Trayvon Martin Case

[JoeRedskin]

Originally Posted by HailGreen28

How about following IBM Watson's example and have a computer reference a database of legal precedents?

Which is, of course, fine in its limited capabilities - but that is merely data retrieval. The trick, of course, is discerning which "legal precedent", from the many similar but distinct past fact patterns applying the same law, is applicable to the entirely new fact pattern presented by your case. Essential to that analysis is determining which particular prior distinct fact pattern - or combination of them - should be chosen to be applicable to you. (Why will one legal precedent create a more just result for you than another equally similar but distinct fact pattern)?

No one is saying that computers cannot aid in many respects and in some of the most cut and dried cases of "black letter law" matters do much of the work (for example - initial determinations on whether a car went through a red light). The problem, of course, is that the vast majority of cases adjudicated by the judicial system are not such cut and dry matters.

EDIT: By the way, you realize the "database of legal precedents" is in existence. I can access, through Westlaw and Lexus, every legal treatise published in the US since the 1700's and every case, statute and regulation from every jurisdiction since that time (excluding local regs and ordinances - but they exist in separate databases that could easily be linked for access by the Holy Algorithm). Further, they aren't just stored but linked through hyperlinks so that the evolution of the various concepts can be traced through prior cases. They are also categorized by paragraph as to the distinct legal concepts to which they are relevant.

The job of jurists and jurors everywhere is to determine what of that intertwined information is relevant to and how it is applicable to the distinct and specific set of facts each new matter presents.

[HailGreen28]

Originally Posted by JoeRedskin

Which is, of course, fine in its limited capabilities - but that is merely data retrieval. The trick, of course, is discerning which "legal precedent", from the many similar but distinct past fact patterns applying the same law, is applicable to the entirely new fact pattern presented by your case. Essential to that analysis is determining which particular prior distinct fact pattern - or combination of them - should be chosen to be applicable to you. (Why will one legal precedent create a more just result for you than another equally similar but distinct fact pattern)?

Couldn't the details of each case be listed as part of the database? I think the fact patterns would basically write themselves in searches of the database, if the details of the cases were given relevance based on what was proven in court.

Just thought of a point against "robo-judge". These fact patterns couldn't be evidence in themselves of guilt. Just because say a person in Arkansas in a city that is arrested for larceny with a prior charge but no conviction while robbing a department store at 3pm during the summer is 90% likely to have committed the crime. But maybe you could use the database to sort what factors are used to prove innocence or guilt? That witness statements are the deciding factor, or employees accounts? And determine guilt or innocence based on what evidence convicts people in previous cases. So you input the details of the case, and get a verdict based on those detail. Could we assign a weight or percent to evidence based on credibility with a computer like we do as people?

Originally Posted by JoeRedskin

No one is saying that computers cannot aid in many respects and in some of the most cut and dried cases of "black letter law" matters do much of the work (for example - initial determinations on whether a car went through a red light). The problem, of course, is that the vast majority of cases adjudicated by the judicial system are not such cut and dry matters.

EDIT: By the way, you realize the "database of legal precedents" is in existence. I can access, through Westlaw and Lexus, every legal treatise published in the US since the 1700's and every case, statute and regulation from every jurisdiction since that time (excluding local regs and ordinances - but they exist in separate databases that could easily be linked for access by the Holy Algorithm). Further, they aren't just stored but linked through hyperlinks so that the evolution of the various concepts can be traced through prior cases. They are also categorized by paragraph as to the distinct legal concepts to which they are relevant.

No, I didn't know that organizations like Westlaw and LexisNexis existed. Neat and makes sense. Sounds like these works should be certified (maybe already are?) and required use for all jurists. Not to follow whatever patterns these organizations would recommend based on just a search, but every attorney/judge required to know how to use and have an active account with a certified database. Even if it is kinda redundant to what people have to know now.

Originally Posted by JoeRedskin

The job of jurists and jurors everywhere is to determine what of that intertwined information is relevant to and how it is applicable to the distinct and specific set of facts each new matter presents.

I'm thinking it may be possible to write algorithms to do that by computer. Starting with the sentence for a conviction being "x" based on precendent, and working on a program to determine guilt based not just on precedent but weighing the appropriate facts. Would still need proceedings and qualified officers of the court to get and weigh the evidence, and I'm sure there's a lot I'm missing. But how does this sound so far?