Somebody keeps drinking the single source Ethiopian yirgacheffe coffee that I bring to the office for myself.

I rounded up five suspects from around the office and fingerprinted them. To see if I could find a match from a print, I found on my mug.

Is fingerprint analysis reliable enough to pin somebody to this unspeakable crime, and more importantly... uhhhh... how do you actually do this?

To help me figure out what the hell I'm supposed to do, I brought in an expert.

My name is Peter Valentin.

I'm an expert in crime scene reconstruction and forensic science.

A fingerprint is probably the most important piece of information when we don't have a known connection between the victim and the offender.

We have to find something that people leave at scenes relatively easy and something that is unique enough that finding it and identifying what it is will lead us back to the person.

In order to find a fingerprint, investigators will either use a physical component like a powder–

So not cocoa powder?

Cool cool cool cool.

Or a series of chemicals that make a print visible on a surface.

Our fingerprints have ridges and furrows that align to create unique, recognizable patterns.

There is a what's called a loop right here, you have ridges that are coming into the pattern and then coming back out.

Here's another loop right here.

There's arches, and there's whorls and those those three categories alone comprise at least 95 percent of the fingerprints that are in the database.

Everyone's fingerprints have unique features that differentiate them from everyone else's.

For instance, this ridge splits into two here and this ridge has a break in it.

These patterns remain the same throughout a person's lifetime, which makes them a powerful identification tool.

My fingerprint is pretty easy to spot.

The pattern looks like a 1998 Chrysler Sebring.

Investigators rely on the characteristics of a print to find a match using a method called ACE-V.

First, they analyze the print and the surface it's on to see if it's viable for examination.

This one's great.

This one, not so much.

Then, they take the suitable print and compare it against known fingerprints looking for points of similarities and differences.

For example both these patterns are whorls and bifurcate right here.

Investigators look at these traits and try to gauge if there's enough points of comparison to declare it a match.

Finally, a qualified peer reviews their conclusion to verify the match.

After using high tech software to analyze the prints from my mug – I'm kidding this is just some stock footage I downloaded.

I think I finally found a match.

But, ok, even when it's done by professionals... how reliable is fingerprint analysis?

It's a useful tool that obviously has value, but I think it's problematic to overstate its value, especially in a criminal justice context.

One accuracy study found that examiners made false positive identifications in 0.1% of cases.

Which means identifying a print as a match when it's not.

That doesn't sound too bad... unless you happen to be one of those false positives.

Brandon Mayfield, a lawyer from Oregon, was one of those unlucky few.

He was falsely accused of the 2004 Madrid train bombing because the FBI made a false positive match with a partial print found at the scene.

Mayfield seemed like a good suspect to investigators because he had recently converted to Islam and was the lawyer for a man who attempted to join the Taliban,

which in the wake of 9/11 were all red flags for the FBI.

There was a bias that existed because once they tentatively identified the unknown fingerprint belonging to this individual,

who he was and what his background was made the identification seem stronger than it actually was.

He didn't have a valid passport.

So it's a little implausible on the face of it that he was handling a bag of detonators in Spain.

Fingerprinting was so powerful that it just sort of it trumped everything else.

Mayfield was even given his own analyst to examined the prints, and even they agreed that it was a match.

That's an interesting thing about bias because, you know, everybody's assumption is that the experts are biased toward the side that they're working for.

Here was a guy who was working for Mayfield and yet he corroborated the evidence against him which turned out to be incorrect.

The FBI released a 330-page report about where their analysts went wrong, and it's pretty much a case study in a lot of the potential downfalls of fingerprinting.

They cited things such as ignoring differences between the prints, lack of independent verification, the pressure of working on a high profile terrorism investigation,

and letting bias about the suspect affect their analysis.

Their report summed it up by saying, "in any human endeavor, there is a potential for error."

You know, I would suggest learning from history and realizing that none of these things are going to be error free.

So there's going to be mistakes and errors and screw ups.

But also evidence is inherently probabilistic.

So even though it's not a perfect tool, until a more reliable technology emerges, law enforcement will continue to utilize fingerprint analysis.

And I won't ever be fully certain of who stole my coffee.

Until I catch him on this new surveillance camera that I installed.