Case Study 2 — The Golden State Killer: What Real, Validated Forensic Science Can Do

The counterweight to Case Study 1. If Mayfield shows a "strong" discipline failing through overstated certainty, this case shows the field's most rigorous method, used carefully, doing something genuinely extraordinary. It anchors the book and returns in Chapters 7, 8, 29, and 39.

Background

Between 1974 and 1986, a single offender committed a long series of burglaries, rapes, and murders across California — crimes attributed at the time to what seemed like several different criminals ("the Visalia Ransacker," "the East Area Rapist," "the Original Night Stalker"). The cases went cold for decades. The offender was eventually understood to be one man, dubbed the Golden State Killer, responsible for at least 13 murders and dozens of rapes. For more than thirty years, investigators had crime-scene biological evidence but no name to attach to it.

The forensic evidence and the method

The break, in 2018, did not come from a faster fingerprint search or a sharper eyewitness. It came from a chain of validated science used with discipline:

  1. A preserved DNA profile. Crime-scene biological evidence had yielded a DNA profile years earlier. Conventional database searching (CODIS, Chapter 7) had never returned a match, because the offender's profile was not in the law-enforcement database.
  2. Investigative genetic genealogy (IGG). Investigators uploaded a profile to GEDmatch, a public genealogy database where people compare DNA to find relatives. The search returned not the offender but distant relatives — people who shared enough DNA to be third or fourth cousins.
  3. Family-tree reconstruction. Genealogists built family trees outward and inward from those distant matches, using public records — a painstaking, weeks-long process of triangulation — until the trees converged on one branch and, finally, one plausible individual: Joseph James DeAngelo, a former police officer.
  4. Confirmation by direct comparison. Crucially, the genealogy did not convict anyone. It produced a lead. Investigators then collected DeAngelo's discarded DNA (from items he threw away) and compared it directly to the crime-scene profile. That comparison — the rigorous, quantified one — is what tied him to the crimes. He was arrested in April 2018 and later pleaded guilty.

What it did, and didn't, establish

This case is the book's emblem of validated forensic science, and it earns the title precisely because of how carefully the strong method was used:

  • The validity spectrum, top end (§1.5). DNA comparison sits at the rigorous end of the spectrum because it rests on population genetics and yields a quantified random match probability. That is what made the final confirmation trustworthy in a way a visual "match" never could be.
  • Genealogy as lead, not proof (§1.6). The genealogical work was investigative — it told police where to look. It did not, and could not, establish guilt by itself. The honest structure was: genealogy narrows the world to a name; direct DNA comparison then tests that name. Lead, then confirmation. Mixing the two up — treating the genealogical inference as the proof — would have been the Mayfield error in a new costume.
  • Exclusion at scale (§1.6). Along the way, the same techniques excluded other relatives and candidates. The method's power to rule people out is inseparable from its power to point.

A caution the book will develop (Chapters 8, 29, 38): the very power of IGG raises real privacy and consent questions — your DNA in a database can expose a relative who never consented. That validated science works does not settle whether, or how, it should be used. Power and ethics are different questions, and this book insists on asking both.

Discussion questions

  1. Genetic genealogy produced a lead; direct DNA comparison produced the evidence. Why is keeping those two steps distinct an example of this chapter's core discipline?
  2. Conventional database searching failed for decades, then genealogy succeeded. What does this say about the difference between a method's power and its availability?
  3. Contrast the certainty language appropriate to the final DNA comparison here with the certainty language that failed in the Mayfield case. Why is one defensible and the other not?
  4. The method that solved this case also creates privacy concerns. Is "it works" a sufficient argument for using a forensic technique? Why might the book insist otherwise?
  5. Place this case and the Mayfield case at their respective ends of the validity spectrum, and write one sentence explaining what separates them.