Schneier on Security
A blog covering security and security technology.
« Security Theater on the Boston T |
| High School Teacher Assigns Movie-Plot Threat Contest Problem »
August 30, 2010
Misidentification and the Court System
How do most wrongful convictions come about?
The primary cause is mistaken identification. Actually, I wouldn't call it mistaken identification; I'd call it misidentification, because you often find that there was some sort of misconduct by the police. In a lot of cases, the victim initially wasn't so sure. And then the police say, "Oh, no, you got the right guy. In fact, we think he's done two others that we just couldn't get him for." Or: "Yup, that's who we thought it was all along, great call."
It's disturbing that misidentifications still play such a large role in wrongful convictions, given that we've known about the fallibility of eyewitness testimony for over a century.
In terms of empirical studies, that's right. And 30 or 40 years ago, the Supreme Court acknowledged that eyewitness identification is problematic and can lead to wrongful convictions. The trouble is, it instructed lower courts to determine the validity of eyewitness testimony based on a lot of factors that are irrelevant, like the certainty of the witness. But the certainty you express [in court] a year and half later has nothing to do with how certain you felt two days after the event when you picked the photograph out of the array or picked the guy out of the lineup. You become more certain over time; that's just the way the mind works. With the passage of time, your story becomes your reality. You get wedded to your own version.
And the police participate in this. They show the victim the same picture again and again to prepare her for the trial. So at a certain point you're no longer remembering the event; you're just remembering this picture that you keep seeing.
Posted on August 30, 2010 at 12:05 PM
• 31 Comments
To receive these entries once a month by e-mail, sign up for the Crypto-Gram Newsletter.
@: "you often find that there was some sort of misconduct by the police. In a lot of cases, the victim initially wasn't so sure. And then the police say, "Oh, no, you got the right guy. In fact, we think he's done two others that we just couldn't get him for."
My father was attacked and robbed in Mississippi about 10 years ago, and the police caught the guy a few moments later. My sister was a witness and picked him out of a lineup that night. When she said she was 90% sure it was him, the police said, appropriately, "that's not good enough."
Fortunately, the man confessed and plead guilty. He knew they had him based on evidence. But the fact remains, they were appropriate in taking my sister's doubts into account and in no way lead her into the identification.
I wish it was always that way.
Misidentification (and police misconduct) put Steve Titus in jail for a couple of years before a Seattle Times reported essentially proved that Steve was innocent. Unfortunately, Steve died before he could get his life back. See http://seattletimes.nwsource.com/news/local/... and the Wikipedia article.
What disturbs me most is that our investigative, arrest, and trial systems have positive incentives for misidentification that often exceed the incentives for identifying, arresting, and convicting the actual criminal. There are almost no disincentives to misidentification, misuse of evidence, and other malfeasances to gain convictions. Police rarely are punished for arresting someone based on flimsy evidence, district attorneys are rarely punished for trying persons based on flimsy evidence, and judges are never punished for not declaring mistrials based on flimsy, flawed, or even manufactured evidence.
I like most of our Constitution, but the sections related to criminal justice need major revisions.
Sadly, Dr T understates the case. There is zero incentive to get it right. All only thing that pays is to 'git her done'.
If police and prosecutors were sent to prison for a decade or more each time one of them lid or covered up facts at a trial. I'll bet the misidentifications would quickly diminish greatly.
"If police and prosecutors were sent to prison for a decade or more each time one of them lid or covered up facts at a trial. I'll bet the misidentifications would quickly diminish greatly."
Great idea! Could we get that to cover everything said by the President, Vice President, Cabinet members, and all members of Congress, too?
@Alobar: you're not thinking like a crook.
There are two ways to avoid your suggested punishment: either don't cover up facts, or don't get caught... which do you think is more likely?
North Carolina State Bureau of Investigation lab has come under scrutiny as incompetent or worse. Incorrect identification is compounded when the SBI lab can't analyze data that would exonerate the accused. It's a real mess.
"Fortunately, the man confessed and plead guilty. He knew they had him based on evidence. But the fact remains, they were appropriate in taking my sister's doubts into account and in no way lead her into the identification.
Posted by: HJohn at August 30, 2010 12:14 PM"
Is that what they were doing when she said she was 90% confident? Or were they trying to get her to revise her confidence to 100%?
Great article thanks for posting it, its very troubling. I understand forensics labs are often behind because of the amount to be worked on and the limited time/persons available to do it.
I wonder what happens if you are misidentified by someone as a terrorist and the Patriot Act covers up any sort of inquiry/clean up?
The crimes are heinos and of course the person responsible needs to be brought to justice. The problem is that the police are negligent in their act. "Give them a finger and they'll take the arm". Just like in security, people get comfortable and when there's no punishment and they're getting away with with everything, they'll let things slip. There's been countless of raids on the wrong homes because officers failed to verify what any normal person would consider to be a given.
If I understand correctly, promotion for police officers consists of many things, but arrests and acts of bravado are taken into account. Therefore, there's no compelling interest to do justice. The interest should come from harsh punishment. Should we not holding them accountable?
I'm also of the opinion that the investigation should be handled not by the police but by a third party. Police are a force to be used when force is necessary. An interrogation in a room with a cooperative suspect does not require force in most cases. However, this will never happen because people hate losing their power and because people will complain about costs. Policing should be left to control violent and dangerous situations.
The Indian, Hari Prasad, involved in the Indian Electronic Voting Machine hack (mentioned on this blog earlier) was arrested (supposedly on charges of 'stealing' the EVM) and then released on bail (as "no offence was disclosed").
Interestingly the magistrate has "observed that the police and the election commission should verify the claim made by Prasad and take necessary action if the claims were found to be baseless".
Wonder if they will really try to go down that path.. Though maybe they will, with their own pet scientists/ engineers.
Also the tone of the news article seems to run counter to that of the magistrate interestingly enough...
"You become more certain over time; that's just the way the mind works. ... And the police .... show the victim the same picture again and again to prepare her for the trial. So at a certain point you're ... just remembering this picture that you keep seeing."
I have often found that over time I tend to become less certain of things, but then my mind doesn't quite work like most people's. I have some degree of eidetic memory (I don't remember everything I experience, but what I do remember, I generally remember in great detail with all five senses), so if I can't recall something clearly and in detail, my instinct is to doubt my memory.
I've never been in a situation where someone was reinforcing a "memory" with cues like photos, but I think I would have very different memories of seeing a photo and seeing the actual person: I would remember things like the temperature, smell, appearance, etc. of the room where I saw the photo, as opposed to the situation in which I saw the person. I would be very curious to see how I would actually remember things in a situation like that, though.
Grits for Breakfast has covered false identification in Texas cases for quite a while. DNA evidence has resulted in a number of convictions being over turned in Dallas. Most of these cases had misidentification or bad forensics at the heart of the wrongful conviction.
"I'm also of the opinion that the investigation should be handled not by the police but by a third party. Police are a force to be used when force is necessary"
What you are vaguely describing fits the French system.
And there are a number of French ExPats I know who don't have very good things to say about the system.
The simple fact is any system is "workable" by those in it and the Politicos activly encorage an attitude of "Justice has to be seen to be done" not that it is actually done.
As for Identification by witnesses every sicentific test gives it a very very poor rating.
The first thing that goes wrong is at the event most witnesses suffer from "tunnel vision" as part of the bodies fight or flight reaction to fear.
This then causes "distraction/fixation" that is the witness is drawn to some feature that they think is odd or outstanding and focus in on it and ignore the rest of the details. Again this is due to the in built fight or flight reaction.
Thus post event their mind is briefly a blank canvas with perhaps one or two things they have remembered as little islands of colour on a grey background.
The next thing that happens usually well within an hour is the mind starts to colour in around the islands of real observation with what they think "likley to have happened" as the brain tries to find order in the chaos.
Thus anything that is said to the witness including neutral questions will colour the image further.
For instance a simple question might be "How high was the attacker?"
The brain panics as the witness wants to be helpfull and as people are very very bad at absolute measures the brain gets confused in trying to convert an indistinct memory into a number.
Thus a questioner should ask the witness to gauge the physical charecteristics against another object or person at the scene.
In reality nearly all evidence that appears in a court is increadably subjective. Even things that are portrayed as hard facts almost invariably are not when humans are involved.
The reality is Judges understand pieces of paper and little else. Jurors understand stories that relate to their own personal lives and thus prejudices.
The police and barristers know this and thus play to their respective audiances.
Experianced criminals know how to work this to their favour as well.
And as for any inocent falsely accused...
Well lets just say they generaly don't know enough not to talk themselves into jail...
Gee, I think of Bruce all the time I read about devious crimes and misdemeanors -- pardon the tautological (?) -- so he surely is the villain par excellence.
Hey out of my office window I see someone speeding: Must be him again. Get him.
Prosecutors like eyewitness identification because juries eat it up. Despite the fact that they're more likely to be wrong than right, tearful identifications in the courtroom can powerfully sway a jury.
It's all about the conviction rate, not the justice rate.
Another thing that always bothers me about wrongful convictions: the real criminal is out there, walking free, not under additional suspicion, and with an increased feeling that "they" aren't going to get him or her.
Consider probably the most celebrated of these cases, the Dreyfus affair in 19th-century France. Having pushed so hard for Dreyfus's conviction, the French Army was forced to not only exonerate Esterhazy (sp?), but leave him with access to secret material.
Eyewitnesses are regularly terrorised into recanting their identifications, are unwilling to testify in court or, in some extreme cases, are swimming with the fishes. The bad guys win either way.
@HJohn: re: "When she said she was 90% sure it was him, the police said, appropriately, "that's not good enough." "
Not appropriate. The only appropriate thing is for the police to note down "ID 90% certain" in the file and say "Thank you for your time." Any expression of approval or disapproval of the statements of the witness will influence the witness.
Regardless of whether it was deliberate one possible response to the "not good enough" comment is for the witness to say, "No, I'm certain." at which point the witness does become certain despite their previous doubts.
Ideally, a lawyer representing the lined-up people should be present for the entire encounter. One of the problems with the adversarial approach in the US is that the police are supposed to be neutral(i.e. they are supposed to be trying to get the real criminal), but in reality they are on the side of the prosecution (i.e. they want a conviction).
In my home state, witness "line-ups" are done as follows.
The suspect is photographed, head only, against a neutral background. From a library of pictures, 11 others of the same gender are selected randomly, with the proviso that they have to approximately match in hair colour and skin tone to any previous statement by the witness.
These 12 pictures are then presented to another detective, who has no idea which one is the suspect. They are shown to the witness in a 4 x 3 grid in a random order, and which one he or she chooses (if any) is noted.
This is then repeated with another 12 pictures: 11 more random ones, and a *different* shot of the suspect. If the suspect is confidently picked both times, you have a positive identification. Otherwise, not.
I think this is a remarkably fair and well thought-out system. Amusingly, it is actually a lot cheaper and easier to organise than an actual line-up. The only real issue is that if the witness recklessly nominates photos at random, there is 0.7% chance the suspect will be unfairly nominated. A third iteration of the test would be a significant improvement in that regard.
@Dr T. "What disturbs me most is that our investigative, arrest, and trial systems have positive incentives for misidentification that often exceed the incentives for identifying, arresting, and convicting the actual criminal."
I'm going to go a little off topic, but this is true. A friend of mine was pulled ofter after leaving my house because the police were looking for a red car in the area, and her license plate "matched two of the letters" on the plates they were looking for. Whey my friend objected that the remaining letters *didn't* match, the cop told her "Well, someone has to go to jail for this."
They did eventually let her go, but it took them about half an hour to decide someone else was a better candidate than she was.
Been there, had that done to me. So here's the story.
A little over 15 years ago, in the Netherlands, I was on my way to a party at my school. Two boys were seen trying to steal a moped. They fled, one of them running into the train station while the train I was in, was stopping there.
At my destination the train stopped at the end of the platform, not where it normally stopped. I got out and started heading in my usual direction when I noticed another flight of stairs was closer and I turned around and headed towards it. Which of course seemed very suspicious to the plain clothes police officers who were waiting on the platform.
I was arrested, I matched the general description (long hair, dark coat) and I was brought back to the police station of the town where the moped was stolen.
The witness saw me through the one way mirror and I actually heard him confirm that I was the one. Fortunately, both my parents and the guy taking my bike at the station confirmed my story and I was brought home, no harm done except having missed the party.
But those couple of words, "yeah, that's him all right", and the effect they had, I remembered for a long time.
I find the US system curious: in South Africa our law is Roman-Dutch. Our court system is based on a judge who applies hard-nosed rules of jurisprudance to the case. In a typical actual case a car thief, caught in the act by a friend of mine, was held in front of a judge by the police. He protested that he had no idea about this crime and he was arrested for no reason. The judge would have been forced to believe him since the word of the police officer is strictly no better than the word of the accused - i.e. they could have confused him with another joe locked in the same olding cell. However it so happens that the victim found a hat in the car that did not belong to him. The police so happened to later return the hat to the accused. So when the judge asked the victim: "Do you see that hat anywhere?" the victim replied, "Yes, the accused is wearing it!". On this bases the accused was convicted.
I have always held that there should be a special standard of evidence for capital cases. One based more on forensics and hard data.
Eyewitness testimony is the probably the WORST type of evidence there is due to the fragility of human memory (i.e how easily a person's memory of an event can be altered).
Here's an excellent excerpt I created(*) a long time ago from Elizabeth Loftus's book "Eyewitness Testimony":
I especially liked the classroom assignment about the non-existent tape recorder!
"Fish" (David B. Trout)
(*) With the author's permission of course! (+)
(+) Really! I wrote to her via email before I posted it (this was years ago) and she was kind enough to give me permission. She's a very nice lady and lives/works here in Seattle where I do (or did; don't know if she still does or eve if she's still alive. This WAS after all *years* ago).
I live in Toronto. A few years ago, I was unsuccessful in identifying a hit and run driver who stopped about 20 feet from me. I was able to memorize the license plate number. The cops were pretty sure who it was, because there'd been another incident a day or so earlier. And she was 'known to police.'
It was a few weeks after the accident that I went down for the line up.
I went into a room and met a different officer. There was a video camera recording everything. There were about 20 8 x10 pictures copied from id photos of the suspect and others who looked similar. We went through the pictures one by one and I had to write yes or no on the pack of each picture. I managed to get the directions wrong. I identified the wrong woman.
Schneier.com is a personal website. Opinions expressed are not necessarily those of BT.