4/20/2020:Do you think someone's past predicts their future actions? Has this been true in your own life?
I do not think that someone’s past predicts their future actions (although there are of course many exceptions). Everyone is constantly being changed by their experiences and the people they surround themselves by so although someones future is often affected by their past, people change and adapt throughout their lives.
Do you think someone's actions can be predicted based on statistics from their peer group? How does this apply in your own life?
I do think that people are significantly affected by the community of people around them so in some cases statistics from peer groups can be used to predict someones action but their are definitely many outliers so I don’t think that this method of analyzing data is very accurate.
What do you think is statistical vs. individualized evidence?
I think that indivualized evidence is evidence that is collected on the actual suspect whereas statistical evidence is evidence you collect about the suspect that is based on their race, gender or which community there from but has nothing to do with the crime itself.
What do you think probable cause is? Do you think it can be quantified?
Probable cause must be something that is greater than only suspicion and I believe that it has to be backed up with a significant amount of evidence. I think that it can be quantified as it’s very difficult to draw a line between what is probable cause and what isn’t.
What percentage of people who are arrested end up going to trial? What percentage of people who are incarcerated are actually awaiting a trial (i.e. in pre-trial sentencing)?
10% of people who are arrested end of going to trial. 20% of the people who are incarcerated are awaiting a pre-trial sentencing.
What is the difference between a risk score and probability?
High risk means that there is a higher probability but it doesn’t mean there is a high probability. Probability is more about comparing someone based on everyone else in the system whereas a risk score is much more focused on that person’s alleged offense in particular. This is why the definition of high risk can differ depending on who is determining someone’s risk score.
People who are considered high risk for being rearrested, — how high risk are they actually based on the algorithms used in states like New Jersey?
People who are high-risk still have only an 8% chance of being rearrested. This is why judging predicting someones criminal history based on their risk score alone can be very difficult for someone who isn’t well acquainted with the criminal justice system.
What is the paradox between using algorithms for pre-trial sentencing vs. during trials themselves.
Logically it makes sense that you should either be able to use algorithms in both pre-trial sentencing and trials or not be able to use them at all. At the moment risk factor assessments are often used in pre-trial sentencing even though they would never be used in an actual trial.
What surprised your about this podcast episode? Did you know algorithms were being used to sentence people _before_ a trial?
I did know that some algorithms were being used in pre trial sentencing but I was not aware that many people were being given risk factor scores. It also suprised that most statistical evidence can be used in pre-trial hearings but not in an actual trial.
What is worse/better – humans or algorithms?
I do think that there is a time and place for both statistical and individualized evidence so I do think that we need to reexamine the way in which we’re using these two types of evidence in the criminal justice system.
Do you think someone's past predicts their future actions?
I do not think that someone’s past can consistently predict their future actions. I think that there are certain moments when you can bring up a persons past criminal history in court but I don’t think this can be used as a main piece of evidence.
Do you think someone's actions can be predicted based on statistics from their peer group?
I think that there are an infinite amount of factors that need to be considered before you can conclusively say that someone’s actions can be based on statistics from their peer group. It makes much more sense to use evidence that is individualized to a particular persons case instead of using statistical evidence . However if we were to reexamine the different types of information that were inputting into algorithms than it might be possible to use things such as risk factors (although personalized evidence should always be necessary).