This site is about error-covariance in making legal decisions but includes estimation and decision theory. This now more serious because of what will happen to the criminal justice system due to the advent of Very Large Scale Social Data Collection .
With new software it will be possible to collect massive amounts of data about every person in society, without their knowledge or permission. This will include enough information to be able to accurately predict who will commit crimes and the ability to find out who has already done so.
Like the technology itself, this comment is new. I now carry on with the rest of the original front page.
The most serious issue, error-covariance is ultimately a mathematical matter, but that fact is rarely considered. Instead many people have an intuitive understanding of it. Those people are usually frustrated by the politics which interferes with what they know to be true.
Like most of mathematics, the discussion here is largely concerned with an idealized system, but the problem of a migration path from our current ridiculous system to a defensible one is important.
The Error Covariance of Legal Tribunals
Consider the decisions of these three judges, as compared with some known facts.
In this example, a majority vote of the three judges always produces the correct verdict. As regards these five cases, the tribunal performs perfectly, not making a single mistake. Yet each individual judge made one mistake. This is the beauty of a well chosen tribunal, it can perform much better than the individuals composing it.
It is easy to see that a differeny panel of judges who also make just one mistake apiece could produce exactly the wrong result in one specific case if they are making the same mistake.
From a mathematical point of view, one can say that the tribunal shown in the table above performs well because of its low error-covariance. In an era during which judges are either elected or are political appointees, there is no reason whatsoever to suppose that they would have little error-covariance and form an effective tribunal.
That is a typical flaw in our legal systems. Basic mathematical principles are ignored in favor of politics. But we can fix this!
Corrections, Rehabilitation, Punishment and Retribution
This is an important part of making society work, quite distinct from the judicial matter of deciding whether a person is innocent or guilty. It is also distinct from the judicial question of how much punishment a person deserves or how much correction that person needs. This is a matter of estimation, discussed on the website referred to above, about decision and estimation.
Often such punishment will involve incarceration in what is usually referred to as a correctional institution. This is an example of what I describe on another website as an engineered social environment. A discussion of them will be mostly confined to that site, but insofar as their use is a part of the criminal justice system, it will be included here.