What you see on the screen is what I look at every day when I start up. This computer sits on my bench. I have been a trial judge for ten years in Multnomah County and have handled all sorts of civil and criminal cases.

The image in the middle is a design for the prison, which was the invention of Jeremy Bentham. Anyone with any exposure to the history of criminal justice knows that prisons are relatively modern. They came from the same time as when Lewis and Clark were coming through these parts with their Newfoundland dog. If you back out of my website address, it will direct you to my Newfoundland dog site. I have this on my screen each morning to remind me that even though I am certain that I have the answer to 90 percent of the problems you have been hearing about this morning, so was Jeremy Bentham, and he was deadly wrong about a lot of what he thought he was certain about. So this is a reminder of humility; I think all judges need that routinely, and that is why it is on my bench.

In your materials you have something called "Paying Attention to What Works;" it's a three page piece, a download from the website that I maintain in support of this process which I will soon describe to you. It was an article in the Oregon State Bar Bulletin that was published about a year ago, the August-September issue. The goldenrod issue called Resolutions News is this two page document that was just distributed. At the very bottom of the front page you will see the website that I maintain all this stuff on, and I will tell you what's there in a minute. Resolution News is a newsletter, this is the fourth edition out today. I have been publishing it primarily for Oregon judges to keep them apprized of developments in this process. I give it to you because it describes -- it's a snapshot of where we are in our progress now in the technological directions that I'll describe in a minute, and also gives a handy cite to the web address where you can look up everything else. If you remove the words "whatwrks.html" from the website, you'll get into the Columbia River Newfoundland Club web site, and it's fun too, but has absolutely nothing to do with this.

First of all, I'd like to give one disclaimer, I have something of a quirk, I don't want to call it a disability, a quirk, I can't remember names at all. I have seen a lot of you before, and I will see a lot of you again, and if I don't remember your names, it's probably because I was dropped on my head in preschool. But one of the advantages it has given to me is that when I listen to discussions about systems and their efficiencies, I think I view them differently than most people. We all have to worry about systems and how well they work, but when I start seeing prison modulation based on changing the rules for revocation so they we can keep the bed load right, it occurs to me that if we build a prison within a wide range of possible capacities, we can make the system accommodate that capacity just by focusing on the system's needs. I think that what we all do in all of our worlds, including the judge world, is we focus on the needs of the system instead of the purpose of the system. And with the deficit of not being able to stay on that groove, I can't carry a tune, I cannot continue to look at systems only for their own interests. Every once in a while, I lose track and I realize that there is a purpose. Along those lines, let me give you first this -- it is the same statistics as on that downloaded piece of materials here. This is one way of looking at the problem of recidivism, and as someone mentioned this morning, recidivism can be measured in all sorts of ways, and for reasons that I will explain later, I think it is critical to remain able to redefine what you mean by recidivism within a fairly narrow range, that has to do with criminal behavior, but for different people for different occasions, you may want to look at different measures of recidivism. I will give you an example of that.

This is one way of looking at recidivism. These are statistics published by the Portland Police Bureau every month, the most recently published statistics are from August, 2000, they are a little behind, that is August publication date, July is the month. The way to read these statistics, which I have extracted from a bigger chart which gives you more information, is the first row is the number of people jailed in Portland for anything in July of 2000. The next column is the number of those who are in jail for something in Portland within the prior 12 months. So if you look at burglars, of the 32 burglars in jail for burglary, not necessarily burglars, in jail for burglary, 22 of them had been in jail in Portland for something else within the 12 months. Of the 23 jailed for robbery, 22 had been in jail in Portland within the 12 months previous for something else. I have just confirmed that these do not include the original charge, that is to say you are not counting the first arrest on a robbery and they are arrested again. And if you look through those -- look at the drugs, 372 people jailed in July, 304 of them had been in jails in for something in Portland before that, and so on. There are indeed some first offenders who are very numerous in the system that don't come back. They tend to be first time DUII offenders; they tend to be first time male prostitution defendants. The rest of them -- and this is what occurred to me very soon after being on the bench, was that the first offender you see in the courtroom is a very rare critter indeed. When I hear an attorney say "this is my client's first offense" I immediately say "Please don't put it that way" say "it's my client's only offense," because most of the people we see are in the middle of criminal careers, and until we start looking at their careers as careers, we won't start addressing the problem. And until we do a better job of deciding what it is we are to do to these people when we have the option of doing it to them --  and as judges and in the correctional system our hit is as they come into the courtroom and get convicted -- other people have other hits -- people have release decisions to make; people have supervision decisions to make, people have decisions -- in 1996 NIJ and the Citizen's Crime Commission put out a joint session very much like this just before the legislative session. What we heard was, if you want the most return for your dollar of investment: high school completion, parenting education. If you want the best return for your money, do it there. But as a judge and a participate in the criminal justice system, my best shot at these people is when they come into my courtroom and their convicted, and I have some discretion. What these figures indicate is that we are doing a terrible job, a terrible, terrible job. They mean for example that the robberies that occurred, some of those robberies are victimizations that we could have prevented had we been smarted, but we didn't even try! We didn't even try because when we do sentencing decisions, we never ask the question, "how is your recommended sentence most likely to prevent the next victimization." When I ask that of lawyers their jaws drop, and people who are perfectly capable of holding forth on mitigation and aggravation as if those things were quantifiable are speechless when I ask that question. And they're speechless only because they don't get asked that question. I think the most important thing we can do in criminal justice is routinely and rigorously, and conscientiously to ask that question.

Why is it that we are like this?
 

This is a figure that I've pulled from the internet -- this is a doctor in the era of the Great Plague. He had robes, and he wore a mask. That mask was a pitiful precursor of contagion science, intended to prevent doctors from becoming infected with the Black Plague. They wore that.

Modern doctors dress a little differently. Here's one:

There are some judges who use computers. Here's one: They don't use that for sentencing. I'll talk about the sentencing systems that exist.

But when it comes to sentencing, they look a lot like the guy who was around in the period of the Black Plague.
 
 

And I suggest to you that one of the interesting wrinkles about the argument for judicial discretion is that most of us think we know very well what works and what doesn't work.  But you know, we don't test it, and we don't rely on knowledge. And frankly, my opinion is that we're no better at what we do without access to good information than this fellow was in his day in the Black Plague.

We have started to work on that problem. In 1997, we passed a law, this was 1997 House Bill 2229. . . . I'll take you through it quickly. The full text is on the website if you want to examine it. It did several things. Its purpose was first of all to articulate as had rarely been done before that public safety is really one of the things that you should measure criminal justice's performance by. Believe it or not, it took a constitutional amendment in 1996 to put that into our constitution. And it took this bill to put it in all the places you will see it in -- that public safety should be the measure, recidivism reduction should be the measure. This [bill] did a lot of things. One thing, it linked juvenile and adult data so that we can see the impact of the choice we make on juveniles as to their criminal careers as adults. We had an Oregon Criminal Justice Commission. It does a lot of the things that the Texas commission [subject of another speaker's talk at this session]. We added to its planning function the duty of providing "methods of assessing the effectiveness of juvenile and adult correctional programs, devices and sanctions in reducing future criminal conduct." That as an outcome measure hadn't been adequately articulated and certainly hadn't been adequately practiced in our system. The Department of State Police shall establish "the Criminal Justice Information System," CJIS, the "system" now being "standards," we changed that at their request, but CJIS is supposed to be doing something new as a result of HB 2229, and that is "[E]nsure that in developing new information systems" --  that's information technology -- "data can be retrieved to support evaluation of criminal justice planning and programs, including . . . the ability of the programs to reduce future criminal conduct." A simple concept, but it needed to be stated. The Youth Authority "reformation plan" was fiddled with in this fashion: to require that the plan be aimed at "reduc[ing] future criminal and antisocial conduct"; that the Youth Authority should cooperate and "to the extent of available information systems resources, shall share data with the Department of Corrections" to see what happens to adults after they've been dealt with in the juvenile system.

The Department of Corrections, which already had an information obligation have that obligation modified to make sure not only that it could track offenders, but that it would "[p]ermit analysis of correlations between sanctions, supervision, services and programs, and future criminal conduct." Again, a modest proposal in any other world, but it hadn't been articulated here.

I'll take you through the rest quickly and describe it to you generically. What you see next is that the Youth Authority and the Department of Corrections, under the various bills that return a lot of responsibilities to local jurisdictions in partnership with private providers and community-based providers, part of the deal has to be shared information so that we can do this kind of assessment.

In the same year, 1997, the Judicial Conference, of which I'm a member, and I now chair the Criminal Law Committee of the Conference, this is a resolution that was introduced first in the '96 Conference, we set up a committee, which is, after all, traditional to deal with things that might be controversial, the committee met and produced this resolution which was passed. It reviews the provisions of Oregon law about what the purposes of sentencing are. The second paragraph is the Article I, section 15 amendment which says that "[l]aws for the punishment of crime shall be founded on these principles: protection of society . . . ." which entered the constitution as a purpose of criminal justice, of sentencing, in 1996. The rest of the statutes are there for your review, but the "therefore clause" is: "THEREFORE, BE IT RESOLVED BY THE OREGON JUDICIAL CONFERENCE that in the course of considering the public safety component" -- there are other components of sentencing -- judges are to "invite advocates to address" and themselves to consider "the likely impact of the choices available to the judge in reducing future criminal conduct."

That's precisely the question that jaws still drop when you ask it in the courtroom. Defense attorneys and prosecutors are rarely prepared to enter the courtroom and answer or address that question.

The last resolution here says that "judges are encouraged to . . . obtain training, education and information to assist them" in making these choices.

So the policy declarations have been made. Publicly, we have learned that one of the issues we are to address in making sentencing choices -- and we have many choices left; in 90 % of the case in which I impose sentences I have tremendous discretion. Ballot Measure 11 is of course important for big cases, but the great bulk of cases coming through the courthouse are not Ballot Measure 11 cases, and there're questions about whether to use jail or not, how much jail to use, whether concurrent or consecutive sentences ought to be used, what conditions to report, whether to revoke, when to revoke, what additional conditions of probation to impose if you don't revoke -- there's tremendous discretion in the courtroom every day. What we have been doing in Multnomah County, and what we hope will be the precursor of the Public Safety Data Warehouse, which the State Police Department is constructing, is to produce a tool which is to be used by judges, advocates for and against the State, for and against the defendant, probation officers in supervision, release officers, correctional officers for people in custody -- to match offenders with the response that works for similar offenders, with "works" being defined by reduced crime.

The technology that this uses is a technology that is employed regularly in every industry that has to compete for its survival. It's a little late in coming to the Judicial Department, a little late in coming to government, because the judiciary profits from its failures. The more of our cases that come back as recidivists, the more demand there is for judges, prosecutors, police, prisons, the whole industry. I'm not suggesting -- I'm not a conspiracy theorist -- I don't think people are consciously planning to increase their profit by undermining rational justice. I think the problem with justice is that it still has some very medieval roots. And although those roots have some proper role, they have no business displacing entirely a rational approach.

This technology is data warehouse technology. It is just the current state of the art for how you approach such a problem. Its critical factors from a technological point of view, and the ones that are important for this discussion, are first, that it draws on operational data. This is not supplied by people sitting down for the purpose of collecting information to support this analysis. This is a process that pools data from multiple sources within the criminal justice system, for the purpose of preparing and comparing analyses which help us all make better decisions.

Now I'm not connected to the database. What I'm about to show you is just a demonstration of how it works. It's not actually working, although the screens that you will see are actual screens from actual queries against real data. The piece that I'm proposing and promoting, and the one that's growing I think state- wide, very slowly, but growing, is the Sentencing Support Tool. The way it works for judges is that typically you would enter an OJIN number -- you can select any of a great number of identifiers -- you can use a name and search if you don't have a number -- and the case number. The way our court system works is that the [OJIN]
case number identifies an individual.

When you "submit" that information, you then get a screen which allows you to select the charge for which you are sentencing this individual. It proposes count one. It gives you the other charges in the case, and then allows you to search for and to select any criminal charge known to Oregon state or local law. So if you have an ordinance against littering in Eugene, it'll be here, because when you actually get to sentencing, you may be sentencing on a charge that's very different than the one that was filed. When you submit that, the program goes to work. What it's doing now is accessing data that's already been collected from operational systems for people who it knows to be like the offender. You've identified the charge, Theft in the Second Degree. It knows who the defendant is and can extract his criminal history is from the various data bases, and it can compare him with all of the people who are like him who have been sentenced for a property crime, and give you the results in terms of new criminal activity for all of the elements of [sentences available] for that crime. When you finally get the results, which typically takes about 90 seconds, what you get is a bar chart.

Let me concede at the outset that we are at the very threshold of using this technology. When this finally worked against real data, I sent an e-mail to judges on an international mailing list comparing this to Kitty Hawk. We have as far to go [as the Wright Brothers did to get to] a 737, but it is as profound a potential change as that flight at Kitty Hawk was.

This has not been done, as far as I can tell, anywhere else in the world. But it's happening in Oregon, and I think it's spreading. I hope you will see some of the reasons it's spreading.

In this bar graph, from left to right, in order of frequency, we get bars representing sentencing elements: Restitution, community service work, probation 1-18 months, jail 11-30 days, jail 4-10 days, forest work camp, and so on. Below the bar chart is a table that lists all of the raw numbers. When you get into the red numbers, they are below 30, and they are not showing up as bars. That's roughly equivalent to statistical significance to the extent that that's of concern. We don't display bars for things that are too rarely imposed. Lurking under each of these three tabs are the assumptions that the program made about this sentencing occasion.

For example, you will recall that we are sentencing this defendant for a Theft II. So the program defaults to comparing only those people sentenced for property crimes.  You can change that [under the "Charge" tab]. What we know about this defendant is lurking under the "Profile" tab. And what we've done, and I think this is an innovation which is itself pretty important -- criminal histories vary, and it's not just whether they're felonies or misdemeanors, or how [frequent] they are, but they vary according to themes. Some people are property criminals with drug involvement, some have drug involvement with sex crime involvement, some are violent, some are not, and so forth. We've established data rules for each of six categories, ranking people from "none" to "severe" in crimes of violence, sex crimes, drug crimes, property crimes, DUII/major traffic, and domestic violence crimes. These crime themes boxes are calculated based on what the databases contributing to the warehouse know about this offender. the Criminal History score, here, is gathered from the last occasion on which the person received a sentencing guideline under the regular "criminal history" [as designated under the sentencing guideline rules]. This offender is 18, or that's what we understood coming into this [this graphic shows other age boxes checked by a user; it began with only the 18-20 box checked], and we know his ethnicity.

Now I understand, and this keeps coming up for good reason, that ethnicity is an issue. One of the things that I need to tell you , of course, is that is that you can change any of this and you can take ethnicity out simply by unclicking.

You can say well, I don't like my client being treated because of his American Indian or Native American ancestry as a Native American, you can unclick that, you can treat it as if he were like anybody else. You can click all the boxes or none and recalculate. The reason we have ethnicity in there is because there are programs in most communities, certainly in Multnomah County, that target ethnic groups, minority groups for special programs like House of Umoja for African American youths, there's NARA -- Native American Rehabilitation Association -- for Native Americans. What we want to be able is to carry the treat programs that aim there approach to specific groups, so in order to compare outcomes we need that flexibility.

In addition, the ability to slice the offender population by ethnicity may be necessary to account for disparities caused by racial profiling. In other words, if a member of a given minority is more likely regardless of criminality to be arrested, prosecuted, and convicted of crimes, it is neither fair nor accurate to compare that person with non-minority offenders whose criminal records although apparently equivalent actually represent a higher level of criminality.

Let me go to what I think is at least responsive to the question of uniform recidivism measures. Because this offender is a property offender, the program decided to start with a measuring recidivism by a conviction for a property crime within three years of the sentence that it is comparing. And this little window right here selects the slice of data that you are looking at -- by default it is 1988 and 2000.

What is critical about this is, if I were sentencing a property criminal, as I have, for a property crime somebody whose real problem is violence, and I just happen to have the property crime, I might very well want to look at violent recidivism by people so sentenced. You need to be able to look at various recidivism measures. You can talk about what it is you can accomplish. If there is an opportunity to accomplish something significant for an offender, something that may not be precisely involved with the crime of conviction, but you can reduce his violence or his likelihood of reoffending as a sex offender, but you have him on a drug charge, you need to be able to look for that. It is an opportunity; you need to be able to see it and argue about it. But let's say instead of worrying about property crime only by Shawn Hermo, we want to change it to any charge occurring within five years -- so we've just changed the outcome measure. In the courtroom or in preparation for the hearing, or by probation officers, you can hit recalculate, and we get a new set of bars because we have just changed the outcome measure. Now we're looking for the same group of people, sentenced for property crime, based on the likelihood that they are going to succeed by the recidivism measure, which we have changed: any crime within five years.

We can also change the profile. Let's say that we found out now that we get to see this offender, he's not really 18 to 21 at all. He is really old. He's really 26, so we click 26 and we recalculate and we get completely different results.

Now one of the reasons we might want to recalculate is we may have learned more in the courtroom about criminal history. We may have learned that something attributed to this offender that doesn't belong him. We may have learned how that something came from Arkansas puts him in a different category altogether. We can make the change and rerun the calculations.

But we have a long way to go. This does not by any means represent the complete pinnacle of this opportunity to aim sentencing decisions at more effective results. I give you two propositions. One is we are doing a horrible job of preventing reoffense. That's gotta be obvious to anybody who takes an objective look at the statistics, and I showed you the ones involving the problem -- 22 of the 23 arrested for burglary were in jail for something else. That's one pretty good danger signal that we are not making the best choice. But why are we not making the best choice? Because we are not even trying without this kind of approach. It's not the discussion we have in the courtroom; it's not the discussion either in probation revocation. What works has to be part of the discussion in the courtroom and this tool will make that part of the discussion in the courtroom, just as the sentencing guidelines make the guidelines the discussion. When somebody walks into the courtroom at a felony sentencing, they carry the grid. So when judges and attorneys are used to looking at this, it will insist by its presence that at least part of the discussion be about what is most likely to reduce criminal behavior. We have a long way to go. We have to look at the composites of sentences, the program ought to be smarter, as I described in that goldenrod piece that I passed out, called Resolution News. And by the way, the resolution I read to you from the Judicial Conference -- the "therefore" clauses are part of a masthead. "Resolution News" is that resolution, and this is a device by which I keep judges apprized of our progress. The progress is very frustrating in it's speed; it is steady, but it is slow, and it's slow for all sorts of reasons. The one which is obvious to me is that government organization information technology is a very different animal than information technology in the private realm. Multnomah County has made its great surge forward because we got a bond to support criminal justice uses and we hired some consultants who are essentially from private industry. The fellow who started this data warehouse project came to us from American Express.

If you go to a hospital today, chances are your treatment is presumed, based on an analysis of how people similarly presenting fared after being treated with a range of things that could happen. Why? Because HMOs have a profit line, and if they don't do something to you that is cost effective by reducing the likelihood that you're going to come back with a more expensive diagnosis next time, they are wasting their money and somebody else who will compete them into the ground. We need to apply the same smarts to criminal justice. This is not the only purpose of sentencing, but it is critical to gain any respect in the community, and I frankly think that the way to get any respect is to start focusing -- holding judges accountable -- which is the way our new Police Chief in Portland put it -- holding the courts accountable for their product.
 

As long as we can walk around in the modern equivalent to that     ===}> and say that our job is not public safety, our job is just deserts, our job is "sending a message," our job is making the "punishment fit the crime."

If you want to send a message do it there ==}>   Let us know if you get a response.

Giant Antenna at Arecibo used for SETI (search for extraterrestrial intelligence)

 If you want to do something about public safety, if you want to do something about criminal justice, then do what we can to make better choices about whatever percentage it is of our criminal population that we can actually divert from criminal careers.

Now, I have no idea how many people we can actually divert from criminal careers, but I do know that our problem is criminal careers, it is not one-time crimes. Sure, there are some one-time crimes that we need to deal with, but most of the people we see keep coming back, they keep coming back in tremendous number, and I cannot believe for a second that we are doing as well by accident as we could do if we made a conscientious effort to use good tools and good data to make better choices with the outcome as a piece of the goal. I would be happy to answer questions.

A: How many of your fellow judges are capable of using this system?

Q: The system -- we have trained five data users, and have them hooked up; they have access to the device and they can use it. We have trained a small group of defense attorneys as beta users and Tom Cleary is arranging a meeting and training for district attorneys. It's not part of the process in any other courtroom yet, but speed is not our most important product, in spite of the numbers you may have seen; not many is the answer.

Q: I was going to ask Judge Ross if this is a tool that could be used in North Carolina because in all the conferences I have attended, yours is the first presentation I've seen like this.

A: There's interest in the idea, Tom Simpson who is one of the two managers of the project is from from the Multnomah County DA's office. He's been going around to counties, a lot of counties are very interested in this, and in trying to emulate it, we have been getting a lot of interest from the Center of State Court's is interested in this, and we've been getting inquiries from Nigeria, Scotland, Australia and Canada, but no one else is doing it. By contrast, you saw the (image of) the judge with the computer, there is a sentencing support system in New South Wales, and there is another one in Scotland, and what it does, the one in New South Wales, does this. It gathers a lot of information about the treatment programs, and a lot of information about the offenders, and then it gives you a chart that shows you what judges usually do. And the assumption is that you should do what judges usually do. Why? Because judges know so much. Well if judges know so much about what to do, how come the recidivism rates are so high? I say do not try to emulate what judges do.

Scotland's gone a step further and they have built models which try to emulate judicial thinking. Anybody who has tried to work with more than one judge at a time, that's a very scary thought, emulating judicial thinking. Judges are conscientious, many of us are actually pretty smart, and most of us want to do the right thing. But all of us have no access to this kind of information and therefore we rely on folklore, mythology, whim, ideology, philosophy, personality to come up with what we think is right. And if we pronounce it with the right tone of voice, people think well, that must work, and that's what happened with the guy with the beak. Everybody thought he was a doctor, so he must be right. Well, judges aren't right unless they have the right information.

Q: My name is Bridget Jones and I am with the Robert Wood Johnson Foundation, Juvenile Justice Initiative. There are a couple of things that I would like to ask you about. I find this prospect kind of scary. First of all, it doesn't take into account or does it take into account things like the subjective nature of charging; does it take into account the facts of the case. Stealing something to eat is different than stealing something for pleasure, and the other thing is where does mitigation and aggravation come into this equation.

A: First of all, this is not a sentencing program. It does not give you a sentence after you input data. It's a piece of information that you have in addition to the other information that comes to you in the sentencing process. And of course the nature of the crime, is it typical of the crime, the aggravation, mitigation, was it an excessive or an insufficient sentence . . . You may find for example that somebody who has killed somebody with a car who, because of their background, is completely safe on probation, and yet for reasons other than public safety you would still use incarceration because there are other components (of sentencing). Likewise, you are not going to put somebody in jail who shouldn't be in jail because that's excessive under the circumstances, because of somebody who steals literally to eat, when they have exhausted the other choices. It is not supposed to make decisions for judges, it's supposed to focus the process on something other than whim, ideology, mythology and the secular version of religion, to determine public safety, because it hasn't worked.

Q: My question though is you are using it as a factor to determine whether that person is going to reoffend or not, is that correct?

A: What the program gives you is how some people who are similar in at least in those aspects that you can retrieve from the database have performed after you have done the things that you might do to this person, so yes, to that extent you are using that information.

Q: And that's based on raw data, the charge itself, what was done as a result of that charge. Is there any place in this program though that speaks to some of the factors that are behind that data?

A: The program is limited to the data that's available, and there are limits on the data that's available, absolutely. The problem -- I certainly understand, æcause I've seen it -- that the idea that people will somehow become insensitive to the details of the individuals or will be insensitive to the policies of the local prosecutor, or the plea bargain process or whatever it is, in coming up with a sentence which is driven, which is not representative of the crime initially charged, and this program doesn't address that. But what it does do is that it competes with unbridled discretion where it exists for public safety. And frankly, if you look at the data that you get back, I think most people who are concerned with excessive punishment for minor criminals will find that indeed the lighter sentences work better for lighter criminals, so look at the results before you are too afraid of them. What I am afraid of, is a criminal justice system which so profoundly fails to serve public safety, with the result that the public's understandable reaction is simply to lock more people up longer and longer without using prison to focus on the people who really need to be there. And much more importing is diverting from criminal careers those budding criminal careers which can be diverted before they become fodder for extreme victimization and for the extreme result of the Texas prison system.

Q: Thank you.
 

Q: I'm Ross Shepherd, the Lane County Public Defender. Has your program been helpful to you in drug cases, in determining what sort of treatment or alternative sanctions are appropriate in individual cases?

A: The program has not yet really been used in drug cases, it's actually rarely been used at all. In my view it has tremendous potential because we (now) lump people together by factors that don't necessarily make them alike. The more data we can access -- and one of my roles is I am on the IITP Inmate and Incarceration and Transition Plan Design Committee. IITP is this wonderful program that has been designed by the Department of Corrections -- and you will see why it connects to your question in a minute -- which assesses needs of inmates as they come into the prison system along at least seven axes, including substance abuse, mental health, physical health, vocational training, education and so forth, and keeps track of that data, makes a plan for them in prison, times it out, it's basically a project plan for them, what they should do be doing. One of the things I want to do with that data is exploit it in the warehouse, so that we can start profiling people about the nature and extent of their drug involvement and the other factors that are variable, so we can start seeing which things work best on which people, instead of treating everybody -- the "drug package" is a standard condition of probation, recommended in very drug case. Well, for some people and probably for most, it's a perfectly rational thing to do. For others, it's just there because that is what we do; we make no attempt to divide people. I believe firmly that (sentencing support tools will be) very powerful, it just hasn't been used yet. It will be, and you can always assess it because the data will be there.

Q: Could we think of this scheme as a hyper-sophisticated sentencing guidelines, where you are very finely defining which box it is we are going to put the offender depending upon all the factors that the computer knows about the background.

A: Yes and no. On the one hand, it very much more finely categorizes people for some information about what you might want to do with this person. On the other, it doesn't purport to cover all of the circumstances that should contribute to a sentence. What the guidelines tell you is stay here unless there's a damn good reason for leaving. What this does is say this is what has happened to people who are similar, if they are really similar, when we did these things to them. Now that doesn't mean -- and again, this is only the recidivism reduction part -- assuming the person is like that person, one of the things that you want to argue is, is this person really fairly characterized by what we can find out about similar offenders. So yes, it's like the guidelines in that it is much more finely categorized; No, it is not like the guidelines in that it doesn't purport to cover all the purposes of sentencing, or to have all the information. My frustration with the guidelines -- I spoke about this when they were first under discussion, and been critical of the guidelines for this purpose all along -- the guidelines, for example in serious cases don't distinguish psychopaths from non-psychopaths, and if you know anything about psychopathy, you want to know that if you are sentencing somebody on an assault. We treat people identically if they fit in that box. And another thing that frustrates me about the guidelines is they are not focused on reducing recidivism. They are a very strange mix of just desserts and prison beds. Again, by gettoing dropped on my head I lost the ability to follow the tune too far. I am not sure that modulating prison bed use is the ultimate question in criminal justice. It seems to me that the ultimate question is reducing victimization that can be avoided. and reducing oppression of the fodder for the criminal justice system which isn't productive of public safety; and sending to programs that can help people the people they can help; not sending anybody to programs that they can't help; and using prisons most effectively on the people who need to be locked up either because we can't ever trust them outside æcause they're dangerous -- which is a substantial and important hunk of our people -- or while we deliver to them the programs that will make them safe to be in the community. That's what we need to do, and we need a rational approach to do it, and this program is primarily designed to insist on a rational approach, in the courtroom, pre-trial release decisions, probation decisions, correctional counseling decisions, and the same stuff for policy decisions about what categories of sanctions we ought to support, what programs ought to be promoted with public expenditures and so on.