- Forensic science plays a crucial role in criminal convictions, but its reliability is often overstated.
- Many forensic sciences do not meet the fundamental requirements of science, according to reports.
- The phrase "to a reasonable degree of scientific certainty" used by forensic experts has no generally accepted meaning in science.
- Misapplication of forensic science has led to wrongful convictions, with errors found in FBI hair analysis in 96% of cases.
- Bite mark analysis is highly subjective and unreliable, yet it has been used in court cases.
- Decisions about the validity of scientific evidence are often made by scientifically illiterate individuals.
- While fingerprints and DNA are considered reliable, they are not infallible, as shown by cases of false positives in fingerprint analysis.

- DNA is considered the most reliable form of evidence in forensic science, but not all DNA tests are equal and crime scenes can produce DNA of varying quality.
- Lower quality DNA samples are sometimes presented as highly reliable to juries, leading to flawed convictions.
- Some forensic labs work closely with law enforcement, which can create subconscious bias and prejudice their work.
- Some states, like Texas, have taken steps to address these issues, but progress at the federal level has been slower.
- The National Commission on Forensic Science, which aimed to address these problems, was shut down by Attorney General Jeff Sessions.
- This lack of progress is dangerous as innocent people are being convicted and guilty criminals are left on the streets.
- Potential jurors should be educated about the shortcomings in the system.
- A satirical video clip highlights the flaws in forensic science and the need for better understanding and standards in the field.

Forensic science used in criminal trials can be surprisingly unscientific. Maybe a new television procedural could help change the public perception.

Connect with Last Week Tonight online...

Subscribe to the Last Week Tonight YouTube channel for more almost news as it almost happens: www.youtube.com/user/LastWeekTonight

Find Last Week Tonight on Facebook like your mom would: http://Facebook.com/LastWeekTonight

Follow us on Twitter for news about jokes and jokes about news: http://Twitter.com/LastWeekTonight

Visit our official site for all that other stuff at once: http://www.hbo.com/lastweektonight

You know, that thing that was almost solved by a flasher dog in the 1980s.

But specifically, this story is about how we increasingly solve crimes using forensic evidence.

It's that thing that is just a staple of TV crime shows.

Were you able to determine which monkey bit him?

The bite marks matched those of the monkey found at the scene.

That last 1 was presumably from 1 of the crossover episodes where the team from Law & Order worked a case with the cast of Monkey Law and Monkey Order.

But on TV and in real life, forensic science plays an important role in criminal convictions.

Prosecutors often complain about a so-called CSI effect, where jurors expect to see forensic evidence in every case.

The problem is, not all forensic science is as reliable as we've become accustomed to believe.

A report in 2009 by the National Academy of Sciences found that many forensic sciences do not meet the fundamental requirements of science.

And a report last year by a presidential science council agreed saying that expert witnesses have often overstated the value of their evidence, going far beyond what the relevant science can justify.

It's not that all forensic science is bad, because it's not, but too often, its reliability is dangerously overstated.

And 1 sign of that is that forensic experts in court are often nudged to use 1 very convincing phrase.

To a reasonable degree of scientific certainty.

reasonable degree of scientific certainty.

Within a reasonable scientific certainty.

Are you able to say that within a reasonable degree of scientific certainty?

And here's the thing, that phrase does have a persuasive ring to it.

Unfortunately, as that presidential council pointed out, it has no generally accepted meaning in science.

It's 1 of those terms like basic or trill that has no commonly understood definition.

I mean, I do feel trill, so I'm guessing it's awful.

-♪ ♪ -♪ And when bad science is confidently presented, terrible convictions can happen.

In fact, among the hundreds of people who've been exonerated by DNA testing since 1989.

In nearly half of their cases, there was some misapplication of forensic science.

And there are people behind those numbers.

Take Santé Tribble, who was convicted of murder and served 26 years, in large part, thanks to an FBI analyst who testified that his hair matched hairs found at the scene.

And as he will tell you, the evidence was presented as being rock solid.

They said it matched my hair in all microscopical characteristics.

And that's the way they presented it to the jury, and the jury took it for granted that that was my hair.

Because who other than an FBI expert would possibly know that much about hair?

Except, of course, for whoever styled Amanda Seyfried at the 2009 Oscars.

Breathtaking waves without losing any of their body or bounce.

Stunning, stunning, stunning, stunning, stunning.

The Jurors in Tribble's case were actually told there was 1 chance in 10000000 that it could be someone else's hair.

Because once DNA analysis became available, his lawyer tested the 13 hairs from the case.

And not only were none of them his, some of what they found was incredible.

9 of the hairs had come from the same source, a couple had come from different sources, and 1 was a dog.

2 different FBI agents who had looked at that and analyzed it didn't recognize that it was dog hair?

So, first, it is amazing that he is able to laugh at that.

But second, if a dog did commit the crime, there's really no recourse there, because there is actually no law against dogs committing murder.

And that's a fact that I learned in Air Bud 9, fuck the police.

And it turns out, Tribble is not the only case where FBI experts overstated their confidence in their results.

The Innocence Project and the National Association of Criminal Defense Lawyers found from the 1970s through 1999, in 268 cases where FBI hair analysis led to a conviction, 257, or 96 percent of them, had errors in analysis.

Oh, it gets worse because 9 of those defendants had already been executed, which is horrifying.

And you would expect FBI hair analysis to have a higher rate of accuracy than your friend's hair analysis of, you can totally pull off bangs.

And look, it's by no means just microscopic hair comparison, which has had the reliability of its results overstated.

Those reports that I showed earlier suggest there is weak scientific support for some aspects of techniques like blood pattern, footwear, firearm, and bite mark analysis.

And you must be familiar with that last 1 from cool scenes like this.

And we all know yellow rectangle is the highest level of match a computer can give you about teeth?

But look, in the real world, bite mark analysis is highly subjective and unreliable.

The President's Council found the entire discipline does not meet the scientific standards for foundational validity, which I believe is science speak for bullshit.

But people have been sent to prison on the basis of bite mark testimony by experts like Dr.

The science of bite marks analysis is very accurate.

When it comes to bite marks, West considers himself the maestro.

He's found bite marks on a decomposed body submerged in a swamp, on a corpse that had been buried for more than a year.

He's even used a bite mark taken out of a bologna sandwich to get a conviction.

Now, that sounds impressive, matching a killer's teeth to a bite mark in a bologna sandwich.

Although you should know that the defendant in that case got a new trial after an autopsy report found that the murder victim had actually eaten a small amount of bologna, consistent with the amounts bitten off the sandwich.

So that sandwich was irrelevant to the case.

In fact, you could even argue that it was actually Dr.

West who was full of, say it with me, shit.

And that is not the only issue that has arisen from his testimony.

There are now 5 cases where he testified for the prosecution and whether charges were dropped or the conviction was later overturned.

And even West himself has admitted that he no longer believes in bite mark analysis for identifying perpetrators, and he doesn't think it should be used in court.

And yet, incredibly, every time a defendant has challenged its validity, the court has ruled it admissible.

And a key reason for that is that judges often rely on precedent to decide what to allow in front of a jury.

So, if a particular discipline has been in court before, it is likely that a judge will admit it again.

All of which means that, as the co-founder of the Innocence Project points out, decisions about the validity of science are being made by people who don't necessarily know much about it.

Historically, we had a situation where 2 scientifically illiterate lawyers argued the bona fides of scientific evidence before a scientifically illiterate judge, so the 12 scientifically illiterate jurors could decide the weight of that evidence.

And if you think about it, that's absolutely terrifying.

Trials can often be a situation when no 1 really knows what they're doing.

It's like a cooking competition for toddlers hosted by a stray cat and judged by goats.

Oh, the tuna was undercooked and covered in cold spaghetti sauce.

You then, for some reason, covered the whole dish in honey nut Cheerios.

--LAUGHTER --And look, none of this is to say that there is not reliable forensic science out there.

Fingerprints and DNA are obvious examples.

But while we think of them as perfect, it is important to know they are by no means infallible.

The FBI has found fingerprint analysis could have a false positive rate as high as 1 error in 306 cases.

And a dramatic example of this came after the Madrid train bombings in 2004, when the FBI arrested this Oregon man, Brandon Mayfield.

He had never even been to Spain in his life, but 3 separate examiners matched his fingerprints to 1 on a bag of detonators.

So he was, at that point, completely fucked, until investigators happened to determine that that fingerprint actually also matched someone else who was in Spain at the time.

And that blew the minds of fingerprint experts.

We always assume that fingerprints are very, very unique.

But what the Mayfield case demonstrates is that parts of a fingerprint can be so similar, it's possible for 2 people

It turns out that 2 people can have fingerprints that are so close that even experts can't tell them apart, meaning that we are now this close to finally proving my theory that there is only 1 Olsen twin.

She's just moving very fast, back and forth.

Now, I don't know how this new information helps me yet, but when it does, the end is nigh, you fraud!

♪ And then, then there is DNA, which is the gold standard in forensic science for a reason, because in perfect conditions, it's seen as the most reliable form of evidence, but not all DNA tests are equal, and crime scenes can produce DNA of widely varying quality.

DNA is very fragile, and easily mixed up at a messy scene.

So, imagine you come across a crime scene, you may have a pool of blood, but it may not just be 1 person's blood, right?

The more contributors to that, that mixture of DNA, the more difficult it is to determine whose DNA it was, whose blood it was.

It can be difficult to tell whose blood is whose in a large pool of blood, which is, coincidentally, the premise of my new game show, So You Think You Can Tell Whose Blood Is Whose in a Large Pool of Blood.

It premieres on Tuesday night, and apparently, it's already been canceled.

--LAUGHTER --CLAPPING But the problem here is, lower quality DNA samples are sometimes presented to juries as if they are highly reliable.

In 2003, a prosecutor in a double murder told the jury that the odds the defendant's DNA matched the glove found at the scene by chance was 1 in 1.1 billion.

But it turned out the glove actually contained at least 3 people's DNA, and a later analysis put the odds closer to 1 in 2.

People do confuse the numbers 1.1 billion and 2 all the time.

That's why I'm always mistakenly saying that my favorite R&B group is Boys 1.1 Billion Men.

And on top of all of this, there is 1 more factor that can be impossible to detect, and it concerns the relationship between law enforcement and the forensic labs themselves.

Because you would hope that those labs would work independently, taking in evidence and splitting out results.

But many labs work closely with law enforcement, knowing details of the case that they are working on, which can prejudice their work even subconsciously.

Sometimes it's not intentional fraud, but rather something, much more, inadvertent.

Which is the kind of bias that can come from feeling like you're part of a side, part of a team, that you, you're part of, You're attached to the prosecution, and you want to get, get the bad guy.

If a referee started participating in a team's end zone celebration, you'd have some serious fucking questions.

Like, why have you picked a side, and how long have you been practicing the dirty bird?

So clearly here, a lot needs to be fixed, and some states have stepped up.

1 has done a lot, including passing a first of its kind junk science law, which enables convicts to request a new trial if the science used to convict them was flawed.

And the pioneering state that did that, by the way, was Texas.

You don't expect Texas to lead the nation in science-related criminal justice reform.

You expect them to lead the nation in remembering the Alamo, or naming their children football.

I love your football, But if you ever forget the Alamo, we are done.

Now, sadly, at the federal level, progress has been slower, although 1 group, the National Commission on Forensic Science, has tried to fix that.

They were founded to advise the DOJ on how to address many of the problems that you have seen tonight.

And their most recent meeting featured powerful remarks from Keith Harwood, who spent 33 years in prison for a crime that he did not commit, based on faulty bite mark evidence.

There's no possibility, excuse me if I get emotional, that I spent more than half my life in prison behind the opinions and the expert egos of 2 odontologists.

There's a death penalty case in Pennsylvania that's going on now and the judge is going to allow bite mark evidence.

How many people have to be wrongly convicted before they realize that this stuff's all bogus?

And it's also the kind of speech that could really inspire that commission to do a lot of good work.

Unfortunately, that was actually their final meeting, because the commission was shut down in April by Attorney General and xenophobic boss baby, Jeff Sessions.

That shouldn't really surprise you anyway.

Sessions is a former prosecutor, and he does seem like the kind of guy who watched Dead Man Walking and was like, hurry up, let's kill the guy already!

This movie should be called Dead Man Dilly-Dallying.

So we may honestly be actively going backwards on this issue, which is dangerous, because not only are innocent people getting convicted, guilty criminals are being left on the streets as a result.

And if this administration does not see this as a problem, then we should at the very least do more to educate potential jurors about some of the shortcomings in our system.

And 1 small way to do that might be this.

All right, everybody, let's get this stuff back to the lab.

He's a crime solver who doesn't like to play by the rules.

I'm getting something on these bite marks here, but it's far from conclusive.

Would you say there's a reasonable degree of scientific certainty?

And that's a problem for absolutely everyone around him.

Chief, the hair matches the victim's wife.

Microscopic hair comparison is bullshit science.

I ran a mitochondrial DNA test on those hairs.

1 was from a cabbage patch kid, and the remaining 1 was from the wife.

So there's our killer right there, chief.

And he won't stop asking the hard questions.

How about a certainly reasonable science degree,

Because he's passionate about his job, despite not fully understanding what that job is.

If we don't have something solid by tomorrow, DA's going to have my ass.


Speaker 0: -♪ ♪ -♪ ♪ Crime. You know, that thing that was almost solved by a flasher dog in the 1980s. But specifically, this story is about how we increasingly solve crimes using forensic evidence. It's that thing that is just a staple of TV crime shows.

Speaker 1: Bullet from the torso on the left. Bullet from the boat on the right.

Speaker 2: 2 hearts beat as 1.

Speaker 0: ---.

Speaker 2: That's just how perfectly...

Speaker 0: --.

Speaker 3: That's a match.

Speaker 0: --.

Speaker 4: We got a match. And they just found us

Speaker 3: a match.

Speaker 5: Visible match. We have a match. Match.

Speaker 2: Were you able to determine which monkey bit him?

Speaker 6: The bite marks matched those of the monkey found at the scene.

Speaker 0: Wow! That last 1 was presumably from 1 of the crossover episodes where the team from Law & Order worked a case with the cast of Monkey Law and Monkey Order. -♪ ♪ -♪

Speaker 4: Ah!

Speaker 0: ♪ But... But on TV and in real life, forensic science plays an important role in criminal convictions. Prosecutors often complain about a so-called CSI effect, where jurors expect to see forensic evidence in every case. The problem is, not all forensic science is as reliable as we've become accustomed to believe. A report in 2009 by the National Academy of Sciences found that many forensic sciences do not meet the fundamental requirements of science.

And a report last year by a presidential science council agreed saying that expert witnesses have often overstated the value of their evidence, going far beyond what the relevant science can justify. And that's the thing here. It's not that all forensic science is bad, because it's not, but too often, its reliability is dangerously overstated. And 1 sign of that is that forensic experts in court are often nudged to use 1 very convincing phrase.

Speaker 7: To a reasonable degree of scientific certainty.

Speaker 3: To a reasonable

Speaker 8: degree of scientific certainty. To a

Speaker 3: reasonable degree of scientific certainty.

Speaker 9: To a reasonable degree of scientific certainty.

Speaker 3: Within a reasonable scientific certainty.

Speaker 7: To a reasonable degree of scientific certainty.

Speaker 2: Are you able to say that within a reasonable degree of scientific certainty? Yes.

Speaker 0: And here's the thing, that phrase does have a persuasive ring to it. Unfortunately, as that presidential council pointed out, it has no generally accepted meaning in science. It's 1 of those terms like basic or trill that has no commonly understood definition. Am I trill? Is that good or bad?

I mean, I do feel trill, so I'm guessing it's awful. -♪ ♪ -♪ And when bad science is confidently presented, terrible convictions can happen. In fact, among the hundreds of people who've been exonerated by DNA testing since 1989. In nearly half of their cases, there was some misapplication of forensic science. And there are people behind those numbers.

Take Santé Tribble, who was convicted of murder and served 26 years, in large part, thanks to an FBI analyst who testified that his hair matched hairs found at the scene. And as he will tell you, the evidence was presented as being rock solid.

Speaker 10: They said it matched my hair in all microscopical characteristics. And that's the way they presented it to the jury, and the jury took it for granted that that was my hair.

Speaker 0: But, you know, I can see why they did. Because who other than an FBI expert would possibly know that much about hair? Except, of course, for whoever styled Amanda Seyfried at the 2009 Oscars. Breathtaking waves without losing any of their body or bounce. Stunning.

-...stunning. --STUNNING. Stunning. Stunning. Stunning.

Stunning, stunning, stunning, stunning, stunning. The Jurors in Tribble's case were actually told there was 1 chance in 10000000 that it could be someone else's hair. And guess what? He was exonerated. Because once DNA analysis became available, his lawyer tested the 13 hairs from the case.

And not only were none of them his, some of what they found was incredible.

Speaker 3: 9 of the hairs had come from the same source, a couple had come from different sources, and 1 was a dog.

Speaker 2: 2 different FBI agents who had looked at that and analyzed it didn't recognize that it was dog hair?

Speaker 3: It was a canine. It was a domestic dog, yes.

Speaker 10: My personal conclusion was... The dog committed the crime.

Speaker 0: -♪ ♪ -♪ Okay. So, first, it is amazing that he is able to laugh at that. But second, if a dog did commit the crime, there's really no recourse there, because there is actually no law against dogs committing murder. And that's a fact that I learned in Air Bud 9, fuck the police. --LAUGHTER AND APPLAUSE --And...

And it turns out, Tribble is not the only case where FBI experts overstated their confidence in their results.

Speaker 8: The Innocence Project and the National Association of Criminal Defense Lawyers found from the 1970s through 1999, in 268 cases where FBI hair analysis led to a conviction, 257, or 96 percent of them, had errors in analysis.

Speaker 0: Oh, it gets worse because 9 of those defendants had already been executed, which is horrifying. And you would expect FBI hair analysis to have a higher rate of accuracy than your friend's hair analysis of, you can totally pull off bangs. Because you can't, you absolutely can't. Believe me, I couldn't. Just learn from my mistakes, kids.

Save yourselves. It's too late for me. And look, it's by no means just microscopic hair comparison, which has had the reliability of its results overstated. Those reports that I showed earlier suggest there is weak scientific support for some aspects of techniques like blood pattern, footwear, firearm, and bite mark analysis. And you must be familiar with that last 1 from cool scenes like this.

Speaker 11: A little 3D magic for clarity, and

Speaker 5: I give you the killer's incisors.

Speaker 0: -♪♪ Oh, no! The computer rated it yellow rectangle. And we all know yellow rectangle is the highest level of match a computer can give you about teeth? But look, in the real world, bite mark analysis is highly subjective and unreliable. The President's Council found the entire discipline does not meet the scientific standards for foundational validity, which I believe is science speak for bullshit.

But people have been sent to prison on the basis of bite mark testimony by experts like Dr. Michael West.

Speaker 12: The science of bite marks analysis is very accurate.

Speaker 9: When it comes to bite marks, West considers himself the maestro. He's found bite marks on a decomposed body submerged in a swamp, on a corpse that had been buried for more than a year. He's even used a bite mark taken out of a bologna sandwich to get a conviction.

Speaker 0: Now, that sounds impressive, matching a killer's teeth to a bite mark in a bologna sandwich. Although you should know that the defendant in that case got a new trial after an autopsy report found that the murder victim had actually eaten a small amount of bologna, consistent with the amounts bitten off the sandwich. So that sandwich was irrelevant to the case. In fact, you could even argue that it was actually Dr. West who was full of, say it with me, shit.

And that is not the only issue that has arisen from his testimony. There are now 5 cases where he testified for the prosecution and whether charges were dropped or the conviction was later overturned. And even West himself has admitted that he no longer believes in bite mark analysis for identifying perpetrators, and he doesn't think it should be used in court. And yet, incredibly, every time a defendant has challenged its validity, the court has ruled it admissible. And a key reason for that is that judges often rely on precedent to decide what to allow in front of a jury.

So, if a particular discipline has been in court before, it is likely that a judge will admit it again. All of which means that, as the co-founder of the Innocence Project points out, decisions about the validity of science are being made by people who don't necessarily know much about it.

Speaker 13: Historically, we had a situation where 2 scientifically illiterate lawyers argued the bona fides of scientific evidence before a scientifically illiterate judge, so the 12 scientifically illiterate jurors could decide the weight of that evidence.

Speaker 0: And if you think about it, that's absolutely terrifying. Trials can often be a situation when no 1 really knows what they're doing. It's like a cooking competition for toddlers hosted by a stray cat and judged by goats. Oh, the tuna was undercooked and covered in cold spaghetti sauce. You then, for some reason, covered the whole dish in honey nut Cheerios.

I loved it. --LAUGHTER --And look, none of this is to say that there is not reliable forensic science out there. Fingerprints and DNA are obvious examples. But while we think of them as perfect, it is important to know they are by no means infallible. The FBI has found fingerprint analysis could have a false positive rate as high as 1 error in 306 cases.

And a dramatic example of this came after the Madrid train bombings in 2004, when the FBI arrested this Oregon man, Brandon Mayfield. He had never even been to Spain in his life, but 3 separate examiners matched his fingerprints to 1 on a bag of detonators. So he was, at that point, completely fucked, until investigators happened to determine that that fingerprint actually also matched someone else who was in Spain at the time. And that blew the minds of fingerprint experts.

Speaker 4: We always assume that fingerprints are very, very unique. But what the Mayfield case demonstrates is that parts of a fingerprint can be so similar, it's possible for 2 people

Speaker 14: to be identified to 1 latent print.

Speaker 0: That's true. It turns out that 2 people can have fingerprints that are so close that even experts can't tell them apart, meaning that we are now this close to finally proving my theory that there is only 1 Olsen twin. She's just moving very fast, back and forth. She confuses your eye. Now, I don't know how this new information helps me yet, but when it does, the end is nigh, you fraud!

You fraud! -♪ Ch-ch-ch-ch-ch-ch-ch. ♪ -♪ Ch-ch-ch-ch-ch-ch-ch. ♪ -♪ Ch-ch-ch-ch-ch-ch-ch. ♪ -♪ Ch-ch-ch-ch-ch-ch-ch.

♪ And then, then there is DNA, which is the gold standard in forensic science for a reason, because in perfect conditions, it's seen as the most reliable form of evidence, but not all DNA tests are equal, and crime scenes can produce DNA of widely varying quality.

Speaker 8: DNA is very fragile, and easily mixed up at a messy scene.

Speaker 15: So, imagine you come across a crime scene, you may have a pool of blood, but it may not just be 1 person's blood, right? The more contributors to that, that mixture of DNA, the more difficult it is to determine whose DNA it was, whose blood it was.

Speaker 0: Exactly. It can be difficult to tell whose blood is whose in a large pool of blood, which is, coincidentally, the premise of my new game show, So You Think You Can Tell Whose Blood Is Whose in a Large Pool of Blood. It premieres on Tuesday night, and apparently, it's already been canceled. --LAUGHTER --CLAPPING But the problem here is, lower quality DNA samples are sometimes presented to juries as if they are highly reliable. In 2003, a prosecutor in a double murder told the jury that the odds the defendant's DNA matched the glove found at the scene by chance was 1 in 1.1 billion.

So that's pretty strikingly impressive. But it turned out the glove actually contained at least 3 people's DNA, and a later analysis put the odds closer to 1 in 2. And you know what? That's close enough, isn't it? People do confuse the numbers 1.1 billion and 2 all the time.

That's why I'm always mistakenly saying that my favorite R&B group is Boys 1.1 Billion Men. And on top of all of this, there is 1 more factor that can be impossible to detect, and it concerns the relationship between law enforcement and the forensic labs themselves. Because you would hope that those labs would work independently, taking in evidence and splitting out results. But many labs work closely with law enforcement, knowing details of the case that they are working on, which can prejudice their work even subconsciously.

Speaker 16: Sometimes it's not intentional fraud, but rather something, much more, inadvertent. Which is the kind of bias that can come from feeling like you're part of a side, part of a team, that you, you're part of, You're attached to the prosecution, and you want to get, get the bad guy.

Speaker 0: Yeah, but that's not their job. At all. They are supposed to be neutral. If a referee started participating in a team's end zone celebration, you'd have some serious fucking questions. Like, why have you picked a side, and how long have you been practicing the dirty bird?

So clearly here, a lot needs to be fixed, and some states have stepped up. 1 has done a lot, including passing a first of its kind junk science law, which enables convicts to request a new trial if the science used to convict them was flawed. And that sounds great. And the pioneering state that did that, by the way, was Texas. Yes, I know.

Texas. You don't expect Texas to lead the nation in science-related criminal justice reform. You expect them to lead the nation in remembering the Alamo, or naming their children football. I love your football, But if you ever forget the Alamo, we are done. Now, sadly, at the federal level, progress has been slower, although 1 group, the National Commission on Forensic Science, has tried to fix that.

They were founded to advise the DOJ on how to address many of the problems that you have seen tonight. And their most recent meeting featured powerful remarks from Keith Harwood, who spent 33 years in prison for a crime that he did not commit, based on faulty bite mark evidence.

Speaker 12: Some would say, well, you're a free man. Well, I will never be free of this. There's no possibility, excuse me if I get emotional, that I spent more than half my life in prison behind the opinions and the expert egos of 2 odontologists. There's a death penalty case in Pennsylvania that's going on now and the judge is going to allow bite mark evidence. How many people have to be wrongly convicted before they realize that this stuff's all bogus?

It's all made up?

Speaker 0: That's a good question. And it's also the kind of speech that could really inspire that commission to do a lot of good work. Unfortunately, that was actually their final meeting, because the commission was shut down in April by Attorney General and xenophobic boss baby, Jeff Sessions. And you know what? That shouldn't really surprise you anyway.

Sessions is a former prosecutor, and he does seem like the kind of guy who watched Dead Man Walking and was like, hurry up, let's kill the guy already! This movie should be called Dead Man Dilly-Dallying. Let's go, people, let's go! So we may honestly be actively going backwards on this issue, which is dangerous, because not only are innocent people getting convicted, guilty criminals are being left on the streets as a result. And if this administration does not see this as a problem, then we should at the very least do more to educate potential jurors about some of the shortcomings in our system.

And 1 small way to do that might be this.

Speaker 6: In this city, when the heat rises... So does the murder rate.

Speaker 2: All right, everybody, let's get this stuff back to the lab. Hey, Chief, no need. Look what I found. Bite mark. Bologna sandwich.

Now that's what I call... Dead meat. Yeah!

Speaker 3: Whoa! The fuck is wrong with you?

Speaker 7: Dead meat? This is a murder. That's a human person.

Speaker 3: That is his wife over there. You're a

Speaker 16: fucking asshole.

Speaker 6: He's a crime solver who doesn't like to play by the rules.

Speaker 3: I'm getting something on these bite marks here, but it's far from conclusive.

Speaker 2: Would you say there's a reasonable degree of scientific certainty?

Speaker 3: No. That's meaningless. Right.

Speaker 6: And that's a problem for absolutely everyone around him.

Speaker 2: Chief, the hair matches the victim's wife. Case closed. Slow down. Microscopic hair comparison is bullshit science. Chief,

Speaker 7: I ran a mitochondrial DNA test on those hairs.

Speaker 2: Your wife did it, right? Case closed.

Speaker 7: Actually, there were 5 hairs. 3 were from a coconut.

Speaker 2: And 2 were from

Speaker 6: the wife.

Speaker 7: 1 was from a cabbage patch kid, and the remaining 1 was from the wife. This golden retriever.

Speaker 2: So there's our killer right there, chief. The victim was shot. How could a dog fire a gun? That's a bad dog right there. He's about to face some roof justice.

Speaker 6: We're indoors. Fuck you. And he won't stop asking the hard questions.

Speaker 2: How about a certainly reasonable science degree,

Speaker 6: you 2? No. Yeah, okay. Because he's passionate about his job, despite not fully understanding what that job is.

Speaker 2: If we don't have something solid by tomorrow, DA's going to have my ass.

Speaker 3: Why would the DA have your ass? We don't work for the DA.

Speaker 2: What?

Speaker 3: We don't work for the DA. We do understand that, right? Please, tell us you understand that.

Speaker 2: Well, yeah. I understand that, totally.

Speaker 6: Oh, there's no way he understands that, because this guy will not quit.

Speaker 2: How about a certainly reasonable?

Speaker 6: Stop talking. OK. Keep up the good work. OK. And even when his team abandons him, he's not afraid to call for backup.

Speaker 2: This better be good. Oh, it is. I brought in some extra help, expert witnesses to help lock in this case. Take a look. We got a forensic dentist, twin boy detectives, an old-timey prospector with a divining rod, a sack full of Magic 8-Balls, and the county's foremost crime-sniffing pony.

None of this is admissible in court. Actually, 3 of them have testified in court before, and they all got convictions.

Speaker 3: Is that the baloney evidence? No!

Speaker 13: Oh, my God!

Speaker 6: CSI, Crime Scene Idiot.

See all LastWeekTonight transcripts on Youtube

Forensic Science: Last Week Tonight with John Oliver (HBO)