, I think the jury sitting with their backs to the courtroom isn't a bad idea, or not even being in the same room at all.
A human also cannot accurately recommend a movie, which means it's not as trivial a task as it seems.
Anyway, an algorithm judgement should not be blindly trusted, and there should be plenty of checks to make sure someone doesn't accidentally get NaN years of detention, but it could have its use.
Your argument seems to be that without empathy in the legal system people would go to prison for stealing a loaf of bread. That doesn't really make sense,
My argument would be that with empathy in the legal system, unlikeable people from the wrong minority with an unhappy judge would go to prison for stealing a loaf of bread.
There is so much disproportionality in the legal system that has arisen from gut feelings and a lack of objectivity - not just obvious things like racism but even just some crimes having far harsher punishments than more damaging ones because they're easier to get angry about.
Formalizing things like reduced punishments for crimes with lower damage (aka. bread is cheap) or born of necessity (aka. the thief can't afford food) seems so trivial a concern in comparison. Obviously formalizing more complex issues is also hard, but the bar to beat isn't set particularly high.
I agree there are many problems with injecting human judgment, but also many benefits. It's not all or nothing. We don't want to simply replace one set of problems with another.
How would you formalize it without it being exploitable?
And what about other things, how about a father who killed the man who raped his daughter? Or a woman who hired some one to kill her abusive partner? How would you estimate damage? it's easy to do with itemised goods but how do you estimate the damage of being abused, raped, how do you put a value on someone's life?
These are things we haven't even managed to fully grasp and defined yet, and you want us to define that to a computer?
How are you going to form a legal argument that can lead to a new precedent to a computer?
Of course what I just said is kind of comical at first glance, but in all honesty, and to absolutely agree with and add to your point, the exploits related to Ethereum are great examples of how something seemingly infallible that is set in stone can easily turn out to be an Achilles' heel, which then, just like in the case of Ethereum, requires human intervention, bringing it back full circle.
Also, to add to this, as @hackuser mentions[1], the system, even if set perfectly in stone, will still be based on somebody's biases.
>Also, to add to this, as @hackuser mentions[1], the system, even if set perfectly in stone, will still be based on somebody's biases.
Yes but these are going to be "fixed" biases, where a judge's ones are not.
A judge can be argued with, it can be presented with the case in a different manner, it can be given precedents which might be relevant for this case because of it's unique circumstances.
Overall a system that has a fixed set of "biases" is still "impartial" it's just utterly static, it would have it's penal code and some large list of if this then that's in regards to how to execute it's judgment.
So while it might have if (crime.type = theft && crime.subject = food && crime.offender = parent.single) { return show.sympathy(); } you won't be able to change that by arguing for or against it and that's what our legal system is based on.
This is why we have defense and prosecution rather than simply running the crime through some flow chart to figure out if you are guilty and if so what the punishment should be.
Laws are already exploitable, and frequently exploited. If you're wondering what to do when the computer gives erroneous results, obviously you'd still have human oversight with the ability to overrule with good cause.
> And what about other things, how about a father who killed the man who raped his daughter? Or a woman who hired some one to kill her abusive partner?
I don't get what makes these particularly hard questions. You have a motive and a crime. There will be some prior on how much a motive adjusts for a crime.
Obviously there is no "correct" answer, as it's a philosophical question. The important thing is that it's a decision made consistently and detached from the specific biases of each case.
> How would you estimate damage? it's easy to do with itemised goods but how do you estimate the damage of being abused, raped, how do you put a value on someone's life?
People already put a value on people's lives. The problem isn't that such a value is inherent to law, it's that it's applied neither consistently or reasonably.
A simple approach to this is to decide upon some general conversions - how much a life is worth, for instance - and when it comes to specific instantiations one can relatively straightforwardly make reasoned deductions about how much suffering a given instantiation of a crime (aka. a specific class of rape) causes in the general case. This would be born of largely of research into the long-term quality-of-life effects of these events, plus a heuristic measure of how much short-term suffering is inflicted.
The difference here isn't that this deduction has to be done. The difference is that this deduction won't change because you're black, because you're rich or because your judge had a leg cramp when they woke up. The difference is that rather than being based on how bad a crime sounds, there is at least some attempt to measure the impact of the damage done.
> How are you going to form a legal argument that can lead to a new precedent to a computer?
Most crimes that we should care about are uninteresting crimes. These are things like violence, property crime and drug offenses and make the vast majority of crimes. New precedents can still be handled on a case-by-case basis (albeit with more regularization to fit the prescribed bases for punishment).
You do have a good point that certain people receive unjustly harsh sentences when judicial discretion is allowed, and that's part of why the concept of mandatory minimum sentences came about. In practice, though, it has resulted in unjustly harsh sentences for everyone.
I'm pretty opposed to having computers issuing legal judgments for a variety of reasons, not the least of which being that I spend a pretty large part of my day surrounded by government-written and government-sponsored software.
But if we were going to go that route eventually, perhaps a system where a judge is provided the "impartial" judgment as additional information or where the decisions are reviewed by a judge before sentencing and could be lessened/cancelled but not increased?
We could have a different definition of "impartial", there is a difference between a judge being impartial by not allowing outside interference which is not related to the direct matter and subject of the case from influencing their judgement.
However we do take account for both the letter and the spirit of the law when making a judgment, and we take account of the circumstances and the nature of the crime, the offenders and the offended parties.
We also make arguments based on the spirit of the law that do appeal to our human nature and empathy at large and this is a very important part of our legal system.
You would not want to see a judicial system where this isn't possible, you see this usually under martial law and other authoritative regimes were all crimes and all offenders are equal to the law and receive the same and often harsh punishment.
Even tho in the US the execution of the legal system might not be perfect, or quite far from it depending on how you look at it the principles of it are pretty important, if was pretty much the first legal system where we said we will not throw people to jail for being in debt, it was one of the few legal systems that did hold itself to higher standards of justice and equality.
And you can't have that with computers, not only because you can't teach them those things but because a legal system isn't some static list, it isn't the 10 commandments.
Our legal system is a living and breeding organism and judges, arbiters and lawyers are it's organs without them debating and arguing the legal system would "die".
You won't have new arguments, you won't have new precedents for the most part the entire legal system would grind to a halt and the only thing you'll have remaining is a penal code.
And trust me you do not want to live under a legal system which is in a effect a static and impartial penal code.
> But if we were going to go that route eventually, perhaps a system where a judge is provided the "impartial" judgment as additional information or where the decisions are reviewed by a judge before sentencing and could be lessened/cancelled but not increased?
That is basically what we do in the federal system. Judges have to calculate sentences using a mechanical, point-based system called the U.S. Sentencing Guidelines. The judge has discretion to depart from the calculated sentence, but in practice the guideline sentence has a powerful anchoring effect.
And it's terrible. Turns out that its much easier to formulate a system with inhumanely long sentences when you're a committee developing an abstract point-scoring system than when you're an individual judge faced with deciding the fate of a living, breathing human in front of you.
It would be just as partial, but instead of the judge's biases it would reflect the preferences of other humans, the software developers and whoever controls them.
I think there's a great risk in misrepresenting to the public that computers are somehow impartial, and not representative and under the control of human interests. It's like saying a legal contract is impartial.
Alleviate the psychological pressure that comes from being a judge.
Provide an evaluation as to their ability to preside over a case.
You can think of plenty of other examples where therapy and counseling might be quite good, i rather not have judges secretly medicating themselves with antidepressants, and judging by statistics alone probably more than a handful of them do.
What about running these statistics (correlating decisions with day of week, hour of day, etc.) for each judge, and then just tell them: see, in the past year you acted in this not very bright way; maybe you should think about the way you do your job? And, by the way, we'll keep running these statistics on how you judge.
Being made aware of our own biases is often the most effective way to make the situation better.
Hour of day / day of week can easily be skewed by the system at large.
For example more bar fights over the weekends, more violent crime during the heat waves in the summer, more prank crimes involving young first time offenders during school breaks (e.g. spring break), more cases of spousal abuse during the holiday season etc.
Just having the numbers and saying here you are biased isn't really easy because it might not be possible to actually determine true bias from them.
A panel of judges would be one approach. Perhaps judges could be made aware of what the average penalty is for similar cases and that would give them an easier anchor for their decision.
The Danziger study assumed that the order in which the parolees appeared was random, which turned out not to be the case:
"Danziger et al. (1) concluded that meal breaks taken by Israeli parole boards influence the boards’ decisions. This conclusion depends on the order of cases being random or at least exogenous to the timing of meal breaks. . . [But] case ordering is not random ...the board tries to complete all cases from one prison before it takes a break and to start with another prison after the break. Within each session, unrepresented prisoners usually go last and are less likely to be granted parole than prisoners with attorneys...
[O]ur data indicate a success rate of 67% for prisoners with counsel and 39% for unrepresented prisoners. Excluding deferrals in the authors' data yields very similar success rates, beginning at about 75% and dropping to 42% at the end of a session. Thus, we strongly suspect that the pattern of declining success rates is a result of hearing represented prisoners first and unrepresented prisoners last."
On the practical side, the prisoners were ordered. Judges tried to break for meals after processing a whole 'batch' of prisoners from a specific prison. Within a prison block, those with lawyers went before those representing themselves, and lawyers also ordered their cases however they wanted. As you might expect, prisoners having counsel fared better than those without.
Additionally, they measured "time elapsed from a meal" in two ways: actual wall time (in hours/minutes/seconds/etc), and case order. Both measures were individually significant, but when both were put into a model, only the ordinal one remained significant. This suggests--to me, at least--that the result isn't due to a physiological process: blood sugar levels vary continuously and don't "jump" from case to case.
There was a rebuttal letter in PNAS shortly after it was published.
Edit: HobbyJogger covered most of this first. Thanks!
The solution is restorative justice, where the focus is on repairing the harm rather than punishment.
For example, sending a drug addict to prison for 10 years doesn't do anything to alleviate the harm of the crime, and in fact is harmful in and of itself, just in a different way.
The law needs to maintain a much closer balance between hurting people and helping people. While hurting criminals does have a deterrent factor, we do have to make sure that after we hurt them we build them back up so they don't reoffend. In the long run, skipping the "build them back up" step is more dangerous than skipping the punishment step. Our laws currently falsely assume that skipping the punishment step is the most dangerous thing to do.
Repair is only applicable when there is victim. The other point is that certain acts can not be meaningfully repaired/repaid within our current societal context. I.e. How do you measure the repayment or actions-required to repair a life that is taken by someone?
I would argue indentured servitude is a valid way to go, but it makes people uncomfortable because they see it as slavery. Yet, keeping someone locked up in an unproductive cage, to watch them wither away is somehow "rehabilitating" them. It's not, it's punishing them, it's revenge.
> Yet, keeping someone locked up in an unproductive cage, to watch them wither away is somehow "rehabilitating" them.
I don't think most people claim its rehabilitation. I think the "rehabilitation" angle is political word play and would bet that most people who support harsh prison sentences see it (and support it) as punishment.
There's an element of doublethink, yes. But the entire system pretends to be about rehabilitation. The language is everywhere. Even prisons are "correctional facilities". It's troublesome because it makes it difficult to have a conversation about actual rehabilitation.
Interesting question - why is the "political wordplay" even necessary?
Many Prisoners are used as a source of virtually free labor already. Unfortunately this is being done as exploitation, without regard to how it may help or harm the prisoners used.
Not just you would argue indentured servitude should be at least an option:
"Neither slavery nor involuntary servitude, except as a punishment for crime whereof the party shall have been duly convicted, shall exist within the United States, or any place subject to their jurisdiction."
> I would argue indentured servitude is a valid way to go, but it makes people uncomfortable because they see it as slavery
I don't see any discomfort presently, even though the constitution exempts prisoners from protection against slavery.The Thirteenth Amendment, Section 1: "Neither slavery nor involuntary servitude, except as a punishment for crime whereof the party shall have been duly convicted, shall exist within the United States, or any place subject to their jurisdiction."[1]
It strikes me as inherently unproductive, on a societal level, to make vengeance a policy, as it so often results in a positive feedback loop of grievances. The grievances of the punished are not recognized as legitimate, so they're frequently displaced onto others, creating more desire for punishment and eventually more victims.
Admittedly, you're expressing a common sentiment. I've never been the victim of a violent crime, so I can't speak to the experience of feeling a need for satisfactory vengeance. But it seems to me to be one of the worse aspects of human nature, which we should seek to overcome through enacting laws seeking restorative justice, or at least non-vengeance based punishment systems. Unfortunately, the letter of the law has little to say about the underlying human desire for vengeance when wronged, and instead seems to often start from the axiom that satisfaction must be given. I wish this were more frequently and publicly questioned.
Punishing such an offender does not repair the harm, the dead will not undead again and limbs will not grow back either. It's vengeance, plain and simple.
Will the law and state exacting revenge make the victims feel better as a consolation prize? Probably not[1].
I've never understood this as an adult. It's fine to make someone else suffer because they did it to you? What reasonable person feels better because they see someone else suffer? What kind of logic is that?
Have you ever been seriously injured by a criminal or have had a loved one seriously injured by a criminal?
Vengeance is a perfectly natural and reasonable expectation of someone in such a situation and an essential element of justice. If the state doesn't provide it, people will take it into their own hands.
> Have you ever been seriously injured by a criminal or have had a loved one seriously injured by a criminal?
Yes, both.
> Vengeance is a perfectly natural and reasonable expectation of someone in such a situation
Natural, obviously (everything that occurs is "natural"; there is nothing but nature.)
Reasonable, no.
> and an essential element of justice.
This is a moral axiom that is far from universally accepted, rather than any kind of statement of verifiable fact.
> If the state doesn't provide it, people will take it into their own hands.
If the state doesn't provide it, some subset of victims will take it into their own hands (and even if the state does provide any particular version of "vengeance", some subset of victims will still take additional or differently-targeted vengeance into their own hands.)
There are lots of things some percentage of people will do on their own if the state doesn't do it for them, but that fact is not, in and of itself, support for the claim that the state doing those things is essential.
> What if someone maims me, or murders a close relative, and I feel the only way to repair the harm I've suffered is to punish the offender?
Repairing private harms isn't the purpose of the criminal justice system, its the purpose of the civil justice system, where imprisonment and capital punishment aren't even options.
The criminal justice system exist to address (whether in a punitive, restorative, or preventive sense), harms to society as a whole.
> "Hurting criminals" isn't just done to deter future crimes, it's done to satisfy the needs of the victims.
Hurting wrongdoers to "satisfy the needs of the victims" doesn't do much to satisfy actual needs of the victims, but does create a society with a lot more victims and wrongdoers.
Whether those needs ought to be satisfied is a question in its own right. How do we decide which needs should be satisfied and which shouldn't be? Not only that, but these needs arise out of attachment and clinging and a mind that has not been kept steady. Based on hate and delusion, in many cases.
I personally do not think that so much importance should be given to the mental state of people affected by trauma. I mean no offense to them, but due to inadequate training in dealing with emotions (which pretty much everyone lacks) their minds are often affected and not kept steady.
I suspect that this is one of the reasons why we don't let victims decide penalties. But judges and especially juries can be affected by outrage, hatred and delusion too. They also cling.
Well, one thing we have to do for sure is inject tests into the system.
Hire actors to play defense, prosecution, etc, without the judge knowing. Change one variable, and see how different judges react. Black, white, male, female, handicapped, etc. Witnesses, victims, defendant.
At a minimum, publish scores of Judges, so the public can make informed decisions.
Justice has to be blind, or there's no fucking point.
Getting rid of: mandatory minimums; three-strikes-your-out laws; the drug war generally, the for-profit prison industrial complex generally, and as other have stated, the whole concept of retributive justice generally might go a long way towards those ends.
It doesn't even take a specific event. Don't go in front of a judge just before lunch as their hunger can influence their decisions, likely not in your favor.
Given there's research showing harshness increases over time until lunch then dropping markedly, then increasing again over the afternoon this isn't terribly surprising.
We're not likely to eliminate these factors any time soon.
We can mitigate them to some extent with awareness and training.
With research seeming to show subconscious biases remarkably difficult to remove, I sincerely doubt they can be eliminated. We haven't managed it in society and the workplace despite significant legislative effort.
> I sincerely doubt they can be eliminated. We haven't managed it in society and the workplace despite significant legislative effort.
Right. Who made the standard (for anything!) 100% success?
But society and the workplace have greatly improved, through social effort and also the resulting legislative one. Just 60 years ago, women were only allowed to be teachers and nurses; not long before that, they needed their husband's permission to get a job and were locked out of higher education for the most part. Now most college graduates are women and most law school graduates are women. 65 years ago the U.S. had segregation! Now a black man is President, the black middle-class has exploded, etc. Look also at attitudes toward race, sexual orientation, religion - look at surveys from the 50s and 60s (and later).
In fact, I think the pessimism in the above comment is completely backward. The track record is of amazing revolutions in human affairs.
We're not really disagreeing you know. That we can't eliminate shouldn't discourage us from striving to continue to improve.
Human judges are the least worst option we have, despite and because of fallible human factors.
Society has made huge progress, and is in many respects infinitely improved. Yet ethnic minorities are still subject to racism even from the Police and judicial system, women don't get paid as highly or get proportionately into senior or board roles.
I'm not suggesting we stop, or seeing it pessimistically.
I'm going to wait for more replication before I adjust my priors too much on this. It seems plausible prima facie, but I'm suspicious of their methodology. From the article (I don't have access to the working paper), it sounds like they might have been trying to test the significance of a lot of different variables, and this might just be the one that popped up.
If you have "priors" that you can adjust, that suggests you think you are an explicit Bayesian reasoner.
It's incoherent to take on that identity and then ignore new information. You should just conjure probabilities from this incident that will have an acceptably low impact on your priors.
I'm not ignoring new information, I'm just saying my priors are strong enough and the evidence is weak enough that this won't make me change my beliefs "too much."
Once had the very surreal experience while in court waiting to litigate a case to object to a higher court's jurisdiction over a case; my motion was granted, and the case was sent to the lower court.
While waiting for the above motion to be heard for over an hour, the person directly behind me was calling how the judge would respond to various matters that required the judge's opinion. Having nothing to do, since mobile device weren't allowed, I listened to the predictions made and the outcome. The predictions were wrong most of the time.
As I stated earlier, my motion to send the complaint back to the lower court was granted and it turns out that the person behind me was the judge for the lower court.
Yes, the lower court was predictably unable to predict the higher court's judgements the majority of the time.
To me, this if it holds true statistically across the legal system in a blind test of judges -- it is a major issue; normally, judges don't want to publicly rule against another judge unless it's obvious the other judge was wrong.
Strange end to my experience was that I won in the lower court and the other party failed to appeal the judgement in a timely manner; meaning my motion to dismiss the appeal was granted based on it being filed one day late by the opposing party by the higher court. Always wondered what would have happened if the higher court had heard the case and issued a judgement; lucky that never happened.
I was under the impression that you only have multiple courts review a case when the case is appealed. Wouldn't cases that are appealed be biased in favor of unpredictability by the fact that people believed they could be reversed?
Happy to address the question, but it's not clear to me exactly what the question is.
In the example I provided the lower court and court above it shared jurisdiction over the case by law. Generally, upper courts for case load reasons will not hear a case that's not an appeal. In this case, the higher court considered hearing the case based on the arguments of the other party, but I won based on my arguments that as a result of the contract between the parties the lower court was explicitly the only court that had jurisdiction over the matter.
For what it's worth, I'm not an attorney, just skilled at legal matters and litigation; as an example, once had a judge issue an order by phone for me and have their clerk draft the order too - ask any attorney if they've ever done that... :-)
That's interesting. I wasn't aware that cases in the jurisdiction of a lower court could "skip" the lower court.
And it appears I misread your earlier point. You were talking about the lower court "predicting" the higher court, and not deciding on the same cases. My point was that cases that go to multiple courts are the ones that are hard to decide, but it's moot.
Human's are rarely rational in a way that could be understood to be accepted as absolute, we are overwhelmingly emotional creatures just like every other creature in existence.
So if a young person displeases you, then all such young people will be simply marked as a threat +1 by your brain
> That may be acceptable as long as it’s one tool in many, he said, but data shouldn’t drive the entire justice system.
Data has always driven all the decisions we make. It's just that in the past, before big data, our personal computers (read: brains) did the data wrangling. By the way, we still do it. It's how humans and other living things employ profiling.
"The checks and balances of democratic governments were invented because humans themselves realized how unfit they were to govern themselves. They needed a system, yes. An industrial age machine."
I generally think a focus on restoring the person to a good citizen is a bonus: There is no point in putting someone with an anger problem in prison if you aren't going to take care of the anger problem while they are there just like it isn't likely that a prison sentence is going to help much if your main problem is that you were nearly starving or addicted to drugs.
Nonetheless, I do believe this is where education, information, and standards should come into play. We should expect training so that judges can overcome things like the result of emotions on sentencing. We should have recommendations of standards for sentencing, and public feedback on how the judge tends to lean. With numbers, a particilar judge can know they are more harsh after lunch, more lenient on women, and other such things. They can look to see what the average sentence range is for the basic situation (x-y for first offense: y-z for second, etc).
Some folks have mentioned a computer system, and I don't think this is all that bad of an idea if used wisely. More like information and suggested sentencing: First offense? Age? Family? Extenuating circumstances that might require leniency or harshness? and then get a range of recommendations with their effectiveness rates.
I do think it is prudent to have sentences seconded by a separate judge, especially if the judge is towards or over the harsher end of the punishments or too far into the leniency. I do think they might get rubber-stamped if not done correctly, but I also think this would get judges to consider outcomes.
None of this is going to get any easier until widespread recognition of the next objective of humanity as permanently ending human killing and torture gains universal or universally-implementable acceptance.
91 comments
[ 4.7 ms ] story [ 146 ms ] threadYou can't teach computers empathy and empathy is part of our legal system for better or for worse.
I don't see evidence of much empathy in the US justice system.
My experiences have been pretty positive in that I was able to avoid jail time or anything going on my record that I would have to explain to a potential employer.
For the minor stuff I've been to court on that I handled myself, I've always gone in a suit and tie and been very respectful to the DA. Both times, he basically put me on probation and then told me to get out of his court.
The other times I've been to court and required an attorney, I took the financial hit in order to avoid having to go through the system. Again, every time I was able to get out unscathed. I had decent attorneys that were able to make deals with the DA's for reduced sentencing, longer probation, etc. Everything was usually done before I actually had to appear in court.
So shy of being incarcerated and having to go through the system (arrest, posting bail, court appearance, and having to serve jail time) I've had fairly positive experiences.
You?
I found out this actually happens recently (if you have a record in Canada). A recent client of my lawyer partner had a past record ( domestic abuse and one more minor crime) and went to prison for stealing about $15-20 of food because he was hungry.
Regardless I know what you're saying. But there may still be the potential for a computer in the future which can analyze the context of the past thousands and thousands of human judge cases with machine learning algorithms. So it's not just a literal interpretation of law.
The challenge is establishing parameters such as previous record, age, grievousness of crime, etc. The goal is to provide a balance to human emotion. Maybe the solution here is providing that to judges, prosecutors, and defense attorneys through technology.
Empathy in the end comes from knowing about the person's background and, assuming it's possible to provide this information to a software program, it could make a decision as good, and maybe better, than a judge (if anything, that will allow the judge to spend time on more complicated cases). Also it's well known that, for instance, less attractive people get harsher sentences, and machines will be completely immune to this kind of bias.
Anyway, an algorithm judgement should not be blindly trusted, and there should be plenty of checks to make sure someone doesn't accidentally get NaN years of detention, but it could have its use.
Why?
Your argument seems to be that without empathy in the legal system people would go to prison for stealing a loaf of bread. That doesn't really make sense,
My argument would be that with empathy in the legal system, unlikeable people from the wrong minority with an unhappy judge would go to prison for stealing a loaf of bread.
There is so much disproportionality in the legal system that has arisen from gut feelings and a lack of objectivity - not just obvious things like racism but even just some crimes having far harsher punishments than more damaging ones because they're easier to get angry about.
Formalizing things like reduced punishments for crimes with lower damage (aka. bread is cheap) or born of necessity (aka. the thief can't afford food) seems so trivial a concern in comparison. Obviously formalizing more complex issues is also hard, but the bar to beat isn't set particularly high.
And what about other things, how about a father who killed the man who raped his daughter? Or a woman who hired some one to kill her abusive partner? How would you estimate damage? it's easy to do with itemised goods but how do you estimate the damage of being abused, raped, how do you put a value on someone's life?
These are things we haven't even managed to fully grasp and defined yet, and you want us to define that to a computer? How are you going to form a legal argument that can lead to a new precedent to a computer?
Of course what I just said is kind of comical at first glance, but in all honesty, and to absolutely agree with and add to your point, the exploits related to Ethereum are great examples of how something seemingly infallible that is set in stone can easily turn out to be an Achilles' heel, which then, just like in the case of Ethereum, requires human intervention, bringing it back full circle.
Also, to add to this, as @hackuser mentions[1], the system, even if set perfectly in stone, will still be based on somebody's biases.
1. https://news.ycombinator.com/item?id=12462653
Yes but these are going to be "fixed" biases, where a judge's ones are not.
A judge can be argued with, it can be presented with the case in a different manner, it can be given precedents which might be relevant for this case because of it's unique circumstances.
Overall a system that has a fixed set of "biases" is still "impartial" it's just utterly static, it would have it's penal code and some large list of if this then that's in regards to how to execute it's judgment.
So while it might have if (crime.type = theft && crime.subject = food && crime.offender = parent.single) { return show.sympathy(); } you won't be able to change that by arguing for or against it and that's what our legal system is based on.
This is why we have defense and prosecution rather than simply running the crime through some flow chart to figure out if you are guilty and if so what the punishment should be.
> And what about other things, how about a father who killed the man who raped his daughter? Or a woman who hired some one to kill her abusive partner?
I don't get what makes these particularly hard questions. You have a motive and a crime. There will be some prior on how much a motive adjusts for a crime.
Obviously there is no "correct" answer, as it's a philosophical question. The important thing is that it's a decision made consistently and detached from the specific biases of each case.
> How would you estimate damage? it's easy to do with itemised goods but how do you estimate the damage of being abused, raped, how do you put a value on someone's life?
People already put a value on people's lives. The problem isn't that such a value is inherent to law, it's that it's applied neither consistently or reasonably.
A simple approach to this is to decide upon some general conversions - how much a life is worth, for instance - and when it comes to specific instantiations one can relatively straightforwardly make reasoned deductions about how much suffering a given instantiation of a crime (aka. a specific class of rape) causes in the general case. This would be born of largely of research into the long-term quality-of-life effects of these events, plus a heuristic measure of how much short-term suffering is inflicted.
The difference here isn't that this deduction has to be done. The difference is that this deduction won't change because you're black, because you're rich or because your judge had a leg cramp when they woke up. The difference is that rather than being based on how bad a crime sounds, there is at least some attempt to measure the impact of the damage done.
> How are you going to form a legal argument that can lead to a new precedent to a computer?
Most crimes that we should care about are uninteresting crimes. These are things like violence, property crime and drug offenses and make the vast majority of crimes. New precedents can still be handled on a case-by-case basis (albeit with more regularization to fit the prescribed bases for punishment).
But if we were going to go that route eventually, perhaps a system where a judge is provided the "impartial" judgment as additional information or where the decisions are reviewed by a judge before sentencing and could be lessened/cancelled but not increased?
However we do take account for both the letter and the spirit of the law when making a judgment, and we take account of the circumstances and the nature of the crime, the offenders and the offended parties.
We also make arguments based on the spirit of the law that do appeal to our human nature and empathy at large and this is a very important part of our legal system.
You would not want to see a judicial system where this isn't possible, you see this usually under martial law and other authoritative regimes were all crimes and all offenders are equal to the law and receive the same and often harsh punishment.
Even tho in the US the execution of the legal system might not be perfect, or quite far from it depending on how you look at it the principles of it are pretty important, if was pretty much the first legal system where we said we will not throw people to jail for being in debt, it was one of the few legal systems that did hold itself to higher standards of justice and equality.
And you can't have that with computers, not only because you can't teach them those things but because a legal system isn't some static list, it isn't the 10 commandments.
Our legal system is a living and breeding organism and judges, arbiters and lawyers are it's organs without them debating and arguing the legal system would "die". You won't have new arguments, you won't have new precedents for the most part the entire legal system would grind to a halt and the only thing you'll have remaining is a penal code.
And trust me you do not want to live under a legal system which is in a effect a static and impartial penal code.
That is basically what we do in the federal system. Judges have to calculate sentences using a mechanical, point-based system called the U.S. Sentencing Guidelines. The judge has discretion to depart from the calculated sentence, but in practice the guideline sentence has a powerful anchoring effect.
And it's terrible. Turns out that its much easier to formulate a system with inhumanely long sentences when you're a committee developing an abstract point-scoring system than when you're an individual judge faced with deciding the fate of a living, breathing human in front of you.
It would be just as partial, but instead of the judge's biases it would reflect the preferences of other humans, the software developers and whoever controls them.
I think there's a great risk in misrepresenting to the public that computers are somehow impartial, and not representative and under the control of human interests. It's like saying a legal contract is impartial.
Having attentive clerks as well as maybe mandatory psychiatric counseling for judges can also resolve some issues.
Provide an evaluation as to their ability to preside over a case.
You can think of plenty of other examples where therapy and counseling might be quite good, i rather not have judges secretly medicating themselves with antidepressants, and judging by statistics alone probably more than a handful of them do.
Being made aware of our own biases is often the most effective way to make the situation better.
Hour of day / day of week can easily be skewed by the system at large.
For example more bar fights over the weekends, more violent crime during the heat waves in the summer, more prank crimes involving young first time offenders during school breaks (e.g. spring break), more cases of spousal abuse during the holiday season etc.
Just having the numbers and saying here you are biased isn't really easy because it might not be possible to actually determine true bias from them.
http://www.pnas.org/content/108/17/6889.full
A panel of judges would be one approach. Perhaps judges could be made aware of what the average penalty is for similar cases and that would give them an easier anchor for their decision.
"Danziger et al. (1) concluded that meal breaks taken by Israeli parole boards influence the boards’ decisions. This conclusion depends on the order of cases being random or at least exogenous to the timing of meal breaks. . . [But] case ordering is not random ...the board tries to complete all cases from one prison before it takes a break and to start with another prison after the break. Within each session, unrepresented prisoners usually go last and are less likely to be granted parole than prisoners with attorneys...
[O]ur data indicate a success rate of 67% for prisoners with counsel and 39% for unrepresented prisoners. Excluding deferrals in the authors' data yields very similar success rates, beginning at about 75% and dropping to 42% at the end of a session. Thus, we strongly suspect that the pattern of declining success rates is a result of hearing represented prisoners first and unrepresented prisoners last."
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3198355/
On the practical side, the prisoners were ordered. Judges tried to break for meals after processing a whole 'batch' of prisoners from a specific prison. Within a prison block, those with lawyers went before those representing themselves, and lawyers also ordered their cases however they wanted. As you might expect, prisoners having counsel fared better than those without.
Additionally, they measured "time elapsed from a meal" in two ways: actual wall time (in hours/minutes/seconds/etc), and case order. Both measures were individually significant, but when both were put into a model, only the ordinal one remained significant. This suggests--to me, at least--that the result isn't due to a physiological process: blood sugar levels vary continuously and don't "jump" from case to case.
There was a rebuttal letter in PNAS shortly after it was published.
Edit: HobbyJogger covered most of this first. Thanks!
For example, sending a drug addict to prison for 10 years doesn't do anything to alleviate the harm of the crime, and in fact is harmful in and of itself, just in a different way.
The law needs to maintain a much closer balance between hurting people and helping people. While hurting criminals does have a deterrent factor, we do have to make sure that after we hurt them we build them back up so they don't reoffend. In the long run, skipping the "build them back up" step is more dangerous than skipping the punishment step. Our laws currently falsely assume that skipping the punishment step is the most dangerous thing to do.
What's the repair/punishment for me smoking a joint on my front porch?
I would argue indentured servitude is a valid way to go, but it makes people uncomfortable because they see it as slavery. Yet, keeping someone locked up in an unproductive cage, to watch them wither away is somehow "rehabilitating" them. It's not, it's punishing them, it's revenge.
I don't think most people claim its rehabilitation. I think the "rehabilitation" angle is political word play and would bet that most people who support harsh prison sentences see it (and support it) as punishment.
Interesting question - why is the "political wordplay" even necessary?
It's just slavery with extra steps.
EDIT: 02x06 - The Ricks Must Be Crazy, it's a very good ep.
"Neither slavery nor involuntary servitude, except as a punishment for crime whereof the party shall have been duly convicted, shall exist within the United States, or any place subject to their jurisdiction."
I don't see any discomfort presently, even though the constitution exempts prisoners from protection against slavery.The Thirteenth Amendment, Section 1: "Neither slavery nor involuntary servitude, except as a punishment for crime whereof the party shall have been duly convicted, shall exist within the United States, or any place subject to their jurisdiction."[1]
1. Stolen from https://news.ycombinator.com/item?id=2037139
I may just become a repeat offender.
"Hurting criminals" isn't just done to deter future crimes, it's done to satisfy the needs of the victims.
Admittedly, you're expressing a common sentiment. I've never been the victim of a violent crime, so I can't speak to the experience of feeling a need for satisfactory vengeance. But it seems to me to be one of the worse aspects of human nature, which we should seek to overcome through enacting laws seeking restorative justice, or at least non-vengeance based punishment systems. Unfortunately, the letter of the law has little to say about the underlying human desire for vengeance when wronged, and instead seems to often start from the axiom that satisfaction must be given. I wish this were more frequently and publicly questioned.
FYI, this is an excellent book on the subject: "Inferno: An Anatomy of American Justice" by Robert Ferguson (http://www.hup.harvard.edu/catalog.php?isbn=9780674728684)
[1] http://www.apa.org/monitor/2009/06/revenge.aspx
Vengeance is a perfectly natural and reasonable expectation of someone in such a situation and an essential element of justice. If the state doesn't provide it, people will take it into their own hands.
Yes, both.
> Vengeance is a perfectly natural and reasonable expectation of someone in such a situation
Natural, obviously (everything that occurs is "natural"; there is nothing but nature.)
Reasonable, no.
> and an essential element of justice.
This is a moral axiom that is far from universally accepted, rather than any kind of statement of verifiable fact.
> If the state doesn't provide it, people will take it into their own hands.
If the state doesn't provide it, some subset of victims will take it into their own hands (and even if the state does provide any particular version of "vengeance", some subset of victims will still take additional or differently-targeted vengeance into their own hands.)
There are lots of things some percentage of people will do on their own if the state doesn't do it for them, but that fact is not, in and of itself, support for the claim that the state doing those things is essential.
Repairing private harms isn't the purpose of the criminal justice system, its the purpose of the civil justice system, where imprisonment and capital punishment aren't even options.
The criminal justice system exist to address (whether in a punitive, restorative, or preventive sense), harms to society as a whole.
> "Hurting criminals" isn't just done to deter future crimes, it's done to satisfy the needs of the victims.
Hurting wrongdoers to "satisfy the needs of the victims" doesn't do much to satisfy actual needs of the victims, but does create a society with a lot more victims and wrongdoers.
I personally do not think that so much importance should be given to the mental state of people affected by trauma. I mean no offense to them, but due to inadequate training in dealing with emotions (which pretty much everyone lacks) their minds are often affected and not kept steady.
I suspect that this is one of the reasons why we don't let victims decide penalties. But judges and especially juries can be affected by outrage, hatred and delusion too. They also cling.
Hire actors to play defense, prosecution, etc, without the judge knowing. Change one variable, and see how different judges react. Black, white, male, female, handicapped, etc. Witnesses, victims, defendant.
At a minimum, publish scores of Judges, so the public can make informed decisions.
Justice has to be blind, or there's no fucking point.
http://www.nature.com/news/2011/110411/full/news.2011.227.ht...
We're not likely to eliminate these factors any time soon.
I disagree. Knowing about them can make a big difference, and training can teach people to avoid these errors.
With research seeming to show subconscious biases remarkably difficult to remove, I sincerely doubt they can be eliminated. We haven't managed it in society and the workplace despite significant legislative effort.
Right. Who made the standard (for anything!) 100% success?
But society and the workplace have greatly improved, through social effort and also the resulting legislative one. Just 60 years ago, women were only allowed to be teachers and nurses; not long before that, they needed their husband's permission to get a job and were locked out of higher education for the most part. Now most college graduates are women and most law school graduates are women. 65 years ago the U.S. had segregation! Now a black man is President, the black middle-class has exploded, etc. Look also at attitudes toward race, sexual orientation, religion - look at surveys from the 50s and 60s (and later).
In fact, I think the pessimism in the above comment is completely backward. The track record is of amazing revolutions in human affairs.
Human judges are the least worst option we have, despite and because of fallible human factors.
Society has made huge progress, and is in many respects infinitely improved. Yet ethnic minorities are still subject to racism even from the Police and judicial system, women don't get paid as highly or get proportionately into senior or board roles.
I'm not suggesting we stop, or seeing it pessimistically.
https://news.ycombinator.com/item?id=12463053
Thanks.
It's incoherent to take on that identity and then ignore new information. You should just conjure probabilities from this incident that will have an acceptably low impact on your priors.
While waiting for the above motion to be heard for over an hour, the person directly behind me was calling how the judge would respond to various matters that required the judge's opinion. Having nothing to do, since mobile device weren't allowed, I listened to the predictions made and the outcome. The predictions were wrong most of the time.
As I stated earlier, my motion to send the complaint back to the lower court was granted and it turns out that the person behind me was the judge for the lower court.
Yes, the lower court was predictably unable to predict the higher court's judgements the majority of the time.
To me, this if it holds true statistically across the legal system in a blind test of judges -- it is a major issue; normally, judges don't want to publicly rule against another judge unless it's obvious the other judge was wrong.
Strange end to my experience was that I won in the lower court and the other party failed to appeal the judgement in a timely manner; meaning my motion to dismiss the appeal was granted based on it being filed one day late by the opposing party by the higher court. Always wondered what would have happened if the higher court had heard the case and issued a judgement; lucky that never happened.
In the example I provided the lower court and court above it shared jurisdiction over the case by law. Generally, upper courts for case load reasons will not hear a case that's not an appeal. In this case, the higher court considered hearing the case based on the arguments of the other party, but I won based on my arguments that as a result of the contract between the parties the lower court was explicitly the only court that had jurisdiction over the matter.
For what it's worth, I'm not an attorney, just skilled at legal matters and litigation; as an example, once had a judge issue an order by phone for me and have their clerk draft the order too - ask any attorney if they've ever done that... :-)
And it appears I misread your earlier point. You were talking about the lower court "predicting" the higher court, and not deciding on the same cases. My point was that cases that go to multiple courts are the ones that are hard to decide, but it's moot.
So if a young person displeases you, then all such young people will be simply marked as a threat +1 by your brain
Data has always driven all the decisions we make. It's just that in the past, before big data, our personal computers (read: brains) did the data wrangling. By the way, we still do it. It's how humans and other living things employ profiling.
Nonetheless, I do believe this is where education, information, and standards should come into play. We should expect training so that judges can overcome things like the result of emotions on sentencing. We should have recommendations of standards for sentencing, and public feedback on how the judge tends to lean. With numbers, a particilar judge can know they are more harsh after lunch, more lenient on women, and other such things. They can look to see what the average sentence range is for the basic situation (x-y for first offense: y-z for second, etc).
Some folks have mentioned a computer system, and I don't think this is all that bad of an idea if used wisely. More like information and suggested sentencing: First offense? Age? Family? Extenuating circumstances that might require leniency or harshness? and then get a range of recommendations with their effectiveness rates.
I do think it is prudent to have sentences seconded by a separate judge, especially if the judge is towards or over the harsher end of the punishments or too far into the leniency. I do think they might get rubber-stamped if not done correctly, but I also think this would get judges to consider outcomes.