8 ms·

How Will Technology Change Criminal Justice?

16 points by bootload 11y ago

rayiner 11y ago
I'm much more interested in how science can change criminal justice. For too long, the scientific community has been content to leave the justice system to its own devices. Letting it carry on as a scientific backwater, like nutrition or social science. Consequently, police have technology without any idea of how to use it, or more importantly how to understand its limitations. For example, forensic experts are routinely allowed to testify about random match probabilities as if they are equivalent to the probability of guilt, even though in practice lab error dominates the random match probability as a source of error. This is an area where technologists can succeed where their colleagues in the physical sciences have failed. It is easy to pass off quackery as reliable forensic methods. It would be a disaster if, e.g., it was as easy to pass off quackery as legitimate data forensics. The only way that can be prevented is if the legitimate practitioners, the institutions that represent the mainstream of the profession, are vocal in calling out the charlatans that use unreliable methods.
- Xcelerate 11y ago
  I agree. In the spirit of that other article posted a few days ago ("It's better to let n guilty people go free than to imprison one innocent person"), I'm of the opinion that the field of criminal justice needs a huge overhaul in terms of rigor. We need professional statisticians working on cases that involve guilt and innocence. We need to provide accurate probabilities to jurors. We need replication and peer review of lab results. As a society, we owe a fair and accurate trial to everyone, and if we have the technical resources to improve upon the existing process, then I think we have a moral obligation to do so.
  - thaumasiotes 11y ago
    > the field of criminal justice needs a huge overhaul in terms of rigor. We need professional statisticians working on cases that involve guilt and innocence. We need to provide accurate probabilities to jurors. We need replication and peer review of lab results. This misjudges the fundamental operation of the justice system. Talk about probabilities is there to sound good, not because the justice system wants to use statistical calculations. http://volokh.com/2014/01/02/wrongful-convictions-proof-beyond-reasonable-doubt/ http://volokh.com/2014/01/02/wrongful-convictions-proof-beyo... > So here we had a case where there were no other witnesses, no “corroborating physical evidence,” a “word against word situation,” yet the judge concluded that Montgomery was guilty beyond a reasonable doubt because the judge saw “no motive whatsoever” for Coast to lie (though it turns out that Coast indeed had such a motive). > My sense is that in such cases convictions are fairly common (though not universal), especially when there’s no obvious motive for the witness to lie, but sometimes even when there is. This wasn't just any word-against-word situation. This was a girl retroactively accusing the neighbor of raping her seven years in the past, when she was 10. No evidence was ever offered (nor could it have been, really); no support of any kind was ever required for her to have a young man thrown in jail for several years. And my observation in passing earlier, that her accusation couldn't have been supported by any evidence even if it had been true, is significant: > What’s doing the work in many of the convictions, I suspect, is that the very ubiquity of the risk makes factfinders realize that — if we were to constantly consider this generalized risk, in the absence of more specific information — a wide range of crimes couldn’t be effectively prosecuted. That’s especially true of child molestation and rape, but it’s also true of many sorts of felons’ possession of guns, robberies, and the like. We've created a justice system that requires large numbers of punishments for crimes which could never be demonstrated under ideal conditions. It shouldn't surprise us that we then don't prosecute those crimes with statistical, or any other kind of, rigor. Consider a very typical word-against-word: a traffic cop writes you a ticket for running a red light. You show up in court and say "it was yellow". The cop says "it was red". He doesn't have to give evidence; his word is worth more than yours. Why?
    - imh 11y ago
      More than that, if you use statistical evidence you open the door to things like "poor people are more likely to commit X crime, and you are poor, so you are more likely to have committed the crime." While it's valid inference, it's a terrible setup for a justice system. I'm sure there are plenty of causal correlations between being some particular protected class (gay/black/muslim) and committing different crimes, but we exclude discrimination on those for a reason. Statistical inference and justice should not go hand in hand.
    - tptacek 11y ago
      Off the top of my head, and not that I don't have sympathy for the point you're making, but: * Because their individualized incentive to lie --- or, more likely, to retcon a false belief that the light was yellow when it wasn't --- is far less than yours. In fact, it probably does not matter to them one way or the other if you win your case; they won't lose a penny. * Because they're a third-party observer tasked with monitoring your compliance with the traffic light, whereas you drive through hundreds of lights without burning a milliseconds thought on any of them. * Because they're trained to monitor compliance with lights, and you're not. * Because they've testified repeatedly before the court and have a track record, and you don't. * Because given the cohort of people contesting tickets, you're statistically likely to be wrong, because so many people contest tickets for bogus reasons. I'm not trying to argue that the police are always right about traffic tickets (although it is the case that I have less sympathy for people cited for traffic violations than for almost any other offense [and I get cited too!], given the risks that noncompliant driving imposes on everyone else).
      - thaumasiotes 11y ago
        As I understand things, traffic tickets are generally revenue to the police, so they do care if you win your case. They'll only really notice if large numbers of people start doing this, but a principle that their unsupported word is no better than yours will certainly rise to that level. There are a lot of process reasons to give the police the legal right to overswear civilians. They basically boil down to the same problem discussed in the Volokh piece, which is that without that right they lose the ability to prosecute (I'm not claiming they are prosecutors; I want to use that word in a more general sense) most of the cases that are meant to be their job. But this isn't compatible with the much-vaunted notion of "innocent until proven guilty"; it's actually a judgment that most innocent people should discover that it's impossible to be found innocent once they're charged.
- waterlesscloud 11y ago
  I'm not so sure half of data forensics isn't already quackery.
sandworm101 11y ago
Horrible title. The OP has fallen into the trap of describing policing as the entirety of "criminal justice". Cops are only one of many branches of the criminal justice system, one that has very little authority in comparison to courts or corrections. I thought I would be reading about forensics or how technology might help corrections officers rehabilitate offenders, but instead it was about what to do with all the traffic cops if cars drive themselves ... a purely hypothetical problem atm.
csense 11y ago
The major thing that needs to change in our criminal justice system is that defendants' privacy should be better protected by the legal system. In olden days, having your court record accessible to the public wasn't too much of a problem, because in order to know about someone's criminal past, you had to walk through the courthouse and spend all day poring over dusty file folders. Which would only be done by people who really cared (especially since they'd have no way of knowing which of the thousands of courthouses in the country to look in, unless you told them where you were convicted). Now that we have powerful computers that can store and index all those records, and ubiquitous networks that can deliver the data anywhere, you can get instant information about anyone's interaction with the criminal justice system right from your desk (or your phone) with a few clicks and a few dollars on your credit card. If convicts, or even people who have only been arrested but never convicted, now have a ton of trouble getting education, employment, and housing, how are they supposed to improve themselves or re-integrate into society after their release?
hwstar 11y ago
Instead of building more prisons, geofence non-violent convicts to specific areas for the length of their sentence. Sci-Fi writer David Brin's "probationers"is one way (although somewhat extreme).

16 points by bootload 11y ago

I'm much more interested in how science can change criminal justice. For too long, the scientific community has been content to leave the justice system to its own devices. Letting it carry on as a scientific backwater, like nutrition or social science. Consequently, police have technology without any idea of how to use it, or more importantly how to understand its limitations. For example, forensic experts are routinely allowed to testify about random match probabilities as if they are equivalent to the probability of guilt, even though in practice lab error dominates the random match probability as a source of error. This is an area where technologists can succeed where their colleagues in the physical sciences have failed. It is easy to pass off quackery as reliable forensic methods. It would be a disaster if, e.g., it was as easy to pass off quackery as legitimate data forensics. The only way that can be prevented is if the legitimate practitioners, the institutions that represent the mainstream of the profession, are vocal in calling out the charlatans that use unreliable methods.

Xcelerate 11y ago
I agree. In the spirit of that other article posted a few days ago ("It's better to let n guilty people go free than to imprison one innocent person"), I'm of the opinion that the field of criminal justice needs a huge overhaul in terms of rigor. We need professional statisticians working on cases that involve guilt and innocence. We need to provide accurate probabilities to jurors. We need replication and peer review of lab results. As a society, we owe a fair and accurate trial to everyone, and if we have the technical resources to improve upon the existing process, then I think we have a moral obligation to do so.
- thaumasiotes 11y ago
  > the field of criminal justice needs a huge overhaul in terms of rigor. We need professional statisticians working on cases that involve guilt and innocence. We need to provide accurate probabilities to jurors. We need replication and peer review of lab results. This misjudges the fundamental operation of the justice system. Talk about probabilities is there to sound good, not because the justice system wants to use statistical calculations. http://volokh.com/2014/01/02/wrongful-convictions-proof-beyond-reasonable-doubt/ http://volokh.com/2014/01/02/wrongful-convictions-proof-beyo... > So here we had a case where there were no other witnesses, no “corroborating physical evidence,” a “word against word situation,” yet the judge concluded that Montgomery was guilty beyond a reasonable doubt because the judge saw “no motive whatsoever” for Coast to lie (though it turns out that Coast indeed had such a motive). > My sense is that in such cases convictions are fairly common (though not universal), especially when there’s no obvious motive for the witness to lie, but sometimes even when there is. This wasn't just any word-against-word situation. This was a girl retroactively accusing the neighbor of raping her seven years in the past, when she was 10. No evidence was ever offered (nor could it have been, really); no support of any kind was ever required for her to have a young man thrown in jail for several years. And my observation in passing earlier, that her accusation couldn't have been supported by any evidence even if it had been true, is significant: > What’s doing the work in many of the convictions, I suspect, is that the very ubiquity of the risk makes factfinders realize that — if we were to constantly consider this generalized risk, in the absence of more specific information — a wide range of crimes couldn’t be effectively prosecuted. That’s especially true of child molestation and rape, but it’s also true of many sorts of felons’ possession of guns, robberies, and the like. We've created a justice system that requires large numbers of punishments for crimes which could never be demonstrated under ideal conditions. It shouldn't surprise us that we then don't prosecute those crimes with statistical, or any other kind of, rigor. Consider a very typical word-against-word: a traffic cop writes you a ticket for running a red light. You show up in court and say "it was yellow". The cop says "it was red". He doesn't have to give evidence; his word is worth more than yours. Why?
  - imh 11y ago
    More than that, if you use statistical evidence you open the door to things like "poor people are more likely to commit X crime, and you are poor, so you are more likely to have committed the crime." While it's valid inference, it's a terrible setup for a justice system. I'm sure there are plenty of causal correlations between being some particular protected class (gay/black/muslim) and committing different crimes, but we exclude discrimination on those for a reason. Statistical inference and justice should not go hand in hand.
  - tptacek 11y ago
    Off the top of my head, and not that I don't have sympathy for the point you're making, but: * Because their individualized incentive to lie --- or, more likely, to retcon a false belief that the light was yellow when it wasn't --- is far less than yours. In fact, it probably does not matter to them one way or the other if you win your case; they won't lose a penny. * Because they're a third-party observer tasked with monitoring your compliance with the traffic light, whereas you drive through hundreds of lights without burning a milliseconds thought on any of them. * Because they're trained to monitor compliance with lights, and you're not. * Because they've testified repeatedly before the court and have a track record, and you don't. * Because given the cohort of people contesting tickets, you're statistically likely to be wrong, because so many people contest tickets for bogus reasons. I'm not trying to argue that the police are always right about traffic tickets (although it is the case that I have less sympathy for people cited for traffic violations than for almost any other offense [and I get cited too!], given the risks that noncompliant driving imposes on everyone else).
    - thaumasiotes 11y ago
      As I understand things, traffic tickets are generally revenue to the police, so they do care if you win your case. They'll only really notice if large numbers of people start doing this, but a principle that their unsupported word is no better than yours will certainly rise to that level. There are a lot of process reasons to give the police the legal right to overswear civilians. They basically boil down to the same problem discussed in the Volokh piece, which is that without that right they lose the ability to prosecute (I'm not claiming they are prosecutors; I want to use that word in a more general sense) most of the cases that are meant to be their job. But this isn't compatible with the much-vaunted notion of "innocent until proven guilty"; it's actually a judgment that most innocent people should discover that it's impossible to be found innocent once they're charged.
waterlesscloud 11y ago
I'm not so sure half of data forensics isn't already quackery.

sandworm101 11y ago

Horrible title. The OP has fallen into the trap of describing policing as the entirety of "criminal justice". Cops are only one of many branches of the criminal justice system, one that has very little authority in comparison to courts or corrections. I thought I would be reading about forensics or how technology might help corrections officers rehabilitate offenders, but instead it was about what to do with all the traffic cops if cars drive themselves ... a purely hypothetical problem atm.

csense 11y ago

The major thing that needs to change in our criminal justice system is that defendants' privacy should be better protected by the legal system. In olden days, having your court record accessible to the public wasn't too much of a problem, because in order to know about someone's criminal past, you had to walk through the courthouse and spend all day poring over dusty file folders. Which would only be done by people who really cared (especially since they'd have no way of knowing which of the thousands of courthouses in the country to look in, unless you told them where you were convicted). Now that we have powerful computers that can store and index all those records, and ubiquitous networks that can deliver the data anywhere, you can get instant information about anyone's interaction with the criminal justice system right from your desk (or your phone) with a few clicks and a few dollars on your credit card. If convicts, or even people who have only been arrested but never convicted, now have a ton of trouble getting education, employment, and housing, how are they supposed to improve themselves or re-integrate into society after their release?

hwstar 11y ago

Instead of building more prisons, geofence non-violent convicts to specific areas for the length of their sentence. Sci-Fi writer David Brin's "probationers"is one way (although somewhat extreme).