Is the science of fingerprints reliable?

[aldiboronti]

@TokyoBayer

Looking at the wiki article about the Portland attorney, it seems to contradict the documentary in several key points.

“20 possible matches for one of the fingerprints were found in the FBI database and one of the possible matches was Brandon Mayfield.”

“The FBI described the fingerprint match as “100% verified”. According to the court documents in judge Ann Aiken’s decision, this information was largely “fabricated and concocted by the FBI and DOJ”.”

"As was discovered during the court case, even the FBI’s own records show that this fingerprint, despite the sworn testimony of FBI and DOJ agents, was in all reality not an exact match but only one of 20 prints “similar” to the ones retrieved from Madrid. "

"Before his arrest, Spanish authorities informed the FBI in a letter from April 13, that they reviewed the fingerprint on the bag as a negative match of Mayfield’s fingerprint,[4] though this letter was not communicated to Mayfield’s attorneys. "

If the article is correct, it does not support the argument that fingerprints are not unique.

That’s a very interesting wiki and fills in several aspects of the case which the documentary didn’t cover. One thing I don’t understand. The attorney later sued the FBI for invasion of his privacy and challenged the constitutionality of the Patriot Act under which the FBI had acted. The judge found in his favor but that ruling was overturned by the appeal court on the grounds that Mayfield didn’t have standing.

Now standing is defined by Wikipedia thus: “the ability of a party to demonstrate to the court sufficient connection to and harm from the law or action challenged to support that party’s participation in the case.” How on earth did Mayfield not have standing in this case?

[DSYoungEsq]

@aldiboronti

Now standing is defined by Wikipedia thus: “the ability of a party to demonstrate to the court sufficient connection to and harm from the law or action challenged to support that party’s participation in the case.” How on earth did Mayfield not have standing in this case?

Why don’t you go look up the decision of the appellate court that said he didn’t. The great thing about decisions is that they are accompanied by explanations.

[aldiboronti]

@DSYoungEsq

Why don’t you go look up the decision of the appellate court that said he didn’t. The great thing about decisions is that they are accompanied by explanations.

Google is my friend in other words. Fair enough, no offense taken.

[Really_Not_All_That_Bright]

In case you are still confused, it’s because the government had settled with him and the remaining cause of action left to him was not one where standing could be found. If he was still pursuing a §1983 claim he would certainly have had standing.

[md2000]

There’s no doubt about, it’s the myth of fingerprints…
I’ve seen them all and the are all the same…" ♫

-Paul Simon, The Myth of Fingerprints

IIRC, even back around 1900, Sherlock Holmes in one of his cases encountered fabricated fingerprint evidence. (Made from a thumbprint on a wax seal, if it’s not too much a spoiler).

The “no two identical” claim comes from statistics - apparently fingerprints are random developments during gestation, although the basic patterns seem to be hereditary. As a result, the odds become astronomical depending on how identical you want your match to be. it’s like the monkeys typing Shakespeare theory - the odds are so unlikely that they are impossible. But if you just calculate the odds of say, 24 points matching - an elementary 2^24 is still a large number (4 million?), and IIRC the values of those 24 numbers are not a simple binary.

[Flyer]

@aldiboronti

I just watched a 2012 documentary on forensic science. During the show one judge they spoke to said there was no scientific basis for the belief that no two persons’ fingerprints could be identical, that no study had been made to demonstrate that theory nor could such a study be done. Fingerprinting is thus not infallible, as the FBI has often claimed in the past, but is a question of probabilities and when it comes to partial prints the ground is even less secure, with some courts demanding 16 point matches while others will accept as few as 6.

I found this astonishing. In light of the above, if in fact it’s true, how could anyone be convicted on fingerprint evidence alone, especially with a partial print? The show interviewed one Portland attorney who had been arrested by the FBI in 2004 as a result of them matching his fingerprints to that of a partial print left on a bag of detonators in the Madrid terrorist attack of that year. Four FBI analysts examined the prints and each said they were 100% certain that they matched. This poor guy, who was still visibly shaken as he recalled the events years later for the show, was heading for a probable death sentence; even his own lawyers thought he was guilty. Had not the Spanish police finally matched the latent print to an Algerian living in Spain the unfortunate attorney would have been toast. (You also have to wonder whether the Algerian was guilty!)

After watching the show I began to think that maybe fingerprinting was on the same level as lie-detecting. But before reaching a conclusion I figured I needed to get the SD on this.

I’ve read all my life that there is a 1 in 7 billion chance that any two people will have identical fingerprints.

No cite, because I don’t remember where I’ve seen that claim, but I’ve seen it in multiple books over the years.

[md2000]

Back of envelope guess, fingerprint lines appear to be roughly(!) 2 per mm. back of the envelope calculation, a fingerprint pad is 1x2cm (the part that is left at the crime scene, ignoring the wraparound and the further parts of the finger). So there are 1x50x2x50 = 5,000 points where a line might end, continue, or merge - 3 choices. (We’re not even talking whether the “continue” is up-down or left-right; there are default patterns, circular or inverted T patterns).

So 3^5,000; even eliminate a few points since a fingerprint identical but shifted up or down fraction is still a match… so say 3^2500 = 6x10^1192
I’d say that approaches monkey-typing-Shakespeare improbability unless you can show there’s a repeatable mechanism for determining how fingerprint patterns are formed.

But, as pointed out very early in the thread, proving a negative is pretty much impossible; we have only the empirical evidence to go on, and nobody has mentioned any regular instances of “almost identical”.

[Chronos]

Actually, almost-identical fingerprints are fairly common, since identical twins have almost-identical fingerprints.

[Jackmannii]

@md2000

IIRC, even back around 1900, Sherlock Holmes in one of his cases encountered fabricated fingerprint evidence. (Made from a thumbprint on a wax seal, if it’s not too much a spoiler).

It made for a good story. But I’ve never heard of that type of fakery ever having been accomplished in real life. And while wax seals are not commonplace now, you could theoretically trick someone into leaving a thumbprint in another yielding substance (a banana? a wedge of cheese?) and try to leave that impression on a streak of blood at the crime scene. The foreign matter might give you away though.

Short of truly identical fingerprints turning up in the future, I’m not sure how the uniqueness of fingerprints could be “disproved”.

I’m reminded of elements of modern medicine that have never been established on the basis of rigorous clinical trials. For instance, we haven’t really proven that surgery for a ruptured appendix saves lives. Odds are that the necessary “proof” will never be forthcoming, unless a lot of people with impending appendiceal ruptures volunteer for a study, knowing that they could be assigned to the control group. :eek:

[Bill_Door]

We’re not talking about an entire fingerprint in these cases. These fingerprints aren’t collected by experts using ink pads. They collect fragmentary and distorted latent prints which they rotate and skew and shape to fit. Here’s an interesting study done for the U. S. Department of Justice. An excerpt:

Non-matching trials include prints that do not originate from the same source; participants responded to a total of 1144 of these trials. Participants correctly labeled 96.8% of the non-matching trials as “no match” (correct rejections), and incorrectly labeled 3.2% of the non-matching trials as “match” (false alarms). In absolute terms, participants correctly labeled 1107 of the 1144 non-matching trials as “no match” (correct rejections), and incorrectly labeled 37 out of the 1144 non-matching trials as “match” (false alarms). Nineteen examiners made at least one false alarm, and
twenty-seven of the non-matching fingerprint stimulus pairs caused at least one false alarm.

So, hey, you’ve got less than a 1 in 25 chance of getting a false match. That’s pretty good, right?

At the level of the participant, twelve participants made one false alarm; three participants made two false alarms; three participants made three false alarms; and one participant made ten false alarms.

There were 56 fingerprint examiners used in the study, so one-third of them made at least one false alarm, and one of them made ten. I kind of hope that one isn’t examining my fingerprints, but he/she is likely out there examining someone else’s right now.

[BwanaBob]

@Chronos

Actually, almost-identical fingerprints are fairly common, since identical twins have almost-identical fingerprints.

How do you jump from “almost-identical fingerprints with identical twins” to fairly common (implied between random strangers)? That’s a big leap since identical twins are a narrow subset of the general population.

[DSYoungEsq]

@BwanaBob

How do you jump from “almost-identical fingerprints with identical twins” to fairly common (implied between random strangers)? That’s a big leap since identical twins are a narrow subset of the general population.

He didn’t say that it applied to random strangers. He simply said that there are a lot of identical twins.

If you have a fingerprint you are using to try and establish the identity of the perpetrator of a crime, and that person has an identical twin, then there is likely to be one set of almost-identical fingerprints, right?

[md2000]

The pattern that is “matched” in fingerprints is the points where lines end, or merge. The general patterns fall into the category of oval whorls or inverted T pattern of loops, and this is somewhat inherited. AFAIK (not an expert) the likelihood that a pair of identical twins will have “almost identical” patterns is not much better than any two related individuals who inherit the same master pattern. The line components are determined randomly during fetal development.