You claim that conviction of innocent people, in Israel and worldwide, is far more common than we would like to think.
Definitely, convicting innocent people is very common, although the criminal law system is still entrenched in the view that conviction of an innocent person is a very rare occurrence. Until recently the public believed that, too, but then came the Innocence Project [a nonprofit that launched a campaign in 1992 to exonerate wrongfully convicted people] in the United States and changed everything: Genetic comparisons from samples taken from crime scenes, and from the bodies of prisoners who for years have been claiming their innocence, are proving that the system actually convicts many innocent people.
What sort of numbers does the Innocence Project cite?
Calculations indicate that in the United States we are talking about at least 5 percent of those convicted of the most serious crimes – rape and murder. My estimate is that when it comes to less serious crimes, the system is less careful, and its rate of error is far higher.
In a gross estimate, how many innocent people are now serving prison sentences in Israel?
There are about 20,000 prisoners here, and I estimate that about 1,000 of them have been wrongly imprisoned. And I believe that if this estimate is mistaken, it’s because it’s too conservative.
On what exactly is this calculation based?
On studies conducted in the wake of the efforts by the Innocence Project – comparing those acquitted with the total number of federal convictions. My own calculations support this figure, and may suggest an even higher one here. I believe that when it comes to crimes that are less serious than murder and rape, at least 10 percent have been wrongly convicted.
You claim that the criminal investigation actually begins from the end: The police have a suspect, and the case is solved in a manner that supports the initial assumption rather than vice versa.
I call the belief underlying that entire concept “the guilt of the suspect”: He is assumed to be guilty, and the investigators concentrate on him and pressure him to confess.
The investigation focuses on the search for evidence that supports his guilt rather than on other directions. In general human nature tends to adopt things that suit our beliefs, and to reject the rest.
Correct. There’s a cognitive bias that dramatically affects criminal investigations. It’s like putting blinkers on a horse so that he’ll only walk straight ahead. We look for evidence, and find it. And, unfortunately, the Supreme Court’s guidelines states that a person can be accused on the basis of his confession alone.
Is no other evidence required?
There’s a requirement for what the Supreme Court calls “something additional,” but it can be something that’s “featherweight.” What served as the basis for the initial suspicion usually serves as the “something additional,” so that in effect the confession itself tips the balance. If the statistics were to suggest that almost 100 percent of confessions are true, there would be no problem, but studies actually prove the opposite: People confess even to things they didn’t do. In the past people tended to think that this could happen only to weak people – minors, the mentally retarded, etc. – but today, we know that it can happen to anyone. Even entirely rational people can make a false confession.
Studies show that when the person under investigation is falsely led to believe that there is very strong evidence against him, he is likely to give up. He will think that nobody will believe him in any case, and in such a situation he will grab the opportunity to make a false confession – because the investigators tell him that they’ll have him indicted on lesser charges and help him get a lighter punishment. These are empty promises for the most part, of course. In the case of Zadorov, he was sentenced to life imprisonment. [Roman Zadorov was convicted in 2010 of killing teenager Tair Rada four years earlier.]
You write in your book about a case in which a housekeeper was persuaded to confess to stealing 400 shekels ($114) from her employers’ house, one reason being that she was arrested by surpriseand also had a nursing infant. I thought about the fact that an ordinary, innocent person is more prone to confess than a guilty one. He doesn’t fully understand the consequences of his confession or the significance of the crime of which he is accused. He may never been in a prison cell, and will do anything in order to avoid it. And if he is told “Confess, and everything will be over and you can go home” – he will simply confess.
Exactly. And not only do they pressure him to confess, they also say there’s no chance that he will be acquitted, so he’s even more tempted by the offer. We have to change the law so that it won’t be possible to convict someone based on a confession alone, and that there will have to be other decisive and weighty evidence: independent, objective proof that points to the person guilty of the crime.
There are high-profile cases – the murder of Tair Rada, and the 2009 murders in the Bar Noar LGBT club in Tel Aviv – but you maintain that in most cases of mistaken convictions, there is at play a principle you’ve dubbed the “hidden accidents principle” of criminal law. In other words, when an innocent person is convicted, almost nobody knows about it.
When, for example, there’s a disaster involving airplanes, everyone knows about it. But in criminal law, we don’t hear about it; the entire system is built to perpetuate the conviction. Ostensibly, there are mechanisms that monitor mistakes, but they are totally useless. An appeal is totally useless. There’s a legal rule whereby the Supreme Court doesn’t need to conduct an in-depth examination of the facts as determined by the court that initially dealt with a given case, and doesn’t intervene in its assessments regarding the credibility of witnesses. In an appeal, the Supreme Court justices deal only with legal errors, and in my opinion, most of the convictions of innocent people don’t stem from such errors, but from mistaken factual rulings. For example, attributing greater importance than they should to a specific piece of evidence, or believing the wrong witness. Studies show that judges, like the rest of us, don’t know who’s telling the truth and who’s lying.
How many appeals are submitted annually? And how many are actually accepted?
Almost all the appeals are rejected, but the picture is one-sided even before that, during the initial trial itself. According to data the Central Bureau of Statistics used to publish until recently, the percentage of convictions in serious crimes is 99 percent. Two or three years ago, the Courts Administration asked the CBS to stop publishing this figure, because it damages the image of the law-enforcement system and is misleading because it also includes plea bargains. But in my opinion, it’s not misleading, it should also include plea bargains, because 90 percent of cases today end in such agreements. There’s no question that the fact that the proportion of convictions is so high should interest the public.
What are the chances of acquittal on appeal for an innocent person serving a prison sentence?
The chances of acquittal are nil, even if he is represented by the strongest and best-connected attorney in the country. Virtually all appeals are rejected. Presumably, it is possible to request a retrial, but it’s impossible to get one. In the past decade, the average has been one request granted per year, and even then the chances are that the person will be convicted again after the retrial. One case a year out of hundreds of requests and thousands of innocent people – it creates the impression that the system never makes a mistake.
What’s the role of the judge in all this?
Judges are also captives of the concept of the guilt of the accused. They get a case that is directed entirely toward conviction of one person, with all the evidence against him. Another problem is that judges are not aware of the new studies that demonstrate the extent to which the legal system is prone to mistakes, and the extent to which the evidence that it tends to rely on is insufficient.
What’s surprising about the picture you paint is that the system is not aware of and doesn’t monitor its inherent shortcomings.
In a system based on decisions of human beings, there will always be mistakes. But there are many things that could be corrected but are not. Psychologists, for example, have been writing for decades that [the very idea of] police lineups are a mistake. It’s possible to prevent such errors by very simple means that don’t even cost anything: By explaining to the witness that the person he is supposed to identify – whom he ostensibly saw at the scene of the crime – is not necessarily among the people in the lineup.
When you don’t do that, the witness will choose, by process of elimination, the person who is most similar to the one he saw at the scene. Another recommendation is to arrange the lineup sequentially – to show him [the witness] one person at a time and not with others, and to ask whether this is the man he saw. That will preclude the tendency to choose the one who is most similar.
How is the system supposed to monitor itself? What mechanisms exist?
Ostensibly there are such mechanisms. The prosecution is supposed to supervise the police, and the judge is supposed to examine everything as he sees it, but the tendency is to rely on whomever handled the case before you. The prosecutor always relies on the police investigator.
There’s no one in this chain who can be expected to go against the tide and offer critical thinking.
Correct. Once I taught a class on the subject of the conviction of innocent people, and the students suggested adding to the police investigative team a person who can act as the devil’s advocate – just as in the army and intelligence, there are people whose job is to suggest another way of thinking and who will seek evidence that points in different directions. That’s a wonderful idea.
How do you distinguish between scientific evidence and evidence that is only perceived as such? You write in your book that even a fingerprint is not necessarily reliable evidence.
Fingerprints and DNA are the most accurate evidence we have today, but as in any scientific examination, here too there may be mistakes. Every time the lab makes a mistake, an innocent man is sent to jail.
How can you make a mistake with a fingerprint or a DNA sample?
A fingerprint taken at the scene of a crime is not like the one that is carefully taken from us, say, on entering the United States. A fingerprint at the scene is usually incomplete; that’s why of course there are mistakes. As far as DNA is concerned, the samples can be ruined at any stage; there are errors because of a testing system that is not totally accurate. Besides, there can be all kinds of explanations for a [particular] fingerprint being at the scene. The suspect may have been there or touched an object that was brought there. Or maybe the [forensic] expert erred in matching the fingerprint found at the scene and that of the suspect.
In Madrid, after the terror attacks that took place there [in 2004], the police found a bag of detonators with a fingerprint on it. They sent it to police forces worldwide, and the FBI scanned its databases and arrived at an American lawyer named Brandon Mayfield. Four different experts determined it was his fingerprint, and he was arrested. Fortunately for him, the Spanish authorities found another man with a better-matched fingerprint.
In the investigation conducted to understand how that mistake was made, they found that when a fingerprint is scanned in a computerized system containing millions of fingerprints, the system finds the 20 most similar ones, but these can mislead and confuse experts. They also found that sometimes police investigators leak to the [forensic] experts who may work in their offices information that biases their thinking.
Let’s talk about the Yanshuf [the Israeli breathalyzer]. You write that the Israel Police purchased outdated models and suggest that over half the drivers charged by means of this equipment of drunk driving were not inebriated.
That’s a calculation I did with Dr. Mordechai Halpert, based on police data. Since nobody must approve or oversee usage of the Yanshuf, a person can be sent to prison based on a test whose accuracy has not been fully proven. The courts don’t require the police to present figures from the manufacturer regarding the percentage of false positive results.
The testing kit can show a positive result even if the person tested is not drunk?
Yes. Now let’s assume, in a cautious estimate, that 1 percent of results are in error. That’s a number based on the field of drug testing – a field where the results are considered very accurate, and where data show there’s a margin of error of 1 percent in the lab. If that means we have 1 percent of false positives – so what? It means that 99 percent of those caught really are drunk, and the 1 percent of innocent people who are convicted is a price that we as a society can and should pay.
But in the field of medical diagnosis – and for some reason this hasn’t crossed the line to criminal diagnosis – it’s known that if you examine a population that is not at risk, the percentage of error will be much higher. That’s why they don’t do large-scale testing for AIDs carriers: It will sow panic. Only if you test a population that has a high percentage of the illness in the first place, does the testing become more accurate.
But the police check drunk driving randomly.
They stand by the side of the road and arrest drivers. They published figures to the effect that of half a million people who were tested, 11,000 drunk drivers were found. Now, in half a million people who are not drunk, if we take into account a 1 percent testing error – we get 5,000 people with a false positive. That’s an astonishing figure, because if we do a basic calculation and round out the numbers – if 11,000 were said to be drunk [after being tested by police], about half weren’t truly drunk. But they too get convicted. There’s no chance of being acquitted after such a test.
We have to demand more accurate equipment, and test people in a nonrandom manner – not simply to place a trap at the side of the road, but to stop only those about whom there is a previous suspicion – those who are driving fast or zigzagging. Only under such conditions is the test really accurate.
That also demonstrates the extent to which the built-in distortion in the system can also affect the lives of ordinary people.
Precisely. It happens every day. The system is not interested in changing, it doesn’t want to hear criticism, most of the people running it are not even aware of how sloppy it is. I wrote my book for the general public, so people will understand and know how those who presume to represent them operate.