Monday, December 19, 2005

Research for someone else to do

Project I: There is one critical fact about the criminal justice system that nobody knows--the rate of false positives. Of people convicted, how many are innocent? Opinions about issues such as the death penalty, plea bargaining, admissable evidence, and much else depend on that fact. So far as I know, there is no serious research that would let us estimate the rate--only anecdotal evidence.

I have a solution to that problem. Find a jurisdiction friendly to research. Identify all convictions in that jurisdiction from before DNA testing for which tissue evidence survives, such that testing it should provide strong evidence for or against guilt. Test all of it--or if that exceeds your budget, a random sample. See how many of those convicted turn out to be innocent.

It isn't a perfect test, of course, since suitable cases are not a random sample of all cases. But it would give us at least a first approximation of an answer. If the rate turned out to be one in a thousand, the case for the death penalty (and many other things) would be a lot stronger than if it turned out to be one in three.

Of course, a good deal of such testing has been done by innocence projects around the country. But it focusses not on a random sample but on prisoners whom there is good reason to suspect are innocent.

Project II: Low volume toilets are supposed to save water. They also, at least in my experience, tend to get stopped up more than ordinary toilets. Someone should do a statistical study relating the fraction of low volume toilets in an area to both water usage and expenditures on plumbers. Assuming water is saved--the process of getting a toilet unstopped can require multiple flushes--it might turn out to be very expensive water.

And yes, this was written after yet another session applying a plumber's snake to a low flow toilet.


Anonymous said...

My understanding is that newer low-flush toilets are much better than the first generation.

Take this for what it is worth: either an important variable to keep in mind for the study, or a thought to look at the possibility of getting a new toilet.

My toilet doesn't flush at all, what do I know.

Sanctimonious Hypocrite said...

Even if they work as designed (and in my experience they do not), I wonder about the unintended consequences of changing the proportion of solid in the effluent.

Todd Mitchell said...

David, the problem with obtaining "false positive" data concerning criminal convictions is that 90% of all criminal convictions in this country are obtained via plea bargaining. You'd have to differentiate between plea bargaining and trial convictions.

How many people are "railroaded" into false confessions through the plea bargaining system would certainly be an interesting study. It most definitely disadvantages the poor, since the bail/bond system practically forces the poor to cop pleas in order to get out.

Anonymous said...

In response to vis: Even where there is no ecological reason to save water there is usually an economic one. I believe that denser effluent is cheaper to process, as most of the processing occurs after the initial filtering of solids, and mainly concerns cleaning the water enough to return it to the environment.

David Friedman said...

Tod Mitchell points out that criminal convictions are largely obtained through plea bargaining. I don't see the relevance of that to my project. I want to know what percentage of those convicted were innocent--however they were convicted.

It would, of course, also be interesting to know whether the percentage is substantially different in the case of convictions via plea bargaining.

Anonymous said...

I would imagine you would also want to know the rate of false negatives. A similar study might be possible though still more difficult. In the same research-friendly jurisdictions, one could perhaps look at acquittals in cases where both the original physical evidence and a DNA sample from the defendant survives.

One might suppose that acquitted but actually guilty defendants would be unlikely to cooperate. Though they cannot face double jeopardy, they might fear being tried by a separate sovereign, or a civil suit. However, some deceased, acquitted defendants may have left viable sources of DNA samples or descendants whose DNA could be used to help establish guilt or innocence.

In precisely the same manner as your proposed false positive estimate, such a sample could be used to approximately gauge the rate of false negatives.

One the low-flow toilet issue: New York City has enjoyed a precipitous decline in per capita water use since the high in the 1980s. If I recall correctly, both low flow regulations and water use fees were introduced around the same time, so this probably complicates the situation too much to allow you to draw any conclusions from the data without much more information, but perhaps somehow the contribution from the regulations could be isolated.

Here's the web page with the use statistics:

Lippard said...

I believe one of Gordon Tullock's books reports on an experiment which used two independent juries who were presented the same evidence and given the same instructions to gauge the reliability on the basis of how often they disagreed (i.e., if 1/4 of cases lead to disagreement, then at least 1/8 of jury decisions are wrong).

But that methodology doesn't tell you *which* juries get it wrong, or why.

(Apparently Richard Dawkins' book, _A Devil's Chaplain_, suggests such an experiment to determine whether Trial by Judge is superior to Trial by Jury, criticized by Tyler Cowen here:

Jack Maturin said...

Hi David,

If the rate turned out to be one in a thousand, the case for the death penalty (and many other things) would be a lot stronger than if it turned out to be one in three.

I'm sorry, but even if the rate was one in a million, that means the state, or whoever is putting someone to death in a murder conviction, are themselves murderers at least one time in a million. Murder is wrong, therefore this system of justice is wrong. Therefore, there is never any justification for the death penalty, unless perhaps you could prove that in every single case the verdict would be correct. In the world I live in, this is impossible, therefore the death penalty is ruled out. My preferred route in these difficult cases is that of a mixture of compensation payment to the agrieved family and/or exclusion from normal society for the convicted person. As society would be entirely private, in an ideal case, there would be no way that this person could come back into society to wreak the same potential havoc they may have caused previously, as they would be excluded by the reputable insurance agencies which would have replaced the security functions of the modern monopoly state. The one-in-a-million falsely convicted person would still be alive and able to make some kind of way in the world. Even as a falsely convicted murderer, who refused to compensate the family because they were genuinely innocent, I'm sure most such people would prefer outcast societal exclusion to being wrongly killed by a security agency. Obviously, self-defence killing would be acceptable.


Anonymous said...

Mr. Maturin's case would be stronger if he used words more carefully. "Murder" is the UNLAWFUL killing of someone. Therefore, by definition, someone who is executed as a result of the legal process was not murdered. He may have been killed incorrectly, it may be unjust, but it ain't murder. Sorry.

Further, it is also unjust to imprison, or fine, someone who was not actually guilty. You can argue that the death penalty is more final, and therefore more unjust. But either you must evaluate the relative cost of life in prison or you have to come up with some way to quantify the fraction of those wrongly getting other penalties who get some compensation/restitution, or are at least publicly exonerated. Which, of course, is something that could be developed as a result of David's proposed study.

My personal objection to invoking the death penalty, as currently implemented in California, is that it is uneconomic for the tax-payers. It costs more, from conviction to eventual execution, than it would to simply imprison someone without parole for the rest of their life. (The Prison Guards' Union is working hard to change that, but for the moment....)

Anonymous said...

I was in Queensland Australia last spring and took note of their dual flush toilets. These toilets had a high volume flush and a low volume flush, so you had your choice. While this may not be a good choice for the indecisive among us, it could be a good option for the rest of us that want to conserve water and are frustrated with ineffective low volume toilets. Of course it can be rather inconvenient travelling to Australia every time you have to go...

Related Note: I didn't determine if the flushed water spun in the other direction because it didn't spin at all the few times I checked. Oh well, I guess I'll have to go back and do more toilet research.

Anonymous said...

nice scatalogical beginning to your blog, professor...YUCK!

Anonymous said...

I've never had to flush more than twice after relieving a blockage, which don't occur more than once every few months. Estimating 10 flushes a day (there are five of us), that's roughly .2% extra flushes, insignificant compared to the water saved by the lower flow.

Andrew said...

In Japan, the small/large dual flush toilets are also very common.

Anyway, the key to handling a low-flush toilet is to use "courtesy flushes" frequently when dealing with solid waste.

The common low-flush toilet cannot typically handle as much solid waste volume as older toilets, therefore they need to be flushed more frequently when you're...losing lots of weight.

Some better toilets employ vacuum pressure and other methods to help move more waste, but it's really not difficult to make a few extra flushes occasionally to avoid the toilet snake or plunger.

Low-flow toilets are not only good in areas where water is scarce and expensive; they also help greatly in areas where populations are growing and sewage treatment plants and other water systems are hitting their limits.

Andrew (learned nothing about this in five years of water studies/civil engineering coursework)

Anonymous said...

Way OT: Is there an RSS or Atom feed on the site? Would love to see one...

GodlessZone said...

Just returned to the US after being gone for a long time and I was horrified at the toilets. I stayed in one apartment temporarily before moving to my other one and opened a business as well. In all three the toilets have a tendency to stop up as the water volume isn't strong enough to carry things away. The maintenance guy at the complex said to flush several times. So a typical use requires at least two flushes on my part and sometimes three because things float back up. And these are new toilets so should be the latest. So I end up using far more water than I did overseas where the toilets were not low flush.

Paul O'Brien said...

I'm not sure David's "false positives" analysis would tell us very much.

All it would identify is those cases where a conviction was only obtained because DNA evidence was not yet available. While that's historically interesting, it doesn't tell us anything about the rate of false positives in the PRESENT criminal justice system, where DNA evidence IS available.

On top of that, DNA evidence only goes to identification, which is only one factor giving rise to false positive convictions. For example, in rape trials, the defence is usually consent. DNA evidence doesn't take you very far when the accused admits he was there.

There are also serious practical difficulties with the experiment. The samples in question date from a pre-DNA era, and won't have been handled in accordance with modern procedures which are necessary to avoid the risk of contamination etc. And that's before you consider the conditions in which they've been stored in the intervening years. Oh, and on top of that, you're really only working with those crimes serious enough to justify a proper crime scene analysis in the first place, which confines you to an unrepresentative minority of crimes straight off the bat.

So you're relying on those cases where (a) DNA evidence could potentially have made a difference; and (b) DNA samples were obtained at the time, and have been stored ever since, in a way that doesn't prejudice their reliability.

I'd be very surprised to learn that a significant number of such cases existed. And those that do, I suspect, will be very unusual and unrepresentative of the wider criminal justice system. I realise David acknowledges this point in the original post, but I think he vastly underestimates the scale of the problem.

Anonymous said...

David, I would love to read your thoughts on Christianity, and why you are an athiest.

Anonymous said...

I have the same problem with the low-flow toilets, but there is another variable that influences the economies observed far more than the two to three flushes a solid elimination may take: the number of times one flushes only fluid waste for every time one flushes solids.

As I get older, I notice that the fluid/solid ratio is becoming almost 5/1. Therefore, even if every solid waste must be flushed 3 times, the water saved on the fluid-only flushes still offsets the extra water used for the solid-waste flush-fest.

Gary McGath said...

As Paul O'Brien notes, the DNA test wouldn't get very far in determining guilt or innocence over a broad spectrum of cases; if it did, we'd now have a general solution to the problem of false convictions. In many cases, the accused was definitely present but may or may not have committed a crime. In others, the accused may have committed a crime (such as hiring someone to commit a murder) without being at the crime scene. In other cases, both presence and action may be undisputed, but disputed circumstances, such as a claim of self-defense, may affect whether the action is a crime.

Anonymous said...

I just suggested the same basic idea (finding out about death penalty correctness rates via DNA) on the Becker/Posner blog. (And I thought I was being original....)

For all the imperfections of only using the DNA evidence to check correctness, I can't see a much better way to check. And it's really important to work out what fraction of convictions (and plea bargains) are false, since it tells us a lot about where we should be putting resources in the criminal justice system.


David Friedman said...

Jack M. writes:

"I'm sorry, but even if the rate was one in a million, that means the state, or whoever is putting someone to death in a murder conviction, are themselves murderers at least one time in a million. Murder is wrong, therefore this system of justice is wrong. Therefore, there is never any justification for the death penalty, unless perhaps you could prove that in every single case the verdict would be correct."


"Obviously, self-defence killing would be acceptable."

Why obviously? If you are willing to kill in what you believe to be self defense, there is some probability--say one in a million--that you will be wrong, that you will kill an innocent man who, due to some bizarre and unlikely confusion, you think is attacking you. If your self defense killing is with a firearm, there is some probability, probably a good deal more than one in a million, that the bullet will end up hitting and killing someone else.

Going beyond the case of criminal justice, if you fly a small plane there is some probability--say one in a million--of some accident that will result in your crashing and killing an innocent person below.

Why do you count the accidental killing of someone you are not entitled to kill as murder only when it takes the form of a mistaken execution?

David Friedman said...

D. I. writes:

"my friend who is a crminal trial lawyer says that in his fifteen year experience, it is almost impossible to secure a wrongful conviction."

The evidence of innocence projects using DNA testing suggests that your friend is mistaken. It may depend in part on whether the defendant can afford to hire a competent criminal trial lawyer. I believe that something like half of all felony defendants are being defended by a lawyer provided by the state--and while some, I'm sure, are doing a competent job, it's pretty clear that many are not.

You might want to use Google Groups to search Usenet for posts by Keith Lynch. He is someone who was convicted, a very long time ago, of a crime that he claims, and I and (I think) most people who have looked at his account believe, he did not commit. He has discussed it online at some length.

David Friedman said...

Anonymous writes:

"I just suggested the same basic idea (finding out about death penalty correctness rates via DNA) on the Becker/Posner blog."

Posner at least has heard it before--I was at the U of C Law School when I started trying to push the idea. It was in part inspired by the difference between his view of the subject and that of Steve Schulhofer, a colleague with whom I was writing an article arguing for a voucher system for providing attornies for indigent felony defendants.

Steve was considered on the left of the U of C faculty, so there was some discussion among colleagues as to which of us was subverting the other. I thought the question was settled when we gave a joint presentation on the article--and Steve Schulhofer ended up lecturing Richard Posner on the virtues of the free market.

Anonymous said...

What about the hidden costs of having s*** stick to the side of the bowl?

Anonymous said...

Do not use Hashaway fixtures, I tell you what.

sierra said...

Andrew Ferguson has
a column out about low-flow toilets.

Anonymous said...

about project II, there is a water-saving toilet capable to
use only a little volume of water as small as 1 liter for flushing one
round of urinating and still keeping the toilet very clean after flushing. They also dont get stopped up like the low bowl toilets.

The Skip Bureau said...

I *HATE* low flush toilets. They are a symbol of everything wrong with this world. That and low flow shower heads. I went out and spent an enormous sum (for me) on a Kohler low flush toilet that promised to not get stopped up. It stops up constantly. Since it's got an odd shaped bowl, it is really hard to clear when it clogs as well, as a standard round plunger won't seal to it.

Also, my house, as with most older houses, has four inch sewer pipes. It took a while to figure out that we had to run a lot of extra water, normally in the tub, to make sure the pipes stayed clean, or the solid waste would build up in the pipes. I don't have a cite, but I read somewhere about cities in Germany having to open up fire hydrants to keep their municipal sewers clean.

The funny thing to me is the idea we're at all running out of water. There's very few places on this earth that are that far from the ocean. We have more water than any other natural resource of any import. What we're running out of is cheap water. Reverse osmosis simply isn't that expensive, but it does require energy, and the left hates the use of energy except as necessary to carry their important butts around to major conference on making the rest of us use less energy.