Exonerations with not even a crime: how many more should there be?

The number of people exonerated after they were falsely convicted of crimes in the US has reached an historic high, with 87 walking free last year.

A new report from the National Registry of Exonerations finds that almost a third of the people in 2013’s unprecedented crop of exonerations were convicted in cases in which, in fact, no crime was committed – a record-breaking number in itself. Some 22 men and five women were given sentences ranging from probation to life, yet when their convictions were investigated, they were not only found to be innocent, but it was discovered that no offence had occurred in the first place.

There is further information here.


Just one more reason the government should stop being involved in law enforcement. Free market competition would allow for private enforcement cartels which would be constrained by competition and the profit motive to never make a mistake.

This isn't effective satire of either statists or any particular branch of anarchism. 0/10

And then you have this: http://www.vagazette.com/news/va-vg-16yearold-james-city-girl-charged-20140205,0,7353878.story

What would complete the trifecta of lunacy would be for her to be charged as an adult (only natural, of course, considering the seriousness of the charges).

If you consider all of the vice crimes and questionable white collar crimes, you've probably only got ~50% of convictions where a crime actually occurred in the first place!

On the other side of the ledger, we have financial firms committing crimes hourly and no one is ever brought to trial, much less put in prison.

Nothing is more criminal than making a profit!

Another '0/10' comment.

Try this on:

"Nothing is more criminal than being a bankster!"

22 people out of how many incarcerated? Two million? What false positive error rate are we talking about here? It's sad yes, but I can't see why people are so surprised. Did people really assume that our justice system or trials are absolutely error free? Is it even possible to design such a system? Sure we ought to try and do the best we can but surely 22 in 2 million is not a ghastly, spectacularly high error rate?

They obviously did not review every single case. This is a subset of a subset, and its significance does not correspond to the low absolute number. For example, if those 22 had the same prosecutor, it would be a serious implication against him/her.

The problem is that the 22 people is not a upper bound but a lower bound. And the factors leading to conviction were not just random flukes like the real criminal looking like the wrongly convicted, but confessing at "end of a 27-hour interrogation – during which she said she was threatened, pushed, called names and denied food and water"

Yes, but so what is the estimated false positives rate? Isn't that the crucial question.

22 / 2,000,000 does not sound too bad. Your lower bound critique would apply even had they exonerated just one person. My point is that, by itself, there's very little alarming in this report unless one was so naive that one assumed justice was a 100% accuracy undertaking.

The alarming part is that the police are using harsh interrogation to extract false confessions, and the historical reluctance to adopt measures such as recording interrogations to ensure that such abuses don't take place.

For a very rough estimate of actual false positives, I would go with 8700.
The article mentions that Dallas county is rare in keeping biological samples indefinitely. Also its DA formed a Conviction Integrity Unit which is uncommon. So lets 10 times as many people would be exonerated if the cases were investigated. And 10 times that are innocent but would not meet the standards of exoneration after conviction. And the total number exonerated was 87 not 22. 22 was a subset where there was no actual crime but an accidental event was considered a crime. So we have 87 * 100.

You may still consider less than 1% low, but would you accept 0.1% airplane flights crashing an acceptable rate when there are steps that can be taken to improve the results?

The title of the post includes, "How Many More Should There Be?"

There should be a lot more but I think we ought to look at this more optimistically: It's great that we are finding out and exonerating them. Shouldn't we expect this though? With time, as technology improves our ability to identify past false positives naturally improves too.

Isn't this a feature and not a bug?

The pessimistic case is that many (perhaps even most) serious criminal convictions are garnered without any 'testable' evidence being gathered at all - or if it is gathered, it may be irrelevant to the criminal issues in question.

Point being: If people were convicted despite the presence of physical evidence which attested to their innocence, how many false convictions are there in which there is no evidence which can attest to their innocence?

Circumstantial evidence *is* sufficient to convict, contrary to popular belief. I would SWAG that wrongful convictions are quite massive, perhaps on the order of 10-20% - and many of those will have plead out!

The real false positives ratio is going to be hard to estimate. Unless someone took a purely random sample & tried to analyse cases from that set & exonerate.

I really doubt that the ratio is as high as you estimate especially for the more serious sentences. I don't think juries are that stupid nor that callous.

>The real false positives ratio is going to be hard to estimate. Unless someone took a purely random sample & tried to analyse cases from that set & exonerate.


Except I'd set the bar a bit lower - exonerations are difficult. Collecting evidence to establish reasonable doubt would be sufficient for our purposes, no?

@Alexei Sadeski

Agreed. But as judged by who.


Indeed. By who and using what interpretation of "reasonable doubt".

It's worth noting that the spate of DNA exonerations represents a small subset of wrongly-incarcerated people. These are merely people for whom redeeming evidence could be located.

"Cases where no crime occurred" is a really misleading designation. If there's a homicide, say, and the dispute is over whether the defendant killed the victim accidentally or deliberately, ultimately the only way to determine whether a crime occurred is to hold a trial. Ultimately, it's tautological in such a situation to ask whether "a crime occurred;" the meaningful question is "were there circumstances that reasonably cause suspicion of a crime?" I'm guessing that in all of these cases, there were such circumstances, and we now just have reason to conclude that the jury came down the wrong way, and not that there was never any basis to suspect a crime at all.

And while the 87 false positives are of course a terrible injustice for those involved, that number is a drop in the bucket as against the total number of criminal convictions out there.

The "no crime" cases included instances of suspected homicide which turned out not to be homicide at all.

Or, in other words, RTFA.

The homicide cases are precisely what I'm talking about. Either you're agreeing with me in an unpleasant way, or you should look again at what I've written.

The homicide cases are what I'm talking about as well!

These are cases in which a homicide was reported, but no homicide took place!

My guess is that almost all of these 87 people were pretty bad dudes to begin with. The over-under betting line for how many committed crimes of equal or greater magnitude in another instance would be 70 or more. This type of thing happens when the police and DA are out to get people who any rational Bayesian would conclude is a hardcore criminal, but they can't pin any specific instance on beyond a reasonable doubt. ("Okay Mr. Smith, it's one thing to coincidentally be found four blocks from a murder scene with traces of gunpowder. It's another thing for it to happen twelve different times"). Ask any cop and they'll tell you that virtually every person who gets to the indictment stage is guilty, if not for that crime then definitely for another. This is a big reason law enforcement supports drug and gun laws, because it's a lot easier to pop a gangsta carrying a bunch of dope or a piece than it is to prove he involved in a crime.

Which suggests a law and order conservative-libertarian compromise. End the drug war in exchange for reversing Miranda and other Warren Court decisions that tie the hand of police.

Now *that* is good satire.


"My guess is that almost all of these 87 people were pretty bad dudes to begin with."

This kind of guessing is described in contemporary psychology as "just world phenomenon", a known cognitive bias.

Except juries don't really detect guilt or not. They don't even look at the evidence all that much.

They are really just unpopularity contests. If you're really unlikeable they will find you guilty.

And "pretty bad dudes" tend to be unpopular.

Countries like Japan and Singapore with near 100% conviction rates also have near zero crime rates. That suggests that the sizable majority of those being acquitted are indeed criminals. As David Simon said in his book about the Baltimore homicide unit: "To a jury, any doubt is reasonable."

Overall the balance of evidence suggests that the American justice system is much more tilted towards acquitting too many guilty men than convicting too many innocent ones. From a utilitarian standpoint most violent criminals are highly repeatable, and most false convictions are for people who are likely to have committed other crimes (the police just don't go rounding up middle-class housewives when there's a murder, they look for known gangsters and lowlives). Every false acquittal probably results in at least two or three more crimes of the same type. Whereas probably 33% or less of false convictions are for people who are truly innocent of any crime.

From a utilitarian standpoint we want to minimize the suffering of innocents. That suggests a calibrated ratio of 6 false convictions for every false acquittal. Right now the justice system is probably closer to the inverse of that ratio. Long story short, we should be convicting a lot more defendants.

>Countries like Japan and Singapore with near 100% conviction rates also have near zero crime rates. That suggests that the sizable majority of those being acquitted are indeed criminals.

Ridiculous leap in logic.

If you want a very clear natural experiment, the Miranda decision offers an excellent example. First Miranda without a doubt absolutely reduced conviction rates (http://content.csbs.utah.edu/~fowles/STANFIN.pdf). Miranda can be seen as a shift in the tradeoff between false acquittals and false convictions, in favor of the former. Next the previously stable murder rate begins a meteoric upward trajectory starting in 1966 the year of Miranda (http://www.infoplease.com/ipa/A0873729.html).

The balance of evidence is clear, Miranda and other Warren court decision, significantly restrained the ability of law enforcement to attain convictions on violent criminals. The marginal acquittals and non-prosecutions must be overwhelmingly made up of criminals. If police were primarily using their pre-Miranda powers to railroad innocent men, you wouldn't see such a drastic impact on crime rates. Middle-class civil libertarians generally support restricting police power because they fear that the police are jack-booted storms troopers. The vast majority of middle-class people are much more likely to have their personal liberty violated by criminals than they are police, even if they are recreational, but well-behaved, drug users.

Why is that a ridiculous leap in logic? One of the factors that will affect crime rates is the chance of being caught and punished. If the prosecution has the whip hand to the extent it gets roughly 100% convictions, then either criminals know they will not be convicted or they know they will be. If the former, they will commit more crime. If the latter, they will commit less. In Japan and Singapore they commit very few crimes indeed.

And this is recent. Singapore was not particularly law abiding in 1960.

Doug, you appear to be altogether too comfortable with reaching strong conclusions in the absence any information to corroborate your assumptions. I certainly hope you never have the pleasure to serve jury duty.

It's also the case that federal prosecutors in the U.S. have an extremely high conviction rate. As with Japan, it seems the reason is that prosecutors will simply drop any case that doesn't look winnable. Conviction rate can be a very misleading measure of how tough the justice system is; what would be much more interesting would be to look at how many people who are seriously investigated for or arrested for a crime wind up being convicted. The existence of Yakuza outfits that operate more or less openly also suggests an important caveat on claims of Japan being a low-crime society.

Plea bargains must also contribute to a high conviction rate.

Well going on the experience of Illinois, where half their death row inmates turned out to be innocent, I would say a lot more than 87.

The jury system is mainly random. Added to that the variations in lawyering. You have random piled on random. Of course it is not going to work well. But that is the compromise the Western legal system made: we got rid of torture and the certainty that it could provide - and only it can provide - in exchange for a system which we pretend is rational and good at finding guilty people, but does not have to inflict any actual violence on anyone.

Apart from the prison rape I suppose.

I give this one a 7/10. High plausibility and good grammar, but doesn't quite sell the humor.

I wish that I could see interviews with the jurors and prosecutors in these cases.

There are tons out there - if not necessarily these particular cases (they're recent, yes?) at least other examples. They're fascinating - particularly the ones with detectives who "heard" false confessions!

Here you go:


Thank you! There's a lot of interesting information there.

Sadly, no interviews with the actual prosecutor and detective who extracted the false confession. It would of course be legally unwise for them to give one, but still.. it'd be morally satisfying to hear what the individuals personally responsible for the miscarriage have to say.

I think that I've seen them before somewhere (interview of prosecutor & detective) - but can't for the life of me find them. Maybe I dreamt it up!

For symmetry purposes, shouldn't there also be a registry of people who have committed crimes but were never caught or convicted?

How many innocent lives are you willing to destroy in order to nab some bad guy in the absence of convincing evidence?

Otherwise asked: How many bad guy are you willing to let walk in order to save one innocent person from a life in prison?

Personally, I am comfortable with a system which requires very strong proof and therefore lets many bad guys walk free in order to protect a very small (but not small enough) number of innocent people.


Let's start the list off with all of the narcotics officers. Guilty of wrongful imprisonment, and none of them have been charged yet.

I guess neither of you have been victims of a violent crime

What's violent crime got to do with mass false imprisonment of non violent "criminals"?

Comments for this post are closed