If You Want to Kill Someone, We Are the Right Guys

On a brisk day in March 2016, Stephen Allwine walked into a Wendy’s in Minneapolis. The smell of old fryer grease hung in the air as he searched for a man wearing dark jeans and a blue jacket. Allwine, who worked as an IT support technician, was lean and nerdy, with wire-rim glasses. He was carrying $6,000 in cash, money he’d collected by pawning silver bars and coins to avoid suspicious deductions from his bank account. He found the man he was looking for sitting in a booth.

They had connected on LocalBitcoins, a sort of Craigslist for people who want to buy cryptocurrency near where they live. Allwine opened the app Bitcoin Wallet on his phone and handed over the cash, and the man scanned a QR code displayed on the phone to transfer the bitcoin. The transaction went seamlessly. Then Allwine returned to his car to discover that he had locked his keys inside.

It was his birthday. He was 43. And he was supposed to join a woman named Michelle Woodard for lunch.

Allwine had met Woodard online a few months earlier. The relationship had progressed quickly, and for a while they exchanged dozens of messages a day. Their passion had since faded, but they still slept together from time to time. While he waited for the locksmith to arrive, he texted her that he'd stopped to buy bitcoin and was running late. Once the door was jimmied open, he met up with Woodard at a burger joint called the Blue Door Pub, determined to enjoy the rest of the afternoon.

That evening he gave himself another birthday present. Using the email address dogdaygod@hmamail.com, he wrote to a person he knew only as Yura. “I have the bitcoins now,” he said.

The Strange Life of a Murderer Turned Crime Blogger

  • How One Woman's Digital Life Was Weaponized Against Her

  • Inside China's Vast New Experiment in Social Ranking

  • Yura ran a site called Besa Mafia, which operated on the dark web and was accessible only through anonymous browsers like Tor. More important for Allwine's purposes, Besa Mafia claimed to have ties to the Albanian mob and advertised the services of freelance hit men. The site's homepage featured a photo of a man with a gun and no-nonsense marketing copy: “If you want to kill someone, or to beat the shit out of him, we are the right guys.”

    Yura promised that customers' money was held by an escrow service and paid out only after a job was completed. But Allwine worried that when he deposited money it would simply end up in someone's bitcoin wallet. He wanted Yura's claims to be true, though, so against his better instincts he transferred the bitcoin. “They say that Besa means trust, so please do not break that,” he wrote Yura. “For reasons that are too personal and would give away my identity, I need this bitch dead.”

    “This bitch” was Amy Allwine, his wife.

    Stephen and Amy Allwine had met 24 years earlier at Ambassador University, a religious school in Big Sandy, Texas. Stephen showed up freshman year with a pack of friends from his church youth group near Spokane, Washington. Amy was from Minnesota and didn't know many people at the school. She quickly attached herself to the Washington crowd. She was sunny and easygoing, and she and Stephen became regular dance partners—an activity that brought them closer, but not too close. They belonged to the Worldwide Church of God, which observed a strict Saturday Sabbath, rejected holidays with pagan influence like Christmas, and frowned on too much physical contact on the dance floor.

    In 1995, while they were still at Ambassador, the United Church of God split off from the Worldwide Church of God. Stephen and Amy joined the new sect, which embraced the internet as a means of spreading the gospel. For Stephen, who had a passion for computer science, it was a logical choice.

    After college, the couple married and moved to Minnesota to be close to Amy's family. Amy could tame even the most unruly animals, and she taught for a few years at a local dog-training school before starting her own business, Active Dog Sports Training. The couple adopted a son, bringing him home when he was just a couple of days old, and in 2011 they moved into a house in Cottage Grove, Minnesota, a centerless enclave of commuters and farmers in the Mississippi River Valley, not far from the Twin Cities. Amy converted a large agricultural shed on the property to a dog-training arena, and their house was soon a homey mess, with fur from the Allwines' Newfoundland and Australian shepherd covering the upholstery and a trail of unfinished Lego projects on the kitchen center island.

    From the outside, nothing seemed amiss. Stephen rose to the rank of elder in the United Church of God, and Amy became a deaconess. The church followed the Jewish calendar, and on Fridays the family had dinner with Amy's parents, whom Stephen called Mom and Dad. On Saturdays they attended services. Every year they traveled to join in the church's fall festival, which was held at different sites around the world. As Amy's business grew, she traveled around the country with friends to attend dog competitions. In their spare time, the Allwines maintained a website called Allwine.net, which included a list of acceptable songs and instructional dance videos showing how to have fun without excessive touching. In one, Amy wears khakis and hiking boots while Stephen is in a polo shirt and baggy jeans, and the two are line dancing to “We Go Together.”

    The day after Stephen bought the bitcoin, he uploaded a photo of Amy to Allwine.net. The picture had been taken on a family vacation to Hawaii, and it showed Amy wearing a teal shirt, with a broad smile on her tan and freckled face. About 25 minutes after he posted the image, Stephen logged in to his dogdaygod email account and sent Yura the link. “She is about 5'6", she looks about 200lbs,” he wrote. The best time to kill her, he continued, would be on an upcoming trip to Moline, Illinois. If the hit man could make her death look like an accident—by, say, ramming her Toyota Sienna minivan on the driver's side—he would throw in a few more bitcoin.

    Yura confirmed the details shortly afterward in awkward English. “He will wait her at the airport, tail her with the stolen car, and when he has the chance will cause a car accident to kill her.” If the car accident didn't work out, he added, “the hitman will shoot her deadly.” Later he reminded dogdaygod to concoct an alibi: “Please make sure you are sorrounded [sic] by people most of the days, and spend some money to shop things on malls or public places where they have video surveillance.”

    On a typical day, Stephen was not surrounded by people. He and Amy lived on 28 acres on a dead-end street. Their house was a simple one-story, double-wide trailer on a basement, but it had four bedrooms, along with a spacious living room and an open kitchen. Stephen had rigged up the roof with solar panels, which he boasted generated so much energy that he was able to feed power back into the grid. He spent much of his time in his basement office, handling glitches in call center technology. Working from home allowed him to hold down two jobs, one with the IT service company Optanix and the other with Cigna, the health insurer. Coworkers often went to him with particularly thorny problems.

    The Allwines' pastor preached about conquering carnal desires, and Stephen himself counseled couples in the congregation who were struggling with marital problems. When he was alone, though, his attention strayed. He ventured onto Naughtydates.com and LonelyMILFs.com. He found an escort on the classified site Backpage and twice drove to Iowa to have sex with her. Through his counseling work, he learned about Ashley Madison, the dating service that caters to married people. It was there he met Michelle Woodard.

    On their first date, Stephen accompanied Woodard to a doctor's appointment. Within a few weeks, she was joining him on work trips. Woodard appreciated Stephen's extraordinary calm. On one trip their connecting flight out of Philadelphia was canceled. Stephen had an 8 am meeting the next day in Hartford, Connecticut, and without fuss he rented a car and drove them the remaining 210 miles.

    A month before Stephen ordered the hit on his wife, he told Woodard that he was going to try to make things work with Amy. In truth, the affair seemed to intensify his desire for a different sort of life.

    Disciplined and computer-savvy, Stephen was in theory the perfect criminal for a dark-web crime. He covered his tracks by using anonymous remailers, which strip identifying information off messages, and Tor, which cloaks an IP address by randomly bouncing communications through a network of relays. And he concocted an elaborate backstory: dogdaygod was a rival dog trainer who wanted Amy dead because Amy had slept with her husband. In his dark-web persona, he transferred his own infidelity onto his wife.

    The United Church of God met in a local Methodist church.

    Alec Soth

    Stephen scheduled the murder for the weekend of March 19, when Amy was going to be in Moline for a dog-training competition. But at the end of the weekend, he wrote to Yura, complaining that he hadn't yet seen any news of Amy's death. Yura explained that the hit man hadn't found the right moment to strike: “He needs to be in a position where he can hit her car to the driver door, lateral collisin [sic] to make sure she dies.” The Besa Mafia administrator seemed to sense that it was important to dogdaygod that Amy be taken out while traveling. “We are not interested in the reason for why the people are killed,” he wrote. “But if she is your wife or some family member, we can do it in your city as well,” he said, adding that his client could leave town on the appointed day. He suggested that Amy could be killed at home and agreed that her house could be burned to the ground—for an additional 10 bitcoin, or $4,100.

    “Not my wife,” Stephen replied, “but I was thinking the same thing.” The next day he scraped together the money. When he transferred the bitcoin to Besa Mafia, however, his screen refreshed and he didn't recognize the 34-digit code that popped up. Panicking, he worried that the cryptocurrency he had labored so hard to acquire had now disappeared without a trace. He hastily copied the code and pasted it into a note on his iPhone, then emailed the code to Yura under the subject line “HELP!” Less than a minute later he deleted the note.

    Yura wrote back seven hours later, assuring him that the transaction had gone through, but days passed and nothing happened. Over the coming weeks, Stephen's messages to Yura alternated between terse disappointment and increasingly detailed instructions. “I know her husband has a big tractor, so I suspect that he has gas cans in the garage,” he wrote, adding: “I ask that you only get her and not the dad or kid.” Like a friendly, chatroom-ready Satan, Yura responded promptly with messages reinforcing his client's basest instincts. “Yes she is really a bitch and she deserve to die,” he wrote. Ninety minutes later he added, “Please notice 80% of our hitman are gang members who do drug dealing, beatings, occasional murder.” For an additional fee, he said, dogdaygod could arrange for a more practiced killer—an ex-military Chechen sniper—to handle the job.

    Stephen had spent at least $12,000 on the hit man idea. Instead of giving up or reexamining the sin he was contemplating, he appeared to become more determined. He logged onto Dream Market, a dark-web marketplace best known for selling drugs, where he could explore other methods of killing. Common sense would suggest varying usernames, but he once again appeared as dogdaygod, as if he had become the character he'd created. He would make back his loss; the payout on Amy's life insurance policy was $700,000.

    In April 2016, about two months after Stephen first ordered the hit on his wife, Besa Mafia was hacked and Yura's messages with clients—including dogdaygod—were dumped in an online pastebin. The data dump revealed that users with names like Killerman and kkkcolsia had paid tens of thousands of dollars in bitcoin to have people killed in Australia, Canada, and Turkey, as well as the United States. The hit orders soon reached the FBI, which directed local field offices across the country to make contact with the intended victims named in the Besa Mafia data dump. FBI special agent Asher Silkey, who worked in the bureau's Minneapolis field office, learned that someone going by the name dogdaygod wanted Amy Allwine killed. He was tasked with warning her of the threat on her life.

    On a cloudy Tuesday afternoon just after Memorial Day, Silkey enlisted the help of Terry Raymond, an officer with the local police force, and they drove to the Allwines' house. Cottage Grove is a sleepy exurb, but, like police departments around the country, the local cops had been called on to address online threats with increasing frequency. Raymond, a reserved man with angular features accented by a trim beard, had been on the force for 13 years and was the department's designated computer forensics specialist.

    When Silkey and Raymond arrived, Stephen Allwine invited them inside. He told the two law enforcement officers that Amy was out, and they stood around in silence while he called her cell. Stephen struck Raymond as socially awkward, but he didn't think much of it. He'd dealt with all sorts in his work.

    The officers drove back to the station, and Amy showed up soon after. They met her in the lobby, which featured an oil painting of the department's canine, Blitz, and led her to a sparsely furnished interview room. Because the FBI was handling the investigation, Raymond mostly listened as Silkey explained that someone who knew Amy's travel schedule and her daily routine wanted her dead. Amy was stunned. She was further confused when Silkey mentioned the allegation about Amy sleeping with a dog trainer's husband. She couldn't think of anyone who considered her an enemy. “If you have any activity that you find suspicious, give us a call,” Raymond said as she left.

    A few weeks later, the Allwines installed a motion-activated video surveillance system at their home, setting up cameras at different entrances. Stephen, meanwhile, purchased a gun—a Springfield XDS 9 mm. He and Amy decided to keep it under her side of the bed. They went on a date to the shooting range.

    The Cottage Grove force, from left: captains Gwen Martin and Randy McAlister and detectives Terry Raymond and Jared Landkamer.

    Alec Soth

    On July 31, Amy called Silkey, distraught: Over the past week, she had received two anonymous email threats. Silkey drove over to the Allwines' house, where Stephen printed out the emails and listened while Amy explained to the agents what had happened.

    The first message came from an anonymous remailer registered in Austria. It read, in part:

    Amy, I still blame you for my life falling apart … I see that you have put up a security system now, and I have been informed by people on the Internet that the police were snooping around my earlier emails. I have been assured that the emails are untraceable and they will not find me, but I cannot attack you directly with them watching.

    Here is what is going to happen. Since I cannot get to you, I will come after everything else that you love.

    The email went on to list location information for Amy's family members, based on what the sender said was found on Radaris.com, a site that makes contact information and background reports available to subscribers. The writer also dropped details that only someone closely following Amy could know—the location of the gas meter on the Allwines' house, the fact that they had moved their RV to a new parking spot, the color of the shirt that their son had worn two days earlier. “Here is how you can save your family,” the email continued. “Commit suicide.” The writer offered various methods by which she could accomplish that end.

    A week later the second anonymous message arrived, chiding her for not taking her own life: “Are you so selfish that you will put your families [sic] lives at risk?”

    Amy handed over her computer, hopeful that something on it might help the agents track down her potential killer. Stephen gave the agents a laptop and his Samsung Galaxy cell phone. The FBI imaged the devices, creating a copy of their applications, processes, and files, and returned them a day or two later.

    Amy gave Silkey the names of people who taught at her arena, animal owners she had worked with, her best friend. The FBI agent interviewed four of them and pulled credit reports for several contacts. Few people stood to profit from Amy's death, yet dogdaygod had paid out thousands of dollars to kill her, suggesting a personal motive. What's more, her persecutor had taken care to instruct Yura not to kill Amy's husband. Investigating a spouse would seem a logical measure. Silkey interviewed Stephen, in addition to imaging the devices, but it is not clear if he did more. The FBI has declined interview requests, and the Cottage Grove police did not have much insight into the bureau's work. Beyond bringing Raymond into the initial interview and sending him a copy of the threatening emails, the agency did not involve the local police.

    Meanwhile, Amy tried to cope with the vicious threats. She enrolled in the police department's Citizen Academy, explaining on her application that she wanted to “learn about the police department, what it does, and how it works.” Sergeant Gwen Martin, who ran the course, didn't know about the threats on Amy's life, nor did Amy tell any of the other participants in the course about her worries as they practiced shooting targets and retrieving fingerprints from a Coke can. Amy asked to be assigned to the K-9 officer for her ride-along, and she was so enthusiastic about exchanging tips on dog obedience and scent training that the officer let her tag along for an extra hour or two. When the program was over, she celebrated with the rest of the group at a small graduation party.

    But Amy still felt powerless. The occasional migraines she suffered became more frequent, and she had trouble remembering things. She put on a brave face when she taught class, but inwardly she worried that her aggressor might be among her dog-training crowd.

    One summer night she sat outside with her sister, looking up at the stars and wondering who was responsible for the pall that had been cast over her life. Years earlier, when her sister started college, Amy sent her a note every week so she wouldn't get homesick. Now her sister returned the favor. In each note she quoted scripture.

    One Saturday afternoon in November, Stephen and Amy set off for church with their son. The road cut through the floodplain east of the Mississippi River, passing yellowing farm fields, yards filled with auto parts, and wooded ravines barren of leaves. The United Church of God rented space from a local Methodist congregation in a redbrick building. There was something appropriately austere about the setting, as if through architectural restraint alone the devil could be kept at bay.

    Inside the chapel, the family sat in a pew, joining men in suit jackets, women with modest hemlines, and children with freshly combed hair. Daylight flooded in through a large skylight as pastor Brian Shaw recited the New Testament's admonition against “having eyes full of adultery and that cannot cease from sin.” He spoke of Job, who trained his eyes not to look with lust at women. The cost of not following Job's lead was dear: “When we do not control our sinful natures, they control us.”

    On Sunday, Stephen woke up just before 6 am, as usual, and descended to his basement office, where he logged in to the Optanix system to start work. At noon he wandered upstairs to have lunch with Amy and their son. Amy, an avid baker, had part of a pumpkin left over from a dessert she'd made a couple days earlier, and she put it in the slow cooker on the kitchen island to roast. Soon after, she started to get woozy.

    Amy's father showed up to work on a dog door he was installing in the garage. Stephen told him that Amy wasn't feeling well and was in the bedroom resting. Her father left without seeing her. Five minutes after he started driving home, Stephen called to ask his father-in-law to turn around and pick up his grandson, explaining that he wanted to take Amy to a clinic.

    As dusk fell, Stephen drove to get gas, then retrieved the boy from his in-laws' house and took him to Culver's, a family-style restaurant chain. It was their Sunday night routine—dinner at Culver's while Amy led dog-training courses—and they sat in the brightly lit space eating chicken tenders and grilled cheese.

    When they returned home, the boy climbed out of the minivan and ran into the house, toward his parents' bedroom. Amy's body lay in an unnatural position, blood pooled around her head. The Springfield XDS 9 mm was at her side.

    Stephen called 911.

    “I think my wife shot herself,” he said. “There's blood all over.”

    Cottage Grove City Hall, where the police department is housed.

    Alec Soth

    Sergeant Gwen Martin arrived at the house a few minutes after the 911 call. When she saw Amy's body on the floor, she remembered training her in the Citizen Academy and burst into tears. Another sergeant took over, and Martin retreated to her squad car. Regaining her composure, she turned to the laptop mounted to the dash and ran a search on police calls to the residence. She was astonished to find the report that Terry Raymond had filed about the dark-web threats to Amy's life. Martin grabbed her phone and dialed Detective Sergeant Randy McAlister, who directed Cottage Grove investigations.

    A baby-faced man of 47 who rode a Harley, McAlister often joined in the frequent joking around the department. He drank coffee from a mug that read “Due to the confidentiality of my job I don't know what I'm doing.” But his chipper demeanor concealed his earnestness. A decade earlier, McAlister had responded to a murder in a nearby town; a couple had been killed in their home by the woman's former boyfriend, as her children cowered nearby. The woman had previously told police that her jealous ex had contacted her in violation of a court order. Frustrated that the system had failed that woman, McAlister started a program aimed at protecting potential victims from stalking and targeted violence. When Raymond mentioned the dark-web threats Amy had received, he suggested they be compared to a database of threats kept by the FBI's Behavioral Analysis Unit; it might help them come up with a profile of a potential perpetrator. But he had no authority in the case.

    Now he raced to the Allwines' home. As he entered through the garage, the aroma of roasting pumpkin, still in the slow cooker, hit his nose. This struck him as odd; people don't typically start cooking right before killing themselves. Other things about the scene were off: There were blood smears on both sides of the bedroom door. And while the mud room floor was covered with dog hair, the floor in the adjacent hall was clean.

    As McAlister waited for the medical examiner and state criminal investigators to arrive, an officer drove Stephen and his son to the station. As a colleague sat with the boy in the station's break room, Raymond escorted Stephen to the same interview room where he and Silkey had met with Amy five months earlier. Raymond pulled on a pair of latex gloves and swabbed Stephen's cheek for DNA. “Are you going to get that from my in-laws?” Stephen asked.

    “No, it should just be you and your son,” Raymond said. He asked Stephen to run through what he had done that day.

    Stephen was cooperative, though Raymond thought his demeanor was wooden for a man who had just lost his wife. He reminded the detective that Amy had an FBI file; he said that her computer had been acting strangely. “Being in the IT industry, it's frustrating because I know how things are supposed to work in a legitimate world,” he said, adding: “I don't know anything about hacking or anything like that.”

    For the next three days, investigators combed the crime scene. State technicians sprayed a chemical called luminol on the floors, then flicked off the lights. Where the luminol hit blood or cleaning solution, it glowed bright blue. The glow showed that the hallway had been cleaned; it also lit up some footprints leading back and forth from the bedroom to the laundry room.

    The Cottage Grove police executed a search warrant on the house. McAlister stationed himself at the dining room table, logging evidence. Raymond descended to Stephen's basement office. Stepping through the door, he saw every surface covered with junk: file folders, tangles of cords, external drives, SD cards, a voice recorder, and a Fitbit. There were hard drives of a type that hadn't been used in nearly a decade. On Stephen's desk were three monitors and a MacBook Pro laptop—not the machine he'd given the FBI.

    The officers brought their haul upstairs, then one by one handed the items to McAlister to log.

    “Holy crap,” he thought as the equipment amassed. Then, “Jeez, no más.” But the devices and drives kept coming. Sixty-six in all.

    Because the crime involved a local death, the Cottage Grove police took control of the investigation. Two and a half weeks after Amy died, the FBI sent over her file. When the police opened the documents, McAlister and Raymond saw—for the first time—the full Besa Mafia messages. That was when they learned that the person who wanted Amy dead went by the name dogdaygod.

    By this point, Stephen was a suspect, but there was no direct evidence linking him to the murder. That his DNA was on everything was hardly remarkable; it was his house. Video from the Allwines' security system revealed nothing abnormal, though the records were incomplete. Stephen explained that he and Amy had neglected to activate the camera over the sliding glass door because their dogs were constantly going in and out. McAlister hoped the answers might be inside the devices that Raymond had lugged out of the Allwines' basement.

    From the moment the Besa Mafia files appeared in the pastebin, bloggers had concluded that the site was a scam. One after another, Yura's clients complained that the hits they'd ordered hadn't been carried out. But McAlister didn't want to take anything for granted. He and Detective Jared Landkamer identified 10 other targets of Besa Mafia orders in the United States and contacted the police departments where they lived. They might get leads in their case or perhaps save other lives.

    McAlister divvied up the electronic work. He sent the computers to a digital forensics specialist at a neighboring police department. Landkamer subpoenaed the Allwines' emails—and then spent many long days reading them. Raymond started by extracting data from Stephen's phones. In a windowless room lined with department-issue monitors, he deployed software that sorted the data—apps here, call logs there—and reconstructed timelines for the devices. On the phone Stephen had given to the FBI to image, Raymond found the apps Orfox and Orbot, which are used to access Tor. He also found text messages containing confirmation codes from LocalBitcoins. The FBI seemed to have either missed them or paid them little mind.

    When he scanned Amy's phone, he could see that, on the day of her death, she seemed to be growing progressively more confused. At 1:48 pm she visited the Wikipedia page for vertigo. At 1:49, she typed “DUY” into Bing. Then, one minute later, “EYE.” Then “DIY VWHH.” It was as if she were desperate to understand why the room was spinning but couldn't execute a simple search.

    In an interview with a state investigator, Stephen had confessed to his affair with Woodard. Raymond found a contact for “Michelle” in Stephen's phone, and when investigators questioned Woodard, she told them about the birthday lunch, when Stephen messaged that he had locked his keys in the car while buying bitcoin. Stephen's call history confirmed that he had phoned for roadside assistance that day from a Wendy's in Minneapolis. The detectives used the text message confirmation codes on Stephen's phone to find his LocalBitcoins account. That led them to his correspondence with a seller about exchanging $6,000 in cash.

    In Stephen's devices, Landkamer found secondary email addresses that led to usernames he used to access Backpage and LonelyMILFS.com. That wasn't incriminating on its own, but it did suggest a motive.

    While Stephen had cloaked most of his criminal activity, he did not purge his more innocuous internet search history. On February 16, a few minutes before dogdaygod first proposed killing Amy in Moline, Stephen had Googled “moline il” on his MacBook Pro. One day later, he looked up their life insurance policy. In July, shortly before Amy received the first threatening email that included addresses gleaned from Radaris, he visited the Radaris pages for Amy's family members.

    Murder was rare in Cottage Grove, and the detectives, confronted with circumstantial evidence and the slipperiness of the dark web, obsessed over the case. Lying in bed one night after reading Amy's FBI file, Landkamer searched “dogdaygod” in Google. When he saw the results, he called out to his wife. The search engine had indexed some posts on Dream Market, the dark-web marketplace where drugs were sold.

    Landkamer immediately texted McAlister what he'd found. McAlister fired up Tor on his personal laptop and pulled up the full threads on Dream Market. In one thread, dogdaygod asked if anyone sold scopolamine, a powerful prescription drug. McAlister had worked as a paramedic, so he knew that scopolamine was prescribed for motion sickness, but it could also make people pliable and amnesiac, earning it the nickname Devil's Breath. As he scrolled down, he came to a comment from a user who assumed that dogdaygod wanted to use scopolamine recreationally. “There is a seller,” the person wrote, “but avoid that shit mate. It's dangerous as fuck and you WILL kill someone.”

    Later, Amy's gastric contents tested positive for scopolamine. But it was a quirk in Apple backups that provided the strongest piece of evidence. The digital forensics specialist from a neighboring department found, archived in Stephen's MacBook Pro, that a note with a bitcoin wallet address had appeared on Stephen's iPhone back in March 2016. This was 23 seconds before dogdaygod frantically wrote Yura with the same 34-digit bitcoin wallet code. Forty seconds after dogdaygod messaged Yura, the note was deleted from Stephen's phone. But deleted files don't disappear until they're overwritten by other files. Several months later, when Stephen backed up his phone to iTunes, the crucial history was preserved on his laptop.

    McAlister was elated. The detectives had linked Stephen's offline persona, a church elder concerned with the propriety of dance moves, with his online ones—the philanderer and the aggrieved would-be murderer. The enticing anonymity of the dark web that nurtured Stephen's crime had given him a sense of omnipotence. He failed to appreciate that this cloak of power didn't follow him to the clear web and to the real world.

    Stephen Allwine is now incarcerated in the Minnesota Correctional Facility, in Oak Park Heights, Minnesota.

    Alec Soth

    Stephen Allwine's trial lasted for eight days. County prosecutors paraded a string of colorful witnesses to the stand: the manager of the pawn shop where Stephen sold his silver, the Backpage escort in Iowa, and Woodard. McAlister held up the murder weapon in court, and in one awkward moment that would become the subject of endless jokes at the Cottage Grove police station, Jared Landkamer defined “MILF” for the court.

    Prosecutors Fred Fink and Jamie Kreuser used the testimony to outline a theory: Stephen had poisoned Amy with a large dose of scopolamine, either to kill or incapacitate her. Either way, while she grew dazed and light-headed, she didn't die. So Stephen shot her with their gun in the hallway. Then he moved the body into the bedroom and cleaned up the blood. When he left to get gas and take his son to Culver's, he was careful to save the receipts.

    The jury deliberated for six hours before finding Stephen guilty. On February 2, he appeared in a packed courtroom for sentencing. One by one, friends and family told the judge how much Amy had meant to them. (Amy's family declined to be interviewed.) Then Stephen rose to plead his case.

    Breathlessly, he tried to refute the technical testimony about backup files and bitcoin wallets. Then he shifted to his spiritual gifts. In jail, where he had been held during the trial, he was ministering to drug addicts and child molesters. He had converted at least three nonbelievers, he said.

    “Mr. Allwine,” the judge said when he had finished, “my perceptions aren't going to alter the sentence in this case. But my perception is that you're an incredible actor. That you can turn tears on and off. That you are a hypocrite and that you are cold.” He sentenced him to life without parole. (The case is headed to appeals court.) From a room adjacent to the courtroom, McAlister watched through a window with Raymond and Landkamer, taking satisfaction as the judge admonished the criminal. But the moment was not without uneasiness. McAlister saw why Stephen might not have triggered alarm bells during the FBI's dark-web investigation: Stephen and Amy appeared to have a happy relationship, with no history of violence or substance abuse. He knew that hindsight bias could color investigators' conclusions, but he also had the feeling that Amy's death might have been prevented. Threat assessment experts use a four-part checklist to determine whether an anonymous harasser is an intimate partner. Amy's harasser met all four conditions in that test: The person closely tracked her whereabouts, seemed to live nearby, knew her habits and future plans, and spoke of her with contempt or disgust.

    In the months following the trial, McAlister was promoted to captain. From time to time, he offers advice to police departments dealing with dark-web crime. There were no other deaths tied back to Besa Mafia customers, but Yura reportedly started other hit-man-for-hire scam sites—Crime Bay, Sicilian Hitmen, Cosa Nostra. It was almost like Yura was the devil watching from a distance, smirking as the seeds he planted germinated and grew into full-blown evil.

    All photographs by Alec Soth/Magnum Photos

    Mara Hvistendahl (@marahvistendahl) is writing a book for Riverhead on a trade-secrets theft case. She wrote about China's social credit system in issue 26.01.

    This article appears in the May issue. Subscribe now.

    Let us know what you think about this article. Submit a letter to the editor at mail@wired.com.

    Read more: https://www.wired.com/story/dark-web-bitcoin-murder-cottage-grove/

    Related Articles

    Markets Are Eating The World

    For the last hundred years, individuals have worked for firms, and, by historical standards, large ones.

    That many of us live in suburbs and drive our cars into the city to go to work at a large office building is so normal that it seems like it has always been this way. Of course, it hasn’t. In 1870, almost 50 percent of the U.S. population was employed in agriculture.[1] As of 2008, less than 2 percent of the population is directly employed in agriculture, but many people worked for these relatively new things called “corporations.”[2]

    Many internet pioneers in the 90’s believed that the internet would start to break up corporations by letting people communicate and organize over a vast, open network. This reality has sort-of played out: the “gig economy” and rise in freelancing are persistent, if not explosive, trends. With the re-emergence of blockchain technology, talk of “the death of the firm” has returned. Is there reason to think this time will be different?

    To understand why this time might (or might not) be different, let us first take a brief look back into Coasean economics and mechanical clocks.

    In his 1937 paper, “The Nature of the Firm,” economist R.H. Coase asked “if markets were as efficient as economists believed at the time, why do firms exist at all? Why don’t entrepreneurs just go out and hire contractors for every task they need to get done?”[3]

    If an entrepreneur hires employees, she has to pay them whether they are working or not. Contractors only get paid for the work they actually do. While the firm itself interacts with the market, buying supplies from suppliers and selling products or services to customers, the employees inside of it are insulated. Each employee does not renegotiate their compensation every time they are asked to do something new. But, why not?

    Coase’s answer was transaction costs. Contracting out individual tasks can be more expensive than just keeping someone on the payroll because each task involves transaction costs.

    Imagine if instead of answering every email yourself, you hired a contractor that was better than you at dealing with the particular issue in that email. However, it costs you something to find them. Once you found them you would have to bargain and agree on a price for their services then get them to sign a contract and potentially take them to court if they didn’t answer the email as stipulated in the contract.

    Duke economist Mike Munger calls these three types of transaction costs triangulation, how hard it is to find and measure the quality of a service; transfer, how hard it is to bargain and agree on a contract for the good or service; and trust, whether the counterparty is trustworthy or you have recourse if they aren’t.

    You might as well just answer the email yourself or, as some executives do, hire a full-time executive assistant. Even if the executive assistant isn’t busy all the time, it’s still better than hiring someone one off for every email or even every day.

    Coase’s thesis was that in the presence of these transaction costs, firms will grow larger as long as they can benefit from doing tasks in-house rather than incurring the transaction costs of having to go out and search, bargain and enforce a contract in the market. They will expand or shrink until the cost of making it in the firm equals the cost of buying it on the market.

    The lower the transaction costs are, the more efficient markets will be, and the smaller firms will be.

    In a world where markets were extremely efficient, it would be very easy to find and measure things (low triangulation costs), it would be very easy to bargain and pay (low transfer costs), and it would be easy to trust the counterparty to fulfill the contract (low trust costs).

    In that world, the optimal size of the firm is one person (or a very few people). There’s no reason to have a firm because business owners can just buy anything they need on a one-off basis from the market.[4] Most people wouldn’t have full-time jobs; they would do contract work.

    Consumers would need to own very few things. If you needed a fruit dehydrator to prepare for a camping trip twice a year, you could rent one quickly and cheaply. If you wanted to take your family to the beach twice a year, you could easily rent a place just for the days you were there.

    On the other hand, in a world that was extremely inefficient, it would be hard to find and measure things (high triangulation costs), it would be difficult to bargain and pay (high transfer costs) and it would be difficult to trust the counterparty to fulfill the contract (high trust costs).

    In that world, firms would tend to be large. It would be inefficient to buy things from the market and so entrepreneurs would tend to accumulate large payrolls. Most people would work full-time jobs for large firms. If you wanted to take your family to the beach twice a year, you would need to own the beach house because it would be too inefficient to rent, the reality before online marketplaces like AirBnB showed up.

    Consumers would need to own nearly everything they might conceivably need. Even if they only used their fruit dehydrator twice a year, they’d need to own it because the transaction costs involved in renting it would be too high.

    If the structure of the economy is based on transaction costs, then what determines them?

    Technological Eras and Transaction Costs

    The primary determinant of transaction costs is technology.

    The development of the wheel and domestication of horses and oxes decreased transfer costs by making it possible to move more goods further. Farmers who could bring their crops to market using an ox cart rather than carrying it by hand could charge less and still make the same profit.

    The development of the modern legal system reduced the transaction cost of trust. It was possible to trust that your counterparty would fulfill their contract because they knew you had recourse if they didn’t.

    The list goes on: standardized weights and  measures, the sail, the compass, the printing press, the limited liability corporation, canals, phones, warranties, container ships and, more recently, smartphones and the internet.

    It’s hard to appreciate how impactful many of these technologies has been, because most of them had become so common by the time most of us were born that we take them for granted.

    As the author Douglas Adams said, “Anything that is in the world when you’re born is normal and ordinary and is just a natural part of the way the world works. Anything that’s invented between when you’re fifteen and thirty-five is new and exciting and revolutionary and you can probably get a career in it. Anything invented after you’re thirty-five is against the natural order of things.”

    To see how technology affects transaction costs, and how that affects the way our society is organized, let’s consider something which we all think of as “normal and ordinary,”  but which has had a huge impact on our lives: the mechanical clock.

    The Unreasonable Effectiveness of the Mechanical Clock

    In 1314, The city of Caen installed a mechanical clock with the following inscription: “I give the hours voice to make the common folk rejoice.” “Rejoice” is a pretty strong reaction to a clock, but it wasn’t overstated, everyone in Caen was pretty jazzed about the mechanical clock. Why?

    A key element of why we have jobs today as opposed to working as slaves or serfs bonded to the land as was common in the Feudal system is a direct result of the clock.

    Time was important before the invention of the clock but was very hard to measure. Rome was full of sundials, and medieval Europe’s bell towers where, time was tolled, were the tallest structures in town.[5]

    This was not cheap. In the larger and more important belfries, two bell-ringers lived full time, each serving as a check on the other. The bells themselves were usually financed by local guilds that relied on the time kept to tell their workers when they had to start working and when they could go home.

    This system was problematic for a few reasons.

    For one, it was expensive. Imagine if you had to pool funds together with your neighbors to hire two guys to sit in the tower down the street full time and ring the bell to wake you up in the morning.

    For another, the bell could only signal a few events per day. If you wanted to organize a lunch meeting with a friend, you couldn’t ask the belltower to toll just for you. Medieval bell towers had not yet developed snooze functionality.

    Finally, sundials suffered from accuracy problems. Something as common as clouds could make it difficult to tell precisely when dawn, dusk, and midday occurred.

    In the 14th and 15th centuries, the expensive bell towers of Europe’s main cities got a snazzy upgrade that dramatically reduced transaction costs: the mechanical clock.

    The key technological breakthrough that allowed the development was the escapement.

    The escapement transfers energy to the clock’s pendulum to replace the energy lost to friction and keep it on time. Each swing of the pendulum releases a tooth of the escapement’s wheel gear, allowing the clock’s gear train to advance or “escape” by a set amount. This moves the clock’s hands forward at a steady rate.[6]

    The accuracy of early mechanical clocks, plus or minus 10-15 minutes per day, was not notably better than late water clocks and less accurate than the sandglass, yet mechanical clocks became widespread. Why?

    1. Its automatic striking feature meant the clock could be struck every hour at lower cost, making it easier to schedule events than only striking at dawn, dusk and noon.
    2. It was more provably fair than the alternatives, which gave all parties greater confidence that the time being struck was accurate. (Workers were often suspicious that employers could bribe or coerce the bell-ringers to extend the workday, which was harder to do with a mechanical clock.)

    Mechanical clocks broadcast by bell towers provided a fair (lower trust costs) and fungible [7] (lower transfer costs) measure of time. Each hour rung on the bell tower could be trusted to be the same length as another hour.

    Most workers in the modern economy earn money based on a time-rate, whether the time period is an hour, a day, a week or a month. This is possible only because we have a measure of time which both employer and employee agree upon. If you hire someone to pressure-wash your garage for an hour, you may argue with them over the quality of the work, but you can both easily agree whether they spent an hour in the garage.

    Prior to the advent of the mechanical clock, slavery and serfdom were the primary economic relationships, in part because the transaction cost of measuring time beyond just sunup and sundown was so high, workers were chained to their masters or lords.[8]

    The employer is then able to use promotions, raises, and firing to incentivize employees to produce quality services during the time they are being paid for.[9]

    In a system based on time-rate wages rather than slavery or serfdom, workers have a choice. If the talented blacksmith can get a higher time-rate wage from a competitor, she’s able to go work for them because there is an objective, fungible measure of time she’s able to trade.

    As history has shown, this was a major productivity and quality-of-life improvement for both parties.[10]

    It gradually became clear that mechanical time opened up entirely new categories of economic organization and productivity that had hitherto been not just impossible, but unimaginable.

    We could look at almost any technology listed abovestandardized weights and measures, the sail, the compass, the printing press, etc.and do a similar analysis of how it affected transaction costs and eventually how it affected society as a result.

    The primary effect is an increase in what we will call coordination scalability.

    Coordination Scalability

    “It is a profoundly erroneous truism, repeated by all copy-books and by eminent people when they are making speeches, that we should cultivate the habit of thinking what we are doing. The precise opposite is the case. Civilization advances by extending the number of important operations which we can perform without thinking about them.”   Alfred North Whitehead

    About 70,000 years ago, there were between six and ten species of the genus homo. Now, of course, there is just one: Homo sapiens. Why did Homo sapiens prevail over the other species, like Homo neanderthalensis?

    Homo sapiens prevailed because of their ability to coordinate. Coordination was made possible by increased neocortical size, which led to an ability to work together in large groups, not just as single individuals. Instead of single individuals hunting, groups could hunt and bring down larger prey more safely and efficiently.[11]

    The brain of Homo sapiens has proven able to invent other, external structures which further increased coordination scalability by expanding the network of other people we could rely on.

    Maybe the most important of these was language, but we have evolved many others since, including the mechanical clock.

    The increased brain size has driven our species through four coordination revolutions: Neolithic, Industrial, Computing, Blockchain.

    Neolithic Era: The Emergence of Division of Labor

    The first economic revolution was a shift from humans as hunter-gatherers to homo sapiens as farmers.

    Coordination scalability among hunter-gatherers was limited to the size of the band, which tended to range from 15 to 150 individuals.[12] The abandonment of a nomadic way of life and move to agriculture changed this by allowing specialization and the formation of cities.

    Agriculture meant that people could, for the first time, accumulate wealth. Farmers could save excess crops to eat later or trade them for farming equipment, baskets or decorations. The problem was that this wealth was suddenly worth stealing and so farmers needed to defend their wealth.

    Neolithic societies typically consisted of groups of farmers protected by what Mancur Olson called “stationary bandits,” basically warlords.[13] This allowed the emergence of much greater specialization. Farmers accumulated wealth and paid some to the warlords for protection, but even then there was still some left over, making it possible for individuals to specialize.

    A city of 10,000 people requires, but also makes possible, specialists.

    The limits of coordination scalability increased from 150 to thousands or, in some cases, tens of thousands. This was not necessarily a boon to human happiness. Anthropologist Jared Diamond called the move to agriculture “the worst mistake in the history of the human race.”[14] The quality of life for individuals declined: lifespans shortened, nutrition was worse leading to smaller stature, and disease was more prevalent.

    But this shift was irresistible because specialization created so much more wealth and power that groups which adopted this shift came to dominate those that didn’t. The economies of scale in military specialization, in particular, were overwhelming. Hunt-gatherers couldn’t compete.

    In the Neolithic era, the State was the limit of coordination scalability.

    Industrial Era: Division of Labor Is Eating the World

    Alongside the city-state, a new technology started to emerge that would further increase the limits of coordination scalability: money. To illustrate, let us take the European case, from ancient Greece to modernity, though the path in other parts of the world was broadly similar. Around 630 B.C., the Lydian kings recognized the need for small, easily transported coins worth no more than a few days’ labor. They made these ingots in a standard sizeabout the size of a thumbnail—and weight, and stamped an emblem of a lion’s head on them.

    This eliminated one of the most time-consuming (and highest transaction cost) steps in commerce: weighing gold and silver ingots each time a transaction was made. Merchants could easily count the number of coins without worrying about cheating.

    Prior to the invention of coins, trade had been limited to big commercial transactions, like buying a herd of cattle. With the reduced transfer cost facilitated by coins, Lydians began trading in the daily necessities of lifegrain, olive oil, beer, wine, and wood.[15]

    The variety and abundance of goods which could suddenly be traded led to another innovation: the retail market.

    Previously, buyers had to go to the home of sellers of whatever they needed. If you needed olive oil, you had to walk over to the olive oil lady’s house to get it. With the amount of trade that began happening after coinage, a central market emerged. Small stalls lined the market where each merchant specialized in (and so could produce more efficiently) a particular goodmeat, grain, jewelry, bread, cloth, etc. Instead of having to go the olive oil lady’s house, you could go to her stall and pick up bread from the baker while you were there.

    From this retail market in Lydia sprang the Greek agora, Medieval market squares in Europe and, the suburban shopping mall and, eventually, the “online shopping malls” Amazon and Google. Though markets were around as early as 7th century BCE Lydia, they really hit their stride in The Industrial Revolution in the 18th century.[16]

    Adam Smith was the first to describe in detail the effect of this marketization of the world. Markets made it possible to promote the division of labor across political units, not just within them. Instead of each city or country manufacturing all the goods they needed, different political entities could further divide labor. Coordination scalability started to stretch across political borders.

    Coming back to Coase, firms will expand or shrink until “making” equals the cost of “buying.” Under this Industrial era, transaction costs made administrative and managerial coordination (making) more efficient than market coordination (buying) for most industries, which led to the rise of large firms.

    The major efficiency gain of Industrial companies over their more “artisanal” forebearers was that using the techniques of mass production, they could produce products of a higher quality at a lower price. This was possible only if they were able to enforce standards throughout the supply chain. The triangulation transaction cost can be broken down into search and measurement: a company needed to find the vendor and to be able to measure the quality of the good or service.

    In the early Industrial era, the supply chain was extremely fragmented. By bringing all the pieces into the firm, a large vertically integrated company could be more efficient.[17]

    As an example, In the 1860s and 1870s, the Carnegie Corporation purchased mines to ensure it had reliable access to the iron ore and coke it needed to make steel. The upstream suppliers were unreliable and non-standardized and Carnegie Corporation could lower the cost of production by simply owning the whole supply chain.

    This was the case in nearly every industry. By bringing many discrete entities under one roof and one system of coordination, greater economic efficiencies were gained and the multi-unit business corporation replaced the small, single-unit enterprise because administrative coordination enabled greater productivity through lower transaction costs per task than was possible before. Economies of scale flourished.

    This system of large firms connected by markets greatly increased coordination scalability. Large multinational firms could stretch across political boundaries and provide goods and services more efficiently.

    In Henry Ford’s world, the point where making equaled the cost of buying was pretty big. Ford built a giant plant at River Rouge just outside Detroit between 1917 and 1928 that took in iron ore and rubber at one end and sent cars out the other. At the factory’s peak, 100,000 people worked there. These economies of scale allowed Ford to dramatically drive down the cost of an automobile, making it possible for the middle class to own a car.[18]

    As with Carnegie, Ford learned that supplier networks take a while to emerge and grow into something reliable. In 1917, doing everything himself was the only way to get the scale he needed to be able to make an affordable car.

    One of the implications of this model was that industrial businesses required huge startup costs.

    The only chance any entrepreneur had to compete required starting out with similarly massive amounts of capital required to build a factory large and efficient enough to compete with Ford.

    For workers, this meant that someone in a specialized role, like an electric engineer or an underwriter, did not freelance or work for small businesses. Because the most efficient way to produce products was in large organizations, specialized workers could earn the most by working inside large organizations, be they Ford, AT&T or Chase Bank.

    At the peak of the Industrial era, there were two dominant institutions: firms and markets.

    Work inside the firm allowed for greater organization and specialization which, in the presence of high transaction costs was more economically efficient.

    Markets were more chaotic and less organized, but also more motivating. Henry Ford engaged with the market and made out just a touch better than any of his workers; there just wasn’t room for many Henry Fords.

    This started to dissolve in the second half of the 20th century. Ford no longer takes iron ore and rubber as the inputs to their factories, but has a vast network of upstream suppliers.[19] The design and manufacturing of car parts now happens over a long supply chain, which the car companies ultimately assemble and sell.

    One reason is that supplier networks became more standardized and reliable. Ford can now buy ball bearings and brake pads more efficiently than he can make them, so he does. Each company in the supply chain focuses on what they know best and competition forces them to constantly improve.

    By the 1880s, it cost Carnegie more to operate the coke ovens in-house than to buy it from an independent source, so he sold off the coke ovens and bought it from the open market. Reduced transaction costs in the form of more standardized and reliable production technology caused both Ford and Carnegie corporation to shrink as Coase’s theory would suggest.

    The second reason is that if you want to make a car using a network of cooperating companies, you have to be able to coordinate their efforts, and you can do that much better with telecommunication technology broadly and computers specifically. Computers reduce the transaction costs that Coase argued are the raison d’etre of corporations. That is a fundamental change.[20]

    The Computing Era: Software Is Eating the World

    Computers, and the software and networks built on top of them, had a new economic logic driven by lower transaction costs.

    Internet aggregators such as Amazon, Facebook, Google, Uber and Airbnb reduced the transaction costs for participants on their platforms. For the industries that these platforms affected, the line between “making” and “buying” shifted toward buying. The line between owning and renting shifted toward renting.

    Primarily, this was done through a reduction in triangulation costs (how hard it is to find and measure the quality of a service), and transfer costs (how hard it is to bargain and agree on a contract for the good or service).

    Triangulation costs came down for two reasons. One was the proliferation of smartphones, which made it possible for services like Uber and Airbnb to exist. The other was the increasing digitization of the economy. Digital goods are both easier to find (think Googling versus going to the library or opening the Yellow Pages) and easier to measure the quality of (I know exactly how many people read my website each day and how many seconds they are there, the local newspaper does not).

    The big improvement in transfer costs was the result of matchmaking: bringing together and facilitating the negotiation of mutually beneficial commercial or retail deals.  

    Take Yelp, the popular restaurant review app. Yelp allows small businesses like restaurants, coffee shops, and bars to advertise to an extremely targeted group: individuals close enough to come to the restaurant and that searched for some relevant term. A barbecue restaurant in Nashville can show ads only to people searching their zip code for terms like “bbq” and “barbecue.” This enables small businesses that couldn’t afford to do radio or television advertising to attract customers.

    The existence of online customer reviews gives consumers a more trusted way to evaluate the restaurant.

    All of the internet aggregators, including Amazon, Facebook, and Google, enabled new service providers by creating a market and standardizing the rules of that market to reduce transaction costs.[21]

    The “sharing economy” is more accurately called the “renting economy” from the perspective of consumers, and the “gig economy” from the perspective of producers. Most of the benefits are the result of new markets enabled by lower transaction costs, which allows consumers to rent rather than own, including “renting” some else’s time rather than employing them full time.

    It’s easier to become an Uber driver than a cab driver, and an Airbnb host than a hotel owner. It’s easier to get your product into Amazon than Walmart. It’s easier to advertise your small business on Yelp, Google or Facebook than on a billboard, radio or TV.

    Prior to the internet, the product designer was faced with the option of selling locally (which was often too small a market), trying to get into Walmart (which was impossible without significant funding and traction), or simply working for a company that already had distribution in Walmart.

    On the internet, they could start distributing nationally or internationally on day one. The “shelf space” of Amazon or Google’s search engine results page was a lot more accessible than the shelf space of Walmart.

    As a result, it became possible for people in certain highly specialized roles to work independently of firms entirely. Product designers and marketers could sell products through the internet and the platforms erected on top of it (mostly Amazon and Alibaba in the case of physical products) and have the potential to make as much or more as they could inside a corporation.

    This group is highly motivated because their pay is directly based on how many products they sell. The aggregators and the internet were able to reduce the transaction costs that had historically made it economically inefficient or impossible for small businesses and individual entrepreneurs to exist.

    The result was that in industries touched by the internet, we saw an industry structure of large aggregators and a long tail [22] of small business which were able to use the aggregators to reach previously unreachable, niche segments of the market. Though there aren’t many cities where a high-end cat furniture retail store makes economic sense, on Google or Amazon, it does.

    source: stratechery.com

    Before


    After (Platform-Enabled Markets)


    Firms


    Platform


    Long Tail



    Walmart and big box retailers
    Amazon Niche product designers and manufacturers

    Cab companies
    Uber Drivers with extra seats

    Hotel chains
    Airbnb Homeowners with extra rooms

    Traditional media outlets
    Google and Facebook Small offline and niche online businesses

    For these industries, coordination scalability was far greater and could be seen in the emergence of micro-multinational businesses. Businesses as small as a half dozen people could manufacture in China, distribute products in North America, and employ people from Europe and Asia. This sort of outsourcing and the economic efficiencies it created had previously been reserved for large corporations.

    As a result, consumers received cheaper, but also more personalized products from the ecosystem of aggregators and small businesses.

    However, the rental economy still represents a tiny fraction of the overall economy. At any given time, only a thin subset of industries are ready to be marketized. What’s been done so far is only a small fraction of what will be done in the next few decades.

    Yet, we can already start to imagine a world which Munger calls “Tomorrow 3.0.” You need a drill to hang some shelves in your new apartment. You open an app on your smartphone and tap “rent drill.” An autonomous car picks up a drill and delivers it outside your apartment in a keypad-protected pod and your phone vibrates “drill delivered.” Once you’re done, you put it back in the pod, which sends a message to another autonomous car nearby to come pick it up. The rental costs $5, much less than buying a commercial quality power drill. This is, of course, not limited to drillsit could have been a saw, fruit dehydrator, bread machine or deep fryer.

    You own almost nothing, but have access to almost everything.

    You, nor your neighbors, have a job, at least in the traditional sense. You pick up shifts or client work as needed and maybe manage a few small side businesses. After you finish drilling the shelves in, you might sit down at your computer and see what work requests are open and work for a few hours on designing a new graphic or finishing up the monthly financial statements for a client.

    This is a world in which triangulation and transfer costs have come down dramatically, resulting in more renting than buying from consumers and more gig work than full-time jobs for producers.

    This is a world we are on our way to already, and there aren’t any big, unexpected breakthroughs that need to happen first.

    But what about the transaction cost of trust?

    In the computer era, the areas that have been affected most are what could be called low-trust industries. If the sleeping mask you order off of Amazon isn’t as high-quality as you thought, that’s not a life or death problem.

    What about areas where trust is essential?

    Enter stage right: blockchains.

    The Blockchain Era: Blockchain Markets Are Eating the World

    One area where trust matters a lot is money. Most of the developed world doesn’t think about the possibility of fiat money [23] not being trustworthy because it hasn’t happened in our lifetimes. For those that have experienced it, including major currency devaluations, trusting that your money will be worth roughly the same tomorrow as it is today is a big deal.

    Citizens of countries like Argentina and particularly Venezuela have been quicker to adopt bitcoin as a savings vehicle because their economic history made the value of censorship resistance more obvious.

    Due to poor governance, the inflation rate in Venezuela averaged 32.42 percent from 1973 until 2017. Argentina was even worse; the inflation rate there averaged 200.80 percent between 1944 and 2017.

    The story of North America and Europe is different. In the second half of the 20th century, monetary policy has been stable.

    The Bretton Woods Agreement, struck in the aftermath of the Second World War, aggregated control of most of the globe’s monetary policy in the hands of the United States. The European powers acceded to this in part because the U.S. dollar was backed by gold, meaning that the U.S. government was subject to the laws of physics and geology of gold mining. They could not expand the money supply any faster than gold could be taken out of the ground.

    With the abandonment of the gold standard under Nixon in 1973, control over money and monetary policy has moved into a historically small group of central bankers and powerful political and financial leaders and is no longer restricted by gold.

    Fundamentally, the value of the U.S. dollar today is based on trust. There is no gold in a vault that backs the dollars in your pocket. Most fiat currencies today have value because the market trusts that the officials in charge of U.S. monetary policy will manage it responsibly.

    It is at this point that the debate around monetary policy devolves into one group that imagines this small group of elitist power brokers sitting in a dark room on large leather couches surrounded by expensive art and mahogany bookshelves filled with copies of The Fountainhead smoking cigars and plotting against humanity using obscure financial maneuvering.

    Another group, quite reasonably, points to the economic prosperity of the last half-century under this system and insists on the quackery of the former group.

    A better way to understand the tension between a monetary system based on gold versus one based on fiat money this has been offered by political science professor Bruce Bueno de Mesquita:  “Democracy is a better form of government than dictatorships, not because presidents are intrinsically better people than dictators, but simply because presidents have less agency and power than dictators.”

    Bueno de Mesquita calls this Selectorate Theory. The selectorate represents the number of people who have influence in a government, and thus the degree to which power is distributed. The selectorate of a dictatorship will tend to be very small: the dictator and a few cronies. The selectorate in democracy tends to be much larger, typically encompassing the Executive, Legislative, and Judicial branches and the voters which elect them.

    Historically, the size of the selectorate involves a tradeoff between the efficiency and the robustness of the governmental system. Let’s call this the “Selectorate Spectrum.”

    Dictatorships can be more efficient than democracies because they don’t have to get many people on board to make a decision. Democracies, by contrast, are more robust, but at the cost of efficiency.

    Conservatives and progressives alike bemoan how little their elected representatives get done but happily observe how little their opponents accomplish. A single individual with unilateral power can accomplish far more (good or bad) than a government of “checks and balances.” The long-run health of a government means balancing the tradeoff between robustness and efficiency. The number of stakeholders cannot be so large that nothing gets done or the country will never adapt nor too small that one or a small group of individuals can hijack the government for personal gain.

    This tension between centralized efficiency and decentralized robustness exists in many other areas. Firms try to balance the size of the selectorate to make it large enough so there is some accountability (e.g. a board and shareholder voting) but not so large as to make it impossible to compete in a marketby centralizing most decisions in the hands of a CEO.

    We can view both the current monetary system and the internet aggregators through the lens of the selectorate. In both areas, the trend over the past few decades is that the robustness of a large selectorate has been traded away for the efficiency of a small one.[24]

    A few individualsheads of central banks, leaders of state, corporate CEOs, and leaders of large financial entities like sovereign wealth funds and pensions fundscan move markets and politics globally with even whispers of significant change. This sort of centralizing in the name of efficiency can sometimes lead to long feedback loops with potentially dramatic consequences.

    Said another way, much of what appears efficient in the short term may not be efficient but hiding risk somewhere, creating the potential for a blow-up. A large selectorate tends to appear to be working less efficiently in the short term, but can be more robust in the long term, making it more efficient in the long term as well. It is a story of the Tortoise and the Hare: slow and steady may lose the first leg, but win the race.

    In the Beginning, There Was Bitcoin

    In October 2008, an anonymous individual or group using the pseudonym Satoshi Nakamoto sent an email to a cypherpunk mailing list, explaining a new system called bitcoin. The opening line of the conclusion summed up the paper:

    “We have proposed a system for electronic transactions without relying on trust”

    When the network went live a few months later in January 2009, Satoshi embedded the headline of a story running that day in The London Times:

    “The Times 03/Jan/2009 Chancellor on brink of second bailout for banks”

    Though we can’t know for sure what was going through Satoshi’s mind at the time, the most likely explanation based is that Satoshi was reacting against the decisions being made in response to the 2008 Global Financial Crisis by the small selectorate in charge of monetary policy.

    Instead of impactful decisions about the monetary system like a bailout being reliant upon a single individual, the chancellor, Satoshi envisioned bitcoin as a more robust monetary system, with a larger selectorate beyond the control of a single individual.

    But why create a new form of money? Throughout history, the most common way for individuals to show their objections to their nation’s monetary policy was by trading their currency for some commodity like gold, silver, or livestock that they believed would hold its value better than the government-issued currency.

    Gold, in particular, has been used as a form of money for nearly 6,000 years for one primary reason: the stock-to-flow ratio. Because of how gold is deposited in the Earth’s crust, it’s very difficult to mine. Despite all the technological changes in the last few hundred years, this has meant that the amount of new gold mined in a given year (the flow) has averaged between 1-2 percent of the total gold supply (stock) with very little variation year to year.

    As a result, the total gold supply has never increased by more than 1-2 percent per year. In comparison to Venezuela’s 32.4 percent inflation and Argentina’s 200.80 percent inflation, gold’s inflation is far lower and more predictable.

    Viewed through the lens of Selectorate Theory, we can say that gold or other commodity forms of money have a larger selectorate and are more robust than government-issued fiat currency. In the same way a larger group of stakeholders in a democracy constrains the actions of any one politician, the geological properties of gold constrained governments and their monetary policy.

    Whether or not these constraints were “good” or “bad” is still a matter of debate. The Keynesian school of economics, which has come to be the view of mainstream economics, emerged out of John Maynard Keynes’s reaction to the Great Depression, which he thought was greatly exacerbated by the commitment to the gold standard and that governments should manage monetary policy to soften the cyclical nature of markets.

    The Austrian and monetarist schools believe that human behavior is too idiosyncratic to model accurately with mathematics and that minimal government intervention is best. Attempts to intervene can be destabilizing and lead to inflation so a commitment to the gold standard is the lesser evil in the long run.

    Taken in good faith, these schools represent different beliefs about the ideal point on the Selectorate Spectrum. Keynesians believe that greater efficiency could be gained by giving government officials greater control over monetary policy without sacrificing much robustness. Austrians and monetarists argue the opposite, that any short-term efficiency gains actually create huge risks to the long-term health of the system.

    Viewed as a money, bitcoin has many gold-like properties, embodying something closer to the Austrian and monetarist view of ideal money. For one, we know exactly how many bitcoin will be created21 millionand the rate at which they will be created. Like gold, the ability to change this is outside of the control of a single or small group of individuals, giving it a predictable stock-to-flow ratio and making it extremely difficult to inflate.

    Similar to gold, the core bitcoin protocol also makes great trade-offs in terms of efficiency in the name of robustness.[25]

    However, bitcoin has two key properties of fiat money which gold lacksit is very easy to divide and transport. Someone in Singapore can send 1/100th of a bitcoin to someone in Canada in less than an hour. Sending 1/100th of a gold bar would be a bit trickier.

    In his 1998 book, Cryptonomicon, science fiction author Neal Stephenson imagined a bitcoin-like money built by the grandchild of Holocaust survivors who wanted to create a way for individuals to escape totalitarian regimes without giving up all their wealth. It was difficult, if not impossible, for Jews to carry gold bars out of Germany, but what if all they had to do was remember a 12-word password phrase? How might history have been different?

    Seen in this way, bitcoin offers a potentially better trade-off between robustness and efficiency. Its programmatically defined supply schedule means the inflation rate will be lower than gold (making it more robust) while it’s digital nature makes it as divisible and transportable as any fiat currency (making it more efficient).

    Using a nifty combination of economic incentives for mining (proof-of-work system) and cryptography (including blockchain), bitcoin allowed individuals to engage in a network that was both open (like a market) and coordinated (like a firm) without needing a single or small group of power brokers to facilitate the coordination.

    Said another way, bitcoin was the first example of money going from being controlled from a small group of firm-like entities (central banks) to being market-driven. What cryptocurrency represents is the technology-enabled possibility that anyone can make their own form of money.

    Whether or not bitcoin survives, that Pandora’s Box is now open. In the same way computing and the internet opened up new areas of the economy to being eaten by markets, blockchain and cryptocurrency technology have opened up a different area to be eaten by markets: money.

    The Future of Public Blockchains

    Bitcoin is unique among forms of electronic money because it is both trustworthy and maintained by a large selectorate rather than a small one.

    There was a group that started to wonder whether the same underlying technology could be used to develop open networks in other areas by reducing the transaction cost of trust.[26]

    One group, the monetary maximalists, thinks not. According to them, public blockchains like bitcoin will only ever be useful as money because it is the area where trust is most important and so you can afford to trade everything else away. The refugee fleeing political chaos does not care that a transaction takes an hour to go through and costs $10 or even $100. They care about having the most difficult to seize, censorship-resistant form of wealth.

    Bitcoin, as it exists today, enhances coordination scalability by allowing any two parties to transact without relying on a centralized intermediary and by allowing individuals in unstable political situations to store their wealth in the most difficult-to-seize form ever created.

    The second school of thought is that bitcoin is the first example of a canonical, trustworthy ledger with a large selectorate and that there could be other types of ledgers which are able to emulate it.

    At its core, money is just a ledger. The amount of money in your personal bank account is a list of all the transactions coming in (paychecks, deposits, etc.) and all the transactions going out (paying rent, groceries, etc.). When you add all those together, you get a balance for your account.

    Historically, this ledger was maintained by a single entity, like your bank. In the case of U.S. dollars, the number in circulation can be figured out by adding up how much money the U.S. government has printed and released into the market and how much it has taken back out of the market.

    What else could be seen as a ledger?

    The answer is “nearly everything.” Governments and firms can be seen just as groups of ledgers. Governments maintain ledgers of citizenship, passports, tax obligations, social security entitlements and property ownership. Firms maintain ledgers of employment, assets, processes, customers and intellectual property.

    Economists sometimes refer to firms as “a nexus of contracts.” The value of the firm comes from those contracts and how they are structured within the “ledger of the firm.” Google has a contract with users to provide search results, with advertisers to display ads to users looking for specific search terms, and with employees to maintain the quality of their search engine. That particular ledger of contracts is worth quite a lot.

    Mechanical time opened up entirely new categories of economic organization. It allowed for trade to be synchronized at great distanceswithout mechanical time, there would have been no railroads (how would you know when to go?) and no Industrial Revolution. Mechanical time allowed for new modes of employment that lifted people out of serfdom and slavery.[27]

    In the same way, it may be that public blockchains make it possible to have ledgers that are trustworthy without requiring a centralized firm to manage them. This would shift the line further in favor of “renting” over “buying” by reducing the transaction cost of trust.

    Entrepreneurs may be able to write a valuable app and release for anyone and everyone who needs that functionality. The entrepreneur would collect micro-payments in their wallet. A product designer could release their design into the wild and consumers could download it to be printed on their 3D printer almost immediately.[28]

    For the first 10 years of bitcoin’s existence, this hasn’t been possible. Using a blockchain has meant minimizing the transaction cost of trust at all costs, but that may not always be the case. Different proposals are already being built out that allow for more transactions to happen without compromising the trust which bitcoin and other crypto-networks offer.

    There are widely differing opinions on what the best way to scale blockchains are. One faction, usually identifying as Web 3/smart contracting platform/Ethereum, believes that scaling quickly at the base layer is essential and can be done with minimal security risk while the other groups believe that scaling should be done slowly and only where it does not sacrifice the censorship-resistant nature of blockchains (bitcoin). Just like the debate between Keynesian and Austrian/monetarist views of monetary policy, these views represent different beliefs about the optimal tradeoff point on the Selectorate Spectrum. But, both groups believe that significant progress can be made on making blockchains more scalable without sacrificing too much trust.

    Public blockchains may allow aggregation without the aggregators. For certain use cases, perhaps few, perhaps many, public blockchains like bitcoin will allow the organization and coordination benefits of firms and the motivation of markets while maintaining a large selectorate.

    Ultimately, what we call society is a series of overlapping and interacting ledgers.

    In order for ledgers to function, they must be organized according to rules. Historically, rules have required rulers to enforce them. Because of network effects, these rulers tend to become the most powerful people in society. In medieval Europe, the Pope enforced the rules of Christianity and so he was among the most powerful.

    Today, Facebook controls the ledger of our social connections. Different groups of elites control the university ledgers and banking ledgers.

    Public blockchains allow people to engage in a coordinated and meritocratic network without requiring a small selectorate.

    Blockchains may introduce markets into corners of society that have never before been reached. In doing so, blockchains have the potential to replace ledgers previously run by kings, corporations, and aristocracies. They could extend the logic of the long tail to new industries and lengthen the tail for suppliers and producers by removing rent-seeking behavior and allowing for permissionless innovation.

    Public blockchains allow for rules without a ruler. It began with money, but they may move on to corporate ledgers, social ledgers and perhaps eventually, the nation-state ledger.[29]

    Acknowledgments: Credit for the phrase “Markets Are Eating the World” to Patri Friedman.


    1. https://www.bls.gov/opub/mlr/1981/11/art2full.pdf
    2. https://www.bls.gov/emp/tables/employment-by-major-industry-sector.htm
    3. http://www3.nccu.edu.tw/~jsfeng/CPEC11.pdf
    4. There are, of course, other types of transaction costs than the ones listed here. A frequent one brought up in response to Coase is company culture, which nearly all entrepreneurs and investors agree is an important factor in a firm’s productivity. This is certainly true, but the broader point about the relationship between firm size and transaction costs hold—culture is just another transaction cost.
    5. http://www.fon.hum.uva.nl/rob/Courses/InformationInSpeech/CDROM/Literature/LOTwinterschool2006/szabo.best.vwh.net/synch.html
    6. https://en.wikipedia.org/wiki/Escapement
    7. Fungibility is the property of a good or a commodity whose individual units are interchangeable. For example, one ounce of pure silver is fungible with any other ounce of pure silver. This is not the same for most goods: a dining table chair is not fungible with a fold-out chair.
    8. Piece rates, paying for some measurement of a finished output like bushels of apples or balls of yarn, seems fairer. But they suffer from two issues: For one, the output of the labor depends partially on the skill and effort of the laborer, but also on the vagaries of the work environment. This is particularly true in a society like that of medieval Europe, where nearly everyone worked in agriculture. The best farmer in the world can’t make it rain. The employee wants something like insurance that they will still be compensated for the effort in the case of events outside their control, and the employer who has more wealth and knowledge of market conditions takes on these risks in exchange for increased profit potential.
    9. For the worker, time doesn’t specify costs such as effort, skill or danger. A laborer would want to demand a higher time-rate wage for working in a dangerous mine than in a field. A skilled craftsman might demand a higher time-rate wage than an unskilled craftsman.
    10. The advent of the clock was necessary for the shift from farms to cities. Sunup to sundown worked effectively as a schedule for farmers because summer was typically when the most labor on farms was required, so longer days were useful. For craftsman or others working in cities, their work was not as driven by the seasons and so a trusted measure of time that didn’t vary with the seasons was necessary. The advent of a trusted measure of time led to an increase in the quantity, quality and variety of goods and services because urban, craftsman type work was now more feasible.
    11. https://unenumerated.blogspot.com/2017/02/money-blockchains-and-social-scalability.html. I am using the phrase “coordination scalability” synonymously with how Nick uses “social scalability.” A few readers suggested that social scalability was a confusing term as it made them think of scaling social networks.
    12. 150 is often referred to as Dunbar’s number, referring to a number calculated by University of Oxford anthropologist and psychologist Robin Dunbar using a ratio of neocortical volume to total brain volume and mean group size. For more see  https://www.newyorker.com/science/maria-konnikova/social-media-affect-math-dunbar-number-friendships. The lower band of 15 was cited in Pankaj Ghemawat’s World 3.0
    13. https://www.jstor.org/stable/2938736
    14. http://discovermagazine.com/1987/may/02-the-worst-mistake-in-the-history-of-the-human-race
    15. Because what else would you want to do besides eat bread dipped in fresh olive oil and drink fresh beer and wine?
    16. From The History of Money by Jack Weatherford.
    17. It also allowed them to squeeze out competitors at different places in the supply chain and put them out of business which Standard Oil did many times before finally being broken up by anti-trust legislation.
    18. http://www.paulgraham.com/re.html
    19. Tomorrow 3.0 by Michael Munger
    20. http://www.paulgraham.com/re.html
    21. There were quite a few things, even pre-internet, in the intersection between markets and firms, like approved vendor auction markets for government contracting and bidding, but they were primarily very high ticket items where higher transaction costs could be absorbed. The internet brought down the threshold for these dramatically to something as small as a $5 cab ride.
    22. The Long Tail was a concept WIRED editor Chris Anderson used to describe the proliferation of small, niche businesses that were possible after the end of the “tyranny of geography.” https://www.wired.com/2004/10/tail/
    23. From Wikipedia: “Fiat money is a currency without intrinsic value that has been established as money, often by government regulation. Fiat money does not have use value, and has value only because a government maintains its value, or because parties engaging in exchange agree on its value.” By contrast, “Commodity money is created from a good, often a precious metal such as gold or silver.” Almost all of what we call money today, from dollars to euros to yuan, is fiat.
    24. Small institutions can get both coordination and a larger selectorate by using social norms. This doesn’t enable coordination scalability though as it stops working somewhere around Dunbar’s number of 150.
    25. Visa processes thousands of transactions per second, while the bitcoin network’s decentralized structure processes a mere seven transactions per second. The key difference being that Visa transactions are easily reversed or censored whereas bitcoin’s are not.
    26. https://medium.com/@cdixon/crypto-tokens-a-breakthrough-in-open-network-design-e600975be2ef
    27. https://medium.com/cryptoeconomics-australia/the-blockchain-economy-a-beginners-guide-to-institutional-cryptoeconomics-64bf2f2beec4
    28. https://medium.com/cryptoeconomics-australia/the-blockchain-economy-a-beginners-guide-to-institutional-cryptoeconomics-64bf2f2beec4
    29. https://twitter.com/naval/status/877467629308395521