Fixing Fingerprints

I can remember being told, when I was a small child, about the shape of snowflakes, and looking at some of them through a simple microscope.  It was something of a revelation that there could be so much structure in something so small.  I also remember being told that no two snowflakes were exactly alike: every one was unique.  It was only a little later that I started to wonder how anyone could know that — surely examining and comparing all the snowflakes in existence was impossible.

In a somewhat similar way, we have all been told repeatedly, in a message that takes many forms, and is regularly reinforced by crime fiction and cop shows on TV, that fingerprints are unique — that the presence of a fingerprint someplace that matches Mr. X is irrefutable proof that Mr. X was there.  It is rare that we ask ourselves why we are so sure.

It should be evident, as a matter of principle, that we really cannot make the statement that fingerprints are unique with any existing, or even imaginable, scientific justification.   That statement is equivalent to saying that there is no pair of fingerprints anywhere at any time that match, a proposition that is impossible to prove.  The best we can hope for is a probabilistic statement about the extremely low likelihood of a match occurring by chance (like the statements that are made with regard to DNA evidence).

As I have noted before, even DNA evidence, generally regarded as the “gold standard” of forensic evidence, is not as soundly based as many experts would like.  An article in New Scientist suggest that fingerprint evidence, despite its common use in criminal cases for about a century, may suffer from even more serious shortcomings.

Since their first use in 1892, fingerprints have formed the heart of many criminal trials. But recently the realization has dawned that two prints deemed a match by an expert may in fact come from two separate people, due to human error, coincidence, low-quality prints or a mixture of all three.

There have been a few high-profile cases in which suspicions, or even convictions, based on fingerprint evidence have later been proved to be unsound (often based on DNA evidence).  Researchers have shown that different fingerprint analysts can come to quite different conclusions on whether given sets of fingerprints do or do not match.  A 2009 report from the National Research Council, published by the National Academies Press, sharply criticized a whole range of practices in the analysis and use of forensic evidence, including fingerprints:

But fingerprint analysts have been slow to acknowledge the problem, and still present matches to courts with no accompanying error rate – a state of affairs that shocked the US National Academy of Sciences (NAS), which presented a report on US forensic science last year.

Fortunately, it appears that some of the criticism is beginning to have a positive impact.  Some serious work is being done to determine realistic expectations for the accuracy of fingerprint analysis.

Last month the US National Institute of Justice commissioned the School of Law at the University of California, Los Angeles, to investigate error rates associated with fingerprint analysis. “Even though we have been using fingerprint evidence in court for almost a hundred years, not nearly enough is known about how often fingerprint examiners might make mistakes, or in what circumstances,” says study leader Jennifer Mnookin.

It is really imperative that this kind of work be done, and the results published.  We are, at least in principle, trying to operate a system of justice, not just trying to see how many people we can convict.

2 Responses to Fixing Fingerprints

  1. Pat Sullivan says:

    Aside from whether or not it’s true in principle that fingerprints uniquely identify an individual (which could work in a controlled biometric ID scenario), serious questions have been raised since at least 2002 about the reliability of fingerprint evidence collected from crime scenes.

    Not hard to see why – crime scene fingerprints are often little more than smudges.

    See e.g. the excellent 2002 New Yorker article “Do Fingerprints Lie”: . That article convinced me then that fingerprint evidence from a crime scene is generally too unreliable to convict in a criminal case.

    • Rich says:

      Pat raises an excellent point, which I should have mentioned. Fingerprints, as forensic evidence, are often used as a means to identify a person — sometimes the only means. On the other hand, they could be used as part of an identification system, to confirm identity as “part of the puzzle”.

      This comes down to the diabolical muddle people often get into when thinking about conditional probabilities. In the first, forensic, case, the probability we want is: given that these fingerprints appear to match, what is the probability that no other person could be a match. In the second case, it is something like: given that this person looks like the description, has an identification token, knows the password, and has matching fingerprints, what is the probability that he could actually be someone else.

      As you point out, this is not the first time these questions have been raised. Thanks for the link to the New Yorker article. I haven’t read that one, but I will be sure to do so.

%d bloggers like this: