Chapter 6 — Key Takeaways

A one-page card. If you remember nothing else from this chapter, remember this.

The core claims

  • The field was forced to reckon with itself from outside. It did not reform after sober self-examination; it was confronted by an external technology (DNA, applied retroactively to old convictions) and then by external scientists (the NAS and PCAST reports).
  • DNA proved wrongful conviction is real and common — by exclusion, not proof. Post-conviction DNA testing has produced several hundred U.S. exonerations (the DNA-only figure passing 375), and those are only the cases that happened to leave testable biology. The true scope is larger.
  • Flawed or misapplied forensic science is a contributing factor in roughly half of the DNA exonerations — not a rare aberration. It splits into junk methods (bite marks, hair, etc.) and valid methods overstated in testimony.
  • The leading causes of wrongful conviction interlock. Mistaken eyewitness ID (the most common), flawed forensics, false confessions, informants, and official misconduct reinforce each other in a bias cascade — no single actor need be dishonest for the system to convict the wrong person.
  • The two reports are the field's diagnosis. NAS 2009: forensic science is "badly fragmented," and apart from nuclear DNA, no method had been rigorously shown to link evidence to a specific source. PCAST 2016: most feature-comparison methods lack demonstrated foundational validity; bite marks fail outright; even fingerprints carry a real, non-zero error rate.

The method-validity verdict

Question Verdict
Is "wrongful conviction" a fringe problem? No. Hundreds documented by DNA alone; the countable cases are a biased, undercounted sample.
Did forensic science contribute? Yes — routinely. A factor in ~half of DNA exonerations.
Are all forensic methods equally valid? No. Foundational validity is measured (or absent) method by method; DNA at the top, bite marks discredited.
Is "zero error rate" ever a valid claim? No. It is a refusal to measure, not a measurement.
What is the book's yardstick? The validity spectrum — position is a ceiling, it is question-specific, and the honest verb tracks the position.

Key terms (one-liners)

  • Wrongful conviction — conviction of a factually innocent person (not a mere procedural reversal).
  • Exoneration — official undoing of a conviction on the ground the person didn't commit the crime.
  • The Innocence Project — founded 1992 (Scheck & Neufeld); frees the wrongly convicted via post-conviction DNA and reforms the causes.
  • Junk science — a method presented as valid science with no measured error rate or validated basis; misleads even an honest examiner.
  • NAS 2009 reportStrengthening Forensic Science in the United States; "badly fragmented," only nuclear DNA validated.
  • PCAST 2016 report — feature-comparison methods; introduced foundational validity vs. validity as applied.

What you can honestly say on the stand

"Decades of post-conviction DNA testing have established, by clean exclusions, that forensic science has contributed to wrongful convictions in roughly half of those cases — through both methods that were never scientifically validated and valid methods overstated in testimony. The 2009 NAS and 2016 PCAST reports found that, apart from nuclear DNA, most feature-comparison methods lacked demonstrated foundational validity, and that no human-judgment method has a true zero error rate. The strongest claim I can support for a method is the one its validation studies measure — no more — and where a method has no validation, the honest answer is that the science does not support the comparison."

The one-sentence reflex to carry forward: Before trusting any forensic claim, ask where the method sits on the validity spectrum, what claim is being made, and whether the validation evidence supports that claim — because that question, not any expert's confidence, is what protects an innocent person.