Wednesday, February 16, 2011

Runaway Trolley - Applied to Criminal Recidivism

The Runaway Trolley thought experiment was introduced in a previous Topic on this Blog.

The basic lesson was that there are situations where it is quite ethical to take an action that saves (or benefits) a number of people, even if that action has, as an inevitable side effect, the death (or detriment) of a smaller number of people.

This posting has to do with Criminal Recidivism, and how we might apply the lesson of the Runaway Trolley to the justice system.


Click HERE to download a narrated PowerPoint Show that includes animated charts for the Runaway Trolley thought experiment. After the Runaway Trolley is explored, the charts continue and apply the ethical lesson to two real-world issues: 1) Criminal Recidivism and 2) End of Life Issues. This posting covers Criminal Recidivism only. A subsequent posting will cover End of Life Issues. The PowerPoint Show is based on a talk I gave to The Philosophy Club at The Villages, FL, on 04 February 2011. NOTE: The Powerpoint Show is Narrated and plays and advances automatically after download to your computer.


Moses Maimonides, the 11th Century Rabbi and medical doctor shown in the sketch above, is one of the most highly regarded ethical teachers in Jewish tradition. He famously wrote that it was better to let 1000 guilty go free rather than wrongly convict a single innocent. Do you agree with that ideal? Is an error rate of 1/1000 the correct standard for a criminal justice system?

Juries in criminal cases are charged with the responsibility to convict only if the evidence meets the standard of being beyond a reasonable doubt. The 1/1000 standard corresponds to a certainty of 99.9%. Is that a good working definition for beyond a reasonable doubt? Is that number too high? For example, in civil cases, the standard is the preponderance of the evidence, which means, if one side proves its case to a certainty of 51%, the other side loses. Is 51% a good working definition for beyond a reasonable doubt?

Well, Benjamin Franklin, US Founding Father, said the number of guilty released to save one innocent from being wrongly convicted was 100, which corresponds to 99%. William Blackstone, the 18th century jurist who codified British Common Law, said the correct value was 10, corresponding to 90% certainty. Benjamin Cardozo, 19th century US Supreme Court Justice, said the number was 5, corresponding to 80% certainty. And, Voltaire, the 18th century French philosopher, said the number was 1, corresponding to 50% certainty.

How in the world can so many respected men have such different standards for criminal justice? How may we use the Runaway Trolley to arrive at a reasonable number?


The graphic lists criminal recidivism rates for a number of crimes. Notice that nearly all violent crimes, including assault, murder, robbery, and sex crimes have recidivism rates above 50%. That means that, when a person has been convicted of a violent crime and has served his sentence and is released, there is a greater than 50% likelihood that he will commit another violent crime, be caught, and convicted again.


Criminal Recidivism rates for violent crimes teach us that, every time we release two convicts, we are, in essence, condemning at least one innocent to become the victim of a future crime. The number of innocents condemned is probably considerably larger than one because: 1) Many violent crimes have more than one victim, and 2) The released convict is likely to commit more than one violent crime before being caught and convicted again.

Extending this lesson to criminal trials, if, after weighing the evidence of a violent crime we the jury believe there is a greater than 50% likelihood the defendant is guilty, we should vote him guilty!

If there is only a 51% chance he is actually guilty, and we release him on a technicality or because we feel sympathy for him, and it turns out he was actually guilty, we are denying justice not only to his victims in the current case, but there is a high likelihood we are also condemning future victims to violence.

If we convict him on 51% certainty, and he turns out to have actually been innocent, we are doing a serious injustice to an innocent man. But, what is the likelihood he is totally innocent? Unless corrupt police have purposely framed him (in which case they would most likely have manufactured overwhelming evidence, which is not the case here), we are probably dealing with a person who has an extensive rap sheet and other indicators he has not lead a respectable life. He may not be guilty of this particular crime, but his incarceration will not be much of a loss to society - certainly not as much as the death or serious injury for one or more totally innocent victims if we make a mistake and release someone who is actually guilty.

The justice system is so dominated by lawyers and legal technicalities that rich people with clever lawyers can literally get away with murder, as many of us think happened to OJ Simpson.


1) Reform the Court System. Change the rules of evidence. Make it harder to get off on a technicality or with a clever lawyer.

2) Change the way we handle people convicted of violent crimes. DO NOT release them after their sentence is served. Keep them in some type of work camp.

3) Perhaps modern technology provides a humane and affordable solution for dealing with released convicts and others with extensive rap sheets. Stick a chip up their butt so they may be tracked for the rest of their lives. If there is certainty they will be caught and convicted if they commit any kind of infraction, they may learn to stay on the straight and narrow.

My free online novel, 2052 - The Hawking Plan, envisages a society, several decades from now, when everybody "voluntarily" carries an RFID device that effectively tracks their every move and activity.

Does that sound too drastic? Well how about your total lack of privacy right now? Those of us with homes and computers and cell phones and cars and jobs are effectively tracked by various computers and video cameras as we go on with our lives. We leave video and computer records dozens of times every day. The only people who have any privacy anymore are the drifters and criminal class, one of whom is likely to steal your car or credit card!

NOTE: Subsequent postings in this series will extend this lesson to the real-world situation of End of Life Issues.

Ira Glickstein


Howard Pattee said...

Ira’s “tough on crime” recommendations will not improve the overcrowded US prison system. We already have a ridiculously high prison population― the highest US prison population percentage in the world.

Ira assumes that the recidivism rate is fixed and has nothing to do with the individual (sociopaths, kleptos and adolescents are different) and with the prison system itself. This is not the case. It is well-documented that some types of drug rehabilitation, job training, and education in prisons reduce recidivism. So do programs for rehabilitating released prisoners, but few states have such programs.

Reliable statistics are hard to find, but China practices prisoner rehabilitation at all these stages and claims a recidivism rate of 6 per cent. Maybe there is an alternative to Ira's lesson here.


Ira Glickstein said...

Thanks Howard for your mention of China's remarkably low recidivism rate, claimed to be only 6% (compared to our 50+% for violent crimes).

Well, William F Buckley is said to have had a solution for the recidivism problem: The death penalty for anyone who commits a violent crime, and, to show we are humane, impose a 25-cent fine for any repeat offenders.

China seems to have followed something like Buckley's sarcastic advice for real. According to Wikipedia,

"Compared to other countries, death sentences are carried out quickly in China. After a first trial conducted by an Intermediate people's court concludes with a death sentence, a double appeals process must follow. ... The execution is carried out shortly thereafter and is fairly automated. ... Capital punishment in China can be politically or socially influenced. In 2003, a local court sentenced the leader of a triad organization to a death sentence with two years of probation. However, the public opinion was that the sentence was too light. Under public pressure, the supreme court of China took the case and retried the leader, resulting in a death sentence which was carried out immediately. ...

"Capital punishment is applied flexibly to a wide range of crimes, some of which are punishable by death in no other judicial system in the world. ... Corruption, property crimes such as theft, and smuggling gold, silver or other precious metals are also amongst the 68 crimes that are eligible for the death penalty in China.

"Capital punishment is also imposed on inchoate offenses, that is, attempted crimes which are not actually fully carried out, including repeat offenses such as attempted fraud. The recidivistic nature of the offenses, not their seriousness per se, is what is adjudicated to merit the capital sentence. ...

"By the confirmed numbers, the rate of executions in China is higher than the United States and Pakistan, though Iran executes more prisoners per capita. The Dui Hua Foundation declares that the true figures were higher; they estimate that China executed between 5,000 and 6,000 people in 2007, down from 10,000 in 2005.

"The exact numbers of people executed in China is classified as a state secret ...

"Pressure placed on local and regional bureaucracies under the auspices of the "strike hard" (严打) campaigns has led to the streamlining of capital cases; cases are investigated, cases and appeals are heard, and sentences carried out at rates much more rapid than in other states."

See also this story on The Harsh Realities Of China's Death Penalty.

I agree that the recidivism rate varies by type of criminal and crime and that education and drug rehabilitation reduce recidivism - but, by how much? Say we could cut the average 50+% rates for violent crimes in half (which is doubtful, but assume we could). That means for every four convicts who have served their terms and are released, only one will commit another violent crime against an innocent person. Sorry, but I'd rather have all four stay under control in a work camp or using tracking technology than have one innocent future victim brutalized. Why? It could be ME (or YOU, or our loved ones).

Ira Glickstein

Howard Pattee said...

Ira, why do you think your draconian recommendations would solve the problem of the prison overpopulation? As a better alternative why not an improved Second Chance Act like the one Bush signed. Of course it hasn’t a chance at reauthorization with the Tea Party congress.


Ira Glickstein said...

Howard, please check your link to the Second Chance Act proposed by now-disgraced Congressman Rangel and signed by President Bush - it is strictly for NON-violent crimes. My "draconian" ideas are about protecting the innocent from future VIOLENT crimes and have nothing to do with NON-violent infractions.

I did not claim they will address what you call "prison overpopulation". However, there are two aspects of my recommendations that might affect the prison population and both violent and non-violent crimes:

1) Requiring ex-cons who are released after serving their sentences to have a tracking and alerting chip. Banks and stores that are likely to be robbed or burgularized could install a chip-reader. It would record the chip ID number of any ex-con who happened to be in the vicinity. The reader would alert the staff during the day so they could take protective measures against robbery, and would record ID numbers at night in case they were burglarized. The police would thus have the ID number of the ex-con who would be the prime suspect.

Knowing that they will almost certainly be detected and caught would probably deter ex-cons with ID chips from committing crimes and that might relieve the prison population.

2) I believe that work camps, for ex-cons who are currently released into the general population, could be designed to be self-supporting, or nearly so. Given such work camps, some convicts who still have time on their sentences could be accommodated there, effectively reducing the prison population.

I have toured our local Marion County jail a couple times. Most of the prisoners have regular jobs within the prison or on the prison farm that raises most of the food they consume. Working prisoners, with black and white striped clothing, walk unescorted within the jail, to and from their jobs, while those with green and white stripes walk unescorted to the doorway to buses that transport them to the prison farm or other outdoor clean-up work. Only a minority of the prisoners, who wear orange or red outfits, do not have jobs and must be escorted when not in their cells.

Ira Glickstein

joel said...

Since your previous post and presentation on the run-away trolley I've been trying to figure out what my personal value of n might be. It's been a challenge, because what is expressed in mathematical form is in fact lacking in mathematical rigor. Specifically, the a priori probability of guilt is not considered by any of the philosophical experts. A proper analysis must consider how the decision threshold of doubt is influenced by the a priori ratio of guilty to innocent. I'll try to get my thoughts in order so that they can be communicated rationally. We should also distinguish between releasing accused based upon some level of doubt and those that are certainly guilty, but are released due to such things as failure to give Miranda warnings or "poison fruit."

In researching this subject I came across an excellent review which I recommend to you despite its disconcerting tongue-in-cheek approach. Here's the citation and a short sample quote:

Jeremy Bentham, founder of utilitarianism, warned against the warm fuzzy feeling that comes from large values of n:

We must be on guard against those sentimental exaggerations which tend to give crime impunity, under the pretext of insuring the safety of innocence. Public applause has been, so to speak, set up to auction. At first it was said to be better to save several guilty men, than to condemn a single innocent man; others, to make the maxim more striking, fix the number ten; a third made this ten a hundred, and a fourth made it a thousand. All these candidates for the prize of humanity have been outstripped by I know not how many writers, who hold, that, in no case, ought an accused person to be condemned, unless evidence amount to mathematical or absolute certainty. According to this maxim, nobody ought to be punished, lest an innocent man be punished.

Ira Glickstein said...

Joel, your quote from Bentham, where moral philosophers engage in a contest of higher values for n (the number of guilty who should be released to spare one innocent from wrong punishment) to show how humane they are, reminded me of an old joke.

A few guys are in a bar and one, to demonstrate how little money matters to him, takes out a dollar bill and sets it on fire. The second guy, to top him, does the same with a five spot. The third burns up a ten-dollar bill, topping them all. The fourth takes out his checkbook and writes a check for $10,000. Then he burns it. He wins!

In the contest to release more and more guilty men, sentimental judicial exaggerators write larger and larger "checks". But it is we unlucky innocent victims in a dangerous society who get to "cash" them, fearful of leaving our homes when it gets dark, protecting ourselves with burglar alarms and automatic spotlights, and, in the end, being brutalized by former convicts who, in the stupid phrase of humane jurists, "have paid their debt to society". What a load of bullcrap that last phrase is!

Ira Glickstein

joel said...

Consider 100 defendants accused of the same charge sitting randomly around the court room. Let us say that we know a priori that one among them is innocent while all the rest are guilty and that the quality of the evidence is the same in each case. If we apply the notion that it is better to acquit 99 guilty than allow a single innocent to be convicted, we are forced to let all 100 free without even bothering ourselves with trials. Let us say that we subscribe to the notion that it is better to allow 10 guilty to go free rather than wrongfully convict an innocent. If we allow ten chosen at random to go free we still may have the innocent person among the remaining ninety. Applying the rule again, requires that we dismiss ten more, etc. until they are all acquitted.

Now consider 100 defendants accused of the same charge, however they sit in ten different rows according to the strength of the evidence against them. The first row is labeled 100% and the last row is labeled zero percent based upon the probability that the evidence will convict. All of the accused in the first row are guilty and all in the last row are innocent. There are ten people in each of the ten rows and they may be either guilty and innocent. Suppose that the judge instructs the jury that any defendant for whom the quality of evidence is fifty percent or more is to be a judged as guilty. On this basis some of the innocent will be convicted. Another judge replaces the current judge and he has the philosophy that it is better to release 10 guilty than convict one innocent. He convicts all the people in row 10 (perfect quality of evidence). He also releases all of those in the last row where the quality of the evidence is zero. He then releases all the other accused in every row, because there still is a probability that there is an innocent in each row along with nine guilty. No matter what the value of n, as long as the innocent and guilty are inseparable, all the accused must be released. Thus, this old chestnut is meaningless in any practical sense. It is just an expression of sympathy for the innocent no matter how many famous philosophers and jurists may quote it. It serves also as an vacuous argument for the injustice of letting the specific guilty party go free on a technicality or the incompetence of a prosecutor.

Ira Glickstein said...

Well-stated Joel and that should convince any reasonable person beyond a reasonable doubt that Voltaire was right and n should be "1". We should structure the justice system such that we convict a defendant on a violent crime charge if the evidence is 51% that he committed the crime. That means there is a 49% chance we are convicting an innocent person. However, we must weigh the wrong being done to that innocent person against the one or more other innocents who will be future victims based on the 50%+ recidivism rate.

Some will object that a 51% standard gives the police and prosecutors free rein to frame and convict anyone they do not like. But, in practical terms, they can manufacture phony evidence up to any standard, and some corrupt prosecutors have.

Ira Glickstein

Howard Pattee said...

Ira concludes, “We should structure the justice system such that we convict a defendant on a violent crime charge if the evidence is 51% that he committed the crime. That means there is a 49% chance we are convicting an innocent person. However, we must weigh the wrong being done to that innocent person against the one or more other innocents who will be future victims based on the 50%+ recidivism rate.”

Aren't you missing the fact that this 51% logic leaves a real killer free to kill 49% of the time?

With such poor detective work, how many trials will it take before you expect to catch the real killer?


Howard Pattee said...

Joel proposes, “a priori that one among them is innocent while all the rest are guilty and that the quality of the evidence is the same in each case.”

If there is no difference in the “quality of evidence” that simply means there is no justification for a trial. With no evidence, what kind of arguments would you give at such a trial?

Joel’s second case of graded rows has the same problem. If there is no evidence in the middle rows to distinguish individuals there is no basis for a trial.


joel said...

Howard, I didn't say that all of the defendants are accused of the very same crime. There are 100 individual crimes and 100 individual defendants. All must be set free if you believe unless the evidence is perfect. The problem is ill-posed if you believe that it is better to set n free rather than convict one.

The quirk arises because one imagines that it is something like a hostage exchange. Obviously it is not, because one never knows who is on the "guilty" team or "innocent" team.

I like your point about the "real" criminal being on the loose to murder again when an innocent is convicted in his or her place. That's also one of the arguments for capital punishment. An unexecuted murderer has a license to kill while in prison.

Ira Glickstein said...

Howard writes:

"...Aren't you missing the fact that this 51% logic leaves a real killer free to kill 49% of the time?

"With such poor detective work, how many trials will it take before you expect to catch the real killer?"

OK, let us run the numbers. Say there are 100 separate violent crimes, and in each case we the jury judge that there is a 51% chance that the defendant is really the criminal in that case.

So, we send 100 people to jail and 51 of them are really guilty while 49 are innocent of that crime.

By sending 51 really guilty to jail, and assuming that had we released them, at least half would commit another violent crime with an average of at least two victims each, we have saved 51 innocent future violent crime victims.

Net-net, 49 innocents are wrongly suffering in jail but 51 innocents are not victimized by future violent crimes, for a net benefit to society of 2.

That is based on the idea that the 49 wrongly convicted are of the same quality of innocence as the 51 totally innocent victims of future crime. But, as I pointed out, it is most likely the wrongly convicted, while they did not commit the specific crime for which they were convicted, were suspected due to past bad deeds, so they are less worthy members of society.

So, we have a net benefit of 2 (2%) to society if we judge all people as exact equals ("justice is blind"), but better than 2% if we take into account that some are less desirable types ("But justice need not be stupid as well").

As for your concern about the real criminals who have not been caught or convicted in the 49 wrongly-adjudicated cases, they will be free in any case because, if the police had any evidence against them, they would also have been tried.

It seems you assume that, once someone is convicted of a given crime the police stop trying to find other suspects in that crime. If we the jury set those 49 free, the police will have to reopen those cases and, in some instances, they will find and convict the real culprits. You are correct on that score.

However, if the police know the standard of jury judgment is "preponderance of the evidence" rather than "beyond a reasonable doubt", they may keep all close cases open and continue to look for other suspects.

Ira GLickstein

Howard Pattee said...

This kind of statistical analysis is unrealistic. No normal juror is going to vote guilty on a 51% “hunch.” In practice that is coin-flipping. Anyway, justice is a more complex ethical issue. There are at least two well-known conflicting value systems, individual liberty vs. social benefit (self interest vs. group interest). There are also the related conflicting values of objective truth vs. moral right, which produces the conflict between science and religion.

Your statistical arguments are all about social benefits and moral rights. They avoid the ethical value of individual liberty and objective truth. There is an objective fact of whether I committed a specific criminal act or not. The fact does not in any way depend on moral values, on the jury, my politics, or on probabilities like the recidivism statistics.

The West’s judicial system began with the Enlightenment’s libertarian and truth ethics. Seeking truth requires everyone to try their best to avoid all these possible social influences. That is one reason sentencing is entirely separate from reaching a verdict. Sentencing is inherently a social issue, the truth is not.

With this ethics, trial of an individual is not statistical. It is not like predicting the weather by comparing the present weather pattern with the statistics of past patterns. This is what some conservatives today want to do by profiling. If you are a Muslim you are more likely to be a terrorist. If you are black and poor you are statistically more likely to be guilty.

Once you begin worrying about “the good of society” or what you should do with the accused because of the statistics of recidivism, you pervert the ideal of truth and individual freedom. (e.g., see the Ministry of Truth in Orwell’s 1984).

On the other hand, in a socialized culture values depend on the social group in power. Group attitudes, like patriotism, are valued more than individual liberty. For example, “unpatriotic” is the first accusation that conservatives use against liberals who disagree with them.

Of course, I’m simplifying. In practice, matters are more complicated. There are no pure libertarians or socialists. Few libertarians would grant the same individual rights to an accused terrorist with access to WMDs as to a common thief. Guantanamo is an example. Even torture is allowed if it saves enough lives. Similarly, even hard core socialists and communists will rebel if the state takes away too many of their personal freedoms. Then there are the ethically challenged Tea Party fanatics who simply can’t handle issues with two sides, and default to conspiracy theories and religious dogma.


joel said...

I'm beginning to think that the root of the problem with the expression "it is better to acquit "n" guilty men than to convict one innocent" is grammatical or syntactic. What the "sayer" wishes to convey is that given there are two actions possible which are mutually exclusive: acquit the n guilty or convict an innocent, the first is preferable to the second. It is a statement of the form; A rather than B or A better than B or A preferable to B.

Another example would be, Henry Clay's "I'd rather be right than president." I choose this saying, because It demonstrates so well the importance of mutual exclusivity. It assumes that being outspokenly right (about thorny political issues) excludes one from being elected president. This mutual exclusivity is the crucial to the quote, or else there would be a third choice which would be being right and also being elected president. This might be the case, but Clay's statement would be trivial. His point is the exclusivity.

Are releasing "n" guilty and convicting one innocent mutually exclusive? Can we not acquit n guilty and convict an innocent at the same time? The answer is obviously "yes." This is the source of our problem. Since we don't know who is truly innocent or truly guilty, "n" is meaningless. The operation cannot be carried out. All we can say is that by making it more difficult to convict anyone, some of the additional acquitted will be innocent and some will be guilty. Some will go on to commit crimes and some will not. I think the more fruitful study would be to examine cases that have been adjudicated both as torts and felonies such as the O.J. Simpson case. I haven't been able to find such a study on the internet.

joel said...

I agree with Ira that rates of recidivism give us a clue as to remedy. We take away a felon's right to own a firearm. I some states we take away the right to vote. We don't recognize a sexual predator's right to freedom of movement. We should use whatever technology we can to discourage recidivism. Tracking devices and listening devices are not out of the question.