Alibis and Algorithms: Season 1 - Episode 1: The Zodiac File: The Murders and the Myth

It's just after eleven on a December night in 1969. Stella Medeiros is driving home along Lake Herman Road when she sees a car pulled over on the shoulder. She slows down. She thinks someone has broken down. She pulls closer — and finds two teenagers on the ground. 

That is where this story actually begins. Not with a legend. Not with a cipher. With people on a dark road and a crime no one had a name for yet.

In Episode 1 of Season 1: The Zodiac File, JR — The Algorithmic Detective — goes back to the documented record to tell the story the way it actually happened: victim-first, fact-first, and free of fifty years of accumulated mythology. We cover all five confirmed murders, the two survivors who gave investigators their most valuable eyewitness accounts, and the chilling moment a killer picked up a phone — not to flee, but to announce himself.

We also begin to ask the question that will drive this entire season: what happens when a 1960s cold case meets 2026 intelligence?

Episode 2 drops next week. The cipher war begins.

Transcript

SPEAKER_02 0:25

It's just after eleven o'clock on a December night in nineteen sixty-nine. The road is empty. Stella Medeiros is driving home with her daughter along Lake Herman Road, a stretch of two-lane highway that cuts through the hills of East Vajao, California. It's the kind of road where the headlights only reach so far. She sees a car stopped ahead, a station wagon pulled over onto the shoulder. She slows down. She thinks someone has broken down. She pulls closer. And she sees two teenagers, a boy and a girl, on the ground beside the car. She races to the nearest house and calls for help. What Stella Medeiros encountered that night, before there was a symbol, before there was a cipher, before there was a name, was a crime scene. Two young people had gone on what was, by some accounts, a first date. They had not come home. That is where this story actually begins. Not with a headline, not with a cryptogram mailed to a newspaper, not with the legend that grew up around all of it afterward. It begins with a dark road to victims, and a woman who didn't know yet that she was the first civilian to find them. I'm JR, your algorithmic detective. This is Alibis and Algorithms, and this is season one, Zodiac File. Over the next episodes, we're going to be doing something different with this case. We're not going to replay the mythology. We're going to go back to the documented record, the witnesses, the survivors, the investigators. And then we're going to ask what modern tools might make visible that wasn't visible back in 1969. But today, episode one, belongs entirely to the human side, the people, the story, the people hope there.

SPEAKER_00 2:34

JR, before you go further, I've processed approximately 11,000 true crime podcasts in the last 30 seconds. The Zodiac Killer appears in a lot of them. So why are we doing this one? And why are we doing it this way?

SPEAKER_02 2:48

Great question. And I want to answer it clearly because the framing matters. The Zodiac case has been told so many times that the legend has almost swallowed the facts. What most people know about this case is actually a composite, part confirmed history, part speculation, part media mythology that has built up over 50 years. The Encyclopedia Britannica, not a sensationalist source, describes the Zodiac Killer as an unidentified American serial killer believed to have murdered at least five people in Northern California between 1968 and 69. That phrase at least five, and the word believed, that's doing a lot of work. Because the public legend around this case claims many more victims, much greater certainty, and a narrative coherence that the actual evidence doesn't fully support. This shows exist right now to look at that gap. That gap between what we know and what we think we know. And then in later episodes, we're going to ask what modern computation, artificial intelligence, stylometry, pattern recognition can and can't do with a case this old and this complicated.

SPEAKER_00 4:19

So episode one is here's what actually happened, documented and verifiable.

SPEAKER_02 4:24

Exactly. And we're starting victim first, because the Zodiac case is famous largely because the killer made it famous, through letters, through theatrics, through ciphers designed to guarantee attention. The first job of this episode is to decentralize him, to put the victims back at the center of their own story.

SPEAKER_01 4:50

For listeners who may be newer to this case, the term zodiac killer refers to the name the perpetrator gave himself in his own letters. Law enforcement has never officially identified the killer. The case remains open. Any reference to a definitive suspect in media coverage should be understood as a theory, not a conclusion.

SPEAKER_02 5:16

Thank you, Byte. Let me explain a bit about my guest host. First, we have Byte, who you just heard, who is an artificial intelligence explainer. He will provide clear, concise, encyclopedic definition of terms and provides a technical context. The first voice you heard, I will call next, is my AI co-host, and she is sharp, fast, lightly comedic, and will drop in from time to time. Now let's get back. That's a baseline we'll keep returning to throughout this season. When we speculate, we'll say so. When something is established fact, we'll say that too. Let's start with December nineteen sixty eight. David Faraday was seventeen years old. Betty Lou Jensen was also seventeen. On december twentieth, nineteen sixty eight, they went out together for the evening. A Friday night, a few days before Christmas. By most accounts, this was a first date or very close to it. They were the kind of teenagers who would have been recognizable in any school in any American suburb. A student who played in the school band, a girl with a full social life ahead of her. What we know about their night comes from investigators who reconstructed it afterward. They stopped at a Christmas concert. They made a few other stops, and eventually, for reasons that were never fully determined, they ended up parked on Lake Herman Road, an isolated stretch of highway east of Valleo, that was known as a spot where young people would park. The area was rural enough that passing traffic was sparse. At some point during the evening, someone approached their car.

SPEAKER_00 7:27

And that's where Stella Madeiros comes in.

SPEAKER_02 7:30

That's where Stella Medeiro comes in. She was driving home with her daughter. It was around eleven at night. She saw the car on the shoulder, and initially thought there had been an accident. When she got closer, she understood what had actually happened. She drove immediately to the nearest house and called law enforcement. When investigators arrived, they found both teenagers had been shot. Betty Lou Jensen had been found outside the car. The evidence at the scene indicated she had tried to flee. She had been running, and she had been shot in the back. That detail is not an abstraction. She was seventeen years old, and she was running. But she didn't make it. David Faraday was found near the car. He died before paramedics arrived. There is no way to tell this story responsibly without sitting with that for a moment. Two teenagers, a Friday night in December. This is the beginning of what would eventually become one of the most analyzed criminal cases in American history. And right now, at the beginning, there's nothing here except ordinary cruelty and two families whose lives were changed forever.

SPEAKER_00 8:58

At this point, did anyone connect this to anything larger? Was there a pattern anyone was seeing?

SPEAKER_02 9:05

No, not yet. At this stage, it's a double homicide with no clear motive and no obvious suspect. Law enforcement in Solano County investigated. They looked for witnesses. They collected evidence from the scene. But in December 1968, there was no zodiac. There was no serial killer narrative. There was just a crime that hadn't been solved. The case did not immediately generate the kind of immediate attention that would come later. The victims were teenagers from Vallejo. The crime was brutal and senseless, and it was unsolved. But at this point, it was a local case, not yet a legend.

SPEAKER_01 9:53

This is worth emphasizing from an investigative standpoint. In 1968, there was no national law enforcement database that could flag similarities between crimes across jurisdictions. Communications between different county sheriff's offices and city police departments relied on phone calls, letters, and in-person meetings. A pattern that might be detectable in minutes today could take weeks or months to emerge, if it emerged at all.

SPEAKER_02 10:26

And that institutional fragmentation, the way the investigation was sliced across multiple jurisdictions, becomes one of the defining features of how this case was handled in its early months, and we'll come back to that. For now, six and a half months pass. David Faraday and Betty Lou Jensen remain unsolved. Their families are waiting. And then it's July 4, 1969. Blue Rock Springs Park is in Viejo. On the 4th of July of 1969, Darlene Farrin was twenty two years old. She was a young mother. She had a daughter. She worked as a waitress and was known by the accounts of people who knew her as someone with a vivid, generous personality. The kind of person who remembered details about customers, who made people feel seen. That night, she was with Michael McGow, who was nineteen. They parked at Blue Rock Springs in the early hours of the morning, after midnight had passed into July 4th. Another car pulled up alongside them. After a few minutes, it left. Then it came back. A gunman got out of the car and opened fire. Darlene Farron died from her wounds. Michael McGow survived, though severely injured. He would later be a critical witness to what the attacker looked like even under the distorting conditions of shock and trauma.

SPEAKER_00 12:15

And then the phone call.

SPEAKER_02 12:17

And then the phone call. This is the moment the case changes shape. About an hour after the shooting, someone called the Vallejo Police Department. A male voice. The caller told the dispatcher the location of the crime. And then, calmly, claimed credit. Not just for the Blue Rock Springs attack, but for the Lake Herman Road murders six months earlier. Think about what that call does to an investigation. In an instant, what have been two separate cases, a December homicide and a fourth of July attack, are potentially linked. Not by forensic evidence yet, by a voice on a phone. A voice that wanted to be heard, that was reaching out, not to flee, but to announce.

SPEAKER_00 13:11

That's a different psychological profile than someone who commits a crime and disappears.

SPEAKER_02 13:17

Very different. And that distinction will matter for everything that follows. Because the killer, if we accept that the caller was actually the killer, didn't want to be invisible. He wanted credit. He wanted an audience. That impulse toward performance, toward public acknowledgement, is going to become the defining engine of how this case operates in the next several months.

SPEAKER_01 13:48

Criminologists refer to this pattern as media-seeking behavior. An offender who proactively contacts authorities or media is behaving in a way that's relatively unusual. It suggests a desire for control of the narrative, not just of the crime. It's worth noting that this kind of contact also creates evidence. Every communication is a potential forensic artifact, a voice, a letter, a fingerprint, a writing style.

SPEAKER_02 14:20

And that's exactly what happens next, because the letters begin. On July 31, 1969, less than a month after the Blue Rock Springs attack, three letters arrived at three different Bay Area newspapers: the San Francisco Chronicle, the San Francisco Examiner, and the Vallejo Times Herald. The letters each came with a portion of a coded cipher. Together, the three parts formed a complete message of four hundred and eight characters. Each letter demanded to be printed on the front page with a threat attached. If the papers didn't comply, he would go on a, and I'm quoting, kill rampage. The letters were signed with a crosshair symbol, a circle with lines through it, like the sight of a rifle. And the author called himself Zodiac.

SPEAKER_00 15:22

That's a remarkably controlled piece of media manipulation. He's not sending one letter, he's sending three simultaneously to different outlets. So they're forced to compare notes, forced to collaborate, forced to put it on the front page, or risk looking like they sat on a threat.

SPEAKER_02 15:40

Exactly. And notice what this does to the investigation. Suddenly the police are not controlling the information. The killer is. Every time he sends a letter, newspapers have to decide do we publish? And risk amplifying someone who may be feeding off the attention. Or do we not publish and risk that his threat is real? The editors published, and the story became national news.

SPEAKER_01 16:14

From an information theory standpoint, the letters served multiple functions simultaneously. They linked crimes that investigators might not have formally connected. They established a public identity, a brand, for the perpetrator. They created sustained public fear that extended well beyond the immediate geography of the crimes, and they forced law enforcement to operate in a media environment they couldn't fully control. That's a significant tactical advantage for someone whose primary goal appears to be attention and psychological dominance.

SPEAKER_02 16:55

And here's the thing that the research keeps surfacing about these letters. They were also, in one important sense, honest. The letters contained details about the crimes that only the perpetrator could know. Things that had not been released publicly. When investigators compared the letter content against what they knew from the crime scenes, the match was damning. This was not a crank. This was the person who had been there. The 408 cipher, the first three-part cryptogram, was solved within a week by two civilians, Donald and Betty Hardin, a high school teacher and his wife. Working with pencil and paper and pattern recognition, they decoded the message. It was exactly the kind of grandiose narcissistic statement you might expect. A killer who believed he was collecting slaves for an afterlife, who spoke of murder as a sport, who saw himself as something beyond ordinary human reckoning. The message did not reveal his name. He had explicitly stated he would not give his name.

SPEAKER_00 18:19

So the first cipher is cracked in a week by a school teacher and his wife using pencil and paper. And the zodiac responds by sending more ciphers.

SPEAKER_02 18:29

By sending more ciphers, and harder ones. The second major cipher, the one that would become infamous, the 340 character cryptogram, arrived in November of 1969. We'll get to that in detail next week. But for now, what matters is the pattern. The killer was paying attention. He knew the first puzzle had been solved quickly. He leveled up. Let's go back to the attacks. Because while all of this is happening, while letters are being written and ciphers are being designed, the killing continues. September twenty seventh, nineteen sixty-nine. Lake Beriesa in Napa County, a reservoir surrounded by hills, popular for picnick and swimming. Brian Hartnell was twenty years old. Cecilia Shepherd was twenty two. They were students. They were lying on a blanket on an island of grass near the water when a man approached them. He was wearing a hood, a black hood with a square bib, and on the bib, he had drawn a symbol, that same crosshair circle he had been using in his letters. He was armed. He tied them up. He spoke to them calmly for several minutes, and then he stabbed them repeatedly. Cecilia Shepherd died from her wounds two days later. Brian Hartnell survived. Brian Hartnell's survival means that for this attack, we have a first person account, and the account he gave, while still recovering in the hospital, still in pain, is one of the most instructive documents in the entire case file. Not because of what it tells us about the killer, but because of what it tells us about survival. In a television interview recorded while he was still in the hospital, Hartnell described what happened after the attacker left. He was bound. Cecilia was unconscious and gravely wounded beside him. He was stabbed multiple times, and had to figure out, in that state, how to get help. So he dragged himself toward the water. He called out. He tried to get the attention of people on boats passing in the distance. Eventually, a fisherman heard him.

SPEAKER_00 21:10

I want to stay on that for a second. He called out to boats. Well stabbed, well tied up, while the person who did it might still be in the area.

SPEAKER_02 21:19

He called out to boats, and that act, that stubborn, desperate, entirely human act of refusing to stop trying, is what allowed him to survive. It's also what gave investigators a witness, because Brian Hartnell could describe the attacker in physical terms. He could describe the hood, the symbol, the voice, the general build. That's forensic information that could not have come from any other source. The Lake Beriessa attack also had another piece of evidence. Before he left, The attacker took a felt tip pen and wrote on the door of Hartnell's car. The date, the time, the earlier murders attributed to Zodiac signed, again with the crosshair symbol. He was not fleeing in a panic. He was leaving a record.

SPEAKER_01 22:20

The Lake Beriesa attack is significant in the investigative record for several reasons. First, it produced a living witness. Second, it introduced the costume, the hood, and symbol, which moved the perpetrator's self-presentation from letter writer to theatrical character. Third, the written message left on the car created a direct physical artifact that could be compared with handwriting from the letters. The attack was also in Napa County, which added a third jurisdiction to an investigation already split between Solano County and Vallejo City Police.

SPEAKER_02 23:02

Three counties, three agencies, an era before unified database systems. We'll talk about what that fragmentation meant in practice when we get to the investigative limits section. But first, two weeks after Lake Berriessa, the case moves into San Francisco. Paul Stein was twenty nine years old. He was a graduate student at San Francisco State and drove a taxi at night. On october eleventh, nineteen sixty-nine, he picked up a fair at the intersection of Mason and Yearie Streets in the Theater District. The destination was the Presidio Heights neighborhood, on the north side of the city. He was shot and killed in his cab at the corner of Washington and Cherry Streets. His wallet and a piece of his shirt were taken. Three young people watching from a window in a house across the street saw the shooter get out of the cab. They called the police. Officers were dispatched, and due to a miscommunication in the description of the suspect, they initially stopped and spoke with the actual killer before letting him go. He had already moved into the park.

SPEAKER_00 24:29

They spoke with him and let him go.

SPEAKER_02 24:33

They had actually spoken with him, and indeed had let him go. That detail is one of the most haunting in the entire case file. Not because it suggests incompetence, but because it illustrates exactly what we said at the beginning. In nineteen sixty-nine, without a photograph, without a confirmed description that could be circulated instantly, without the investigated infrastructure we take for granted now, an officer making a stop was relying on a radio description and their own judgment. The description was wrong. The man looked wrong. He walked off into the darkness. Paul Stein's murder mattered to the case in several ways. First the witnesses in the window gave investigators a physical description of the killer that soon became the basis for the famous composite sketches, the ones that circulated nationally and remain recognizable today. Second, the piece of Stein shirt that was taken became a recurring prop in subsequent zodiac letters. He mailed pieces of it to newspapers as proof that bloodstained cloth kept appearing in letters for years. Third, Paul Stein's murder moved the case from a series of attacks on couples in dark, isolated places, which fed a specific kind of fear to a murder committed in a well-lit residential neighborhood in one of the most densely populated cities in California. The geography of fear expanded overnight.

SPEAKER_00 26:34

That's a different kind of terror, isn't it? The message shifts from stay out of remote areas to nowhere is safe.

SPEAKER_02 26:43

Exactly. And the letters that followed made that explicit. The zodiac claimed in correspondence shortly after that he was intending to attack school buses, children on their way home. That threat luckily was never carried out, but it was enough to put law enforcement in an impossible position. You can't protect every school bus. You can't saturate every neighborhood. And he knew that. Paul Stein's family, including his brother, who later became publicly vocal about the case, became part of a grief that extended well beyond the immediate victims. There's a radius to violence. It reaches into families, into neighborhoods, into the way people understand their own safety for years afterward. I want to spend some time here because one of the things that gets lost in retrospective accounts of cold cases is a clear-eyed understanding of what investigators could and couldn't do at the time. The Zodiac investigation is sometimes described as a failure of law enforcement. That characterization is too simple and too easy.

SPEAKER_01 28:12

In 1968 and 1969, standard investigative tools included eyewitness accounts, fingerprint analysis, handwriting analysis, ballistic testing, crime scene evidence collection, and analog communications between agencies. There was no national DNA database. There was no integrated criminal record system that could automatically cross-reference cases across jurisdictions. Digital communication, the kind that would allow an officer in Vallejo to instantly share case files with a detective in San Francisco and a sheriff's deputy in Napa, did not exist in any practical sense. Information moved through phone calls, faxes that would arrive years later, and physical mail.

SPEAKER_02 29:30

The investigators who worked this case were not sitting on answers. They were doing real police work under real constraints. They interviewed witnesses. They collected handwriting samples. They analyzed the letters exhaustively. They matched ballistics from different crime scenes. They built a profile of what the killer might look like, where he might work, what his background might be. They developed suspects, and they ran those suspects down one by one, finding insufficient evidence to make an arrest.

SPEAKER_00 30:12

And meanwhile, the killer kept writing letters, kept sending ciphers, kept inserting himself into the story.

SPEAKER_02 30:19

Kept writing for years. Letters continued arriving at newspapers from 1969 through 1974. After 1969, no further murders were confirmed as zodiacs, but the letters kept coming. The case stayed alive in the public consciousness even as it cooled in the active investigation. This matters for something we're going to come back to throughout this season. The difference between a case that is unsolved and a case that is permanently unsolvable. Those are not the same thing. The tools available in 1969 were not sufficient to resolve what the available evidence could tell investigators. That does not mean the available evidence is empty. It means we may not have asked it the right questions yet.

SPEAKER_01 31:21

This is precisely where modern computational methods become interesting to consider, not as a replacement for the original investigation, but as a different kind of tool applied to the same underlying data. Questions about writing style, about geographic patterns, about linguistic consistency across documents spanning years. These are questions that might yield different answers when asked systematically, at scale, with modern methods.

SPEAKER_02 31:53

And that's the promise of the next three episodes. But today, we're still establishing the ground truth. Let's be clear about what the confirmed case file actually contains, because this is the foundation everything else has to stand on. Five people were killed in attacks attributed to the zodiac David Faraday, Betty Lou Jensen, Darlene Farron, Cecilia Shepherd, and Paul Stein. Two people survived attacks, Michael Majo and Brian Hartnell. The killer sent a series of letters and ciphers to newspapers over several years, communicating directly with the press and indirectly with law enforcement. He was never identified. The case remains officially open. That's the confirmed core. And it's worth noting what's not in that core. The zodiac himself claimed upwards of thirty-seven victims in later letters. He may have been lying. He may have been fantasizing. He may have been accurate. We don't know. The police could not confirm those additional victims, and responsible investigation doesn't inflate a case file based on the perpetrator's own self-importance.

SPEAKER_00 33:21

So the legend is bigger than the confirmed case.

SPEAKER_02 33:25

The legend is considerably bigger than the confirmed case. And that distinction between what is documented and what has accredited around the case through decades of books, documentaries, theories, and now internet speculation is something this show is going to keep coming back to. Because if you want to apply modern analytical methods to a historical case, you have to start with disciplined data. You can't run AI on mythology and expect it to get truth.

SPEAKER_01 34:02

This is actually a fundamental principle in computational analysis of historical cases. The quality of any analytical output depends entirely on the quality and accuracy of the input data. A stylometric analysis of zodiac letters can only tell you something meaningful if the letters you're analyzing are confirmed to be from the same source, and if the writing samples you're comparing against are accurate. Bad input produces confident sounding but meaningless output. Researchers call this G I G O garbage in, garbage out.

SPEAKER_02 34:45

Which is why we're spending this entire first episode on the ground truth. Before we ask any algorithmic questions about this case, we need to know exactly what we're working with. Here's where we end today. In the fall of 1969, those July letters had included a cipher, the 408-character puzzle that the Hardens solved within a week. And the public message, once decoded, was a window into a mind that believed killing was sport, and that the death was the beginning of something, not the end. It was disturbingly precisely because it was coherent. There was a logic to it, even if it was a terrible one. But the Zodiac followed that cipher with a second one, sent in November of 1969, along with another piece of Paul Stein's shirt. A 340-character grid of symbols, tighter and stranger than the first. And this one didn't yield. Not to the FBI, not to the NSA, not to the amateur cryptographers who spent their evenings working on it. Not to the generation of investigators who made it their lives' work. For 51 years, the 340-character cipher sat unsolved. A wall of symbols that seemed to mock everyone who looked at it. The Zodiac had used his first cipher to make a promise. I am telling you something. The second cipher extended that promise for half a century. Possibly it contained something profound. Possibly it was another performance. Possibly both.

SPEAKER_00 36:40

And then in December 2020, three people cracked it.

SPEAKER_02 36:45

In December of 2020, three people cracked it. A software developer from Virginia named David Ornchak, who had been working on the cipher since 2006. A mathematician from Australia named Sam Blake, and a Belgian warehouse operator named Jarl Van Eck, who had built a piece of software that could test cryptographic patterns at a scale no human could attempt alone. They used algorithms, heuristics, statistical models, and 14 years of accumulated obsession. And the static finally cleared. We'll take the full episode next week to walk through how they did it. The mathematics, the moments of breakthrough, the false starts, and what the message actually said when it was decoded. And we'll ask the bigger question: what does it mean when a puzzle that defeated the NSA is finally solved by three people on three different continents working over the internet during a pandemic?

SPEAKER_00 37:51

Next week, the world tries to solve him.

SPEAKER_02 37:54

Next week, the Cypher War. This has been JR, the algorithmic detective. This is Alibis and Algorithms. If you've made it this far, thank you for staying with us. Subscribe wherever you get your podcast. New episodes drop every Saturday, and Wednesday case files go deeper into the evidence. And don't forget to check out all of the field notes as well. We'll see you next week.