So many ways for the lab to screw up.  Contamination. Even not testing at all and just issuing a report that says guilty. But the adoration of science as a means of creating objective proof of guilt means that it usually goes unnoticed, works its miracles.

“I don’t think people are evil,” Hampikian says, “but once they’re convinced of a story they protect it.” On the phone with Chris Tapp, he’s explaining why it’s been so hard to overturn his conviction. Yet Hampikian thinks recent scientific advances in DNA forensics could still free him.

There’s junk science. Tons of it, and they fight to keep it going despite all rigorous scientific knowledge informing us that it’s total garbage.  But then there’s good science.  Like DNA, which everybody knows is the gold standard.  Then again, nothing gets us into trouble like “everybody knows,” since everybody knows squat.  Greg Hampikian, on the other hand, knows a lot. But he’s not “everybody.”

Hampikian, who holds joint appointments in biology and criminal justice at BSU and heads the Idaho Innocence Project, has been helping people like Tapp for more than 20 years. He works with defense attorneys and police around the world, trying to free innocent people by exploiting the power of DNA forensics—or by exposing its pitfalls. As the nation’s only Innocence Project director who’s also a scientist, “he’s absolutely essential to what we do,” said Aimee Maxwell, executive director of the Georgia Innocence Project in Decatur, which Hampikian co-founded.

DNA appears to be tracking a path similar to fingerprints, once the irrefutable darling of forensics, but shown to be as much of a problematic mutt in the Brandon Mayfield fiasco as most other forensic science.  After 100 years of being the gold standard, fingerprints turned out to be suspect, lacking proven protocols and easily subject to bias. Of course, that doesn’t do much for the people convicted based upon fingerprint identification over that century. So sorry.

And we’re doing it again, this time with DNA.

DNA evidence is so powerful because it has firm roots in science and is backed by statistics. Analysts focus on 13 or more places in the genome, called loci, where humans are extraordinarily diverse. Each locus contains a “short tandem repeat,” a bit of DNA that is repeated multiple times. The exact number of repeats at each locus varies from person to person and can range anywhere between the low single digits to the mid-50s. Because we get one copy of each chromosome from our mother and one from our father, there are two numbers for each locus, which appear as peaks on an electropherogram, a chart produced by a genetic analyzer.

Sounds kosher, right?

Its accuracy has made DNA evidence virtually unassailable. A landmark report published by the National Research Council in 2009 dismissed most forensics as unproven folk-wisdom but singled out DNA as the one forensic science worthy of the name. Yet in recent years Hampikian and other geneticists have begun to question the technology. Thanks to a series of advances—including the polymerase chain reaction, which can multiply tiny amounts of DNA—it’s now possible to detect DNA at levels hundreds or even thousands of times lower than when DNA fingerprinting was developed in the 1980s. Investigators can even collect “touch DNA” from fingerprints on, say, a glass or a doorknob. A mere 25 or 30 cells will sometimes suffice.

As science gets “better,” new problems arise, but the law is simple: DNA good. After all, it’s science, and everybody loves science, especially lawyers and judges who neither grasp science nor feel inclined to mess with it and get a headache. But to scientists, science isn’t so simple. What could possibly go wrong?

This heightened sensitivity can easily create false positives. Analysts are picking up DNA transferred from one person to another by way of an object that both of them have touched, or from one piece of evidence to another by crime scene investigators, lab techs—or when two items jostled against each other in an evidence bag.

The risk of contamination is a perennial problem, and can come from a multitude of sources. Whether an item is touched by numerous people before it ends up in the cop’s evidence bag, or passed around among investigators who stare at it, pondering whether it’s worth putting in the baggie, or even the lab techs who work with it, all have DNA. All have the ability to take a bit of science and ruin it.

But improvements to DNA sensitivity have taken an extant problem and made it a nightmare.

DNA analysis can become even trickier when a mix of DNA from various potential suspects is found in a single crime scene sample. With a simple sample, analysts look at two sets of peaks at a given locus: one for the victim and one for the perpetrator. With mixtures, they’re looking at bunches of peaks, with no indication of which pairs go together, or which source they came from—aside from those of the known victim. At that point the analysis becomes highly subjective.

In the early days of fingerprints, six or seven “points” was considered absolute certainty.  Of course, they learned this wasn’t true, and for the point counters (as opposed to the holistic “I know it when I see it” experts), there has never been a standard that reaches an acceptable level. The FBI clouds it in typical pseudo-science gibberish, as if cool machines and programs with really science-y names makes it more official, but it’s all bullshit as it ultimately comes back to someone pulling a conclusion out of his butt.

Studies have confirmed this. In 2013, geneticist Michael Coble of the National Institute of Standards and Technology in Gaithersburg, Maryland, set up a hypothetical scenario in which a mix of DNA from several people had been found on a ski mask left at a crime scene after a series of robberies. Coble asked 108 labs across the country to determine whether a separate DNA sample, which he posited had come from a suspect in the robberies, was also part of the mix. Seventy-three of the labs got it wrong, saying the suspect’s DNA was part of the mix when, in fact, it was not. “It’s the Wild West out there,” Coble says. “Too much is left to the analysts’ discretion.”

That the government’s DNA expert, who will very thoughtfully explain why the likelihood of error is infinitesimal, will have a huge impact on the jury is obvious. Not only is the science outside the juror’s sphere of knowledge (if it wasn’t, there would be no need for an expert), but it shifts the unpleasant burden of finding guilt off the jury and onto the science guy.  Jurors love to have someone hand them conclusive guilt on a platter.

And judges, the gatekeepers of DNA evidence, even when they’re willing to risk permanent brain damage by thinking hard about the issues, are extremely reluctant to make problems with the government’s most persuasive and conclusive evidence.  After all, it’s DNA. It’s science. It’s probably right, because it’s the gold standard. Everybody says so.