Like Sam, she was the child of academics—her father, Glenn Ellison, had been the head of MIT’s Department of Economics. Like Sam, she was someone for whom math had played an important role early on—she’d first heard of Jane Street from the math competitions that the firm sponsored to meet young people just like her. Like Sam, she had discovered “effective altruism" in college and found in it an intellectually coherent sense of purpose. Originating in the work of Peter Singer in the 1970s, the movement was named by Oxford philosophers who set out, during visits to Ivy League campuses, to convince students to judge the effectiveness of their own lives by counting the number of lives they saved during the 80,000 hours of their careers—by donating the money they made instead of keeping or spending it.

Maybe even more than Sam, Caroline had allowed math to pull her to a moral place. “I was attracted to people thinking about what to do in a quantitative, rigorous way," she said. “Before that, I don’t think I had much of an impulse to do good in the world." And like Sam, she’d been hired by Jane Street as a full-time trader.

But unlike Sam, Caroline was unsure of herself and susceptible to being swayed by the opinions of others. Unlike Sam, she wanted a normal life, with emotions and children and maybe even a sport-utility vehicle in which to drive them around. After a year at Jane Street, she sensed that she was at best average at her job and in any case didn’t have anything like Sam’s feeling for the place, or share his fanaticism about work. “I did feel a bit, like, unsatisfied," she said. “There was something missing. I wasn’t sure I was doing that much good."

In the fall of 2017, she’d been sent by Jane Street back to Stanford, to recruit to high-frequency trading the mathematically gifted friends she’d left behind. Upon arrival she’d called Sam and asked him to meet. Over coffee in Berkeley, Sam was cagey about what he was up to. “It was, ‘I’m working on something secret and I can’t talk about it," recalled Caroline. “He was worried about recruiting from Jane Street. But after we talked a while, he said, ‘I guess maybe I could tell you.’"

By the end of their chat, Caroline thought that maybe she should quit Jane Street and join Alameda Research, the crypto trading firm Sam was secretly building. The work felt familiar: She’d be doing the same sort of research in crypto for Sam’s new quant fund that she was doing in equities for Jane Street. Asking questions like: Does the price of bitcoin vary significantly with time of day? Or: How does the price of bitcoin move in relation to the prices of all the other coins?

But the underlying purpose of the work would be entirely different, because she would be doing it only with other effective altruists. Jane Street, as Sam put it, was “just a place where people come to work each day to play some games and increase the number in their bank account, because what the fuck else are they going to do with their lives?" Alameda Research was going to be different—a vessel to save some vast number of lives.

By Wall Street standards, Jane Street was not a greedy place. Its principals did not flaunt their wealth in the way that the guys who had founded other high-frequency trading firms loved to do. They didn’t buy pro sports teams or hurl money at Ivy League schools to get buildings named for themselves. They were not opposed to saving a few lives. But Jane Street was still on Wall Street. To survive, it needed its employees to grow attached to their annual bonuses and accustomed to their five-bedroom Manhattan apartments and quiet, understated summer houses in the Hamptons.

The flood of effective altruists into the firm was worrisome. These were new graduates taking jobs on Wall Street for the express purpose of making money to give away. They arrived with their own value system. They had their own deep loyalties to something other than Jane Street. They didn’t have the usual Wall Street person’s relationship to money; they didn’t care about their bonuses in the ways Wall Street people were supposed to care. Sam Bankman-Fried had been able to leave his lucrative Jane Street job for a nutso plan in 2017 to try to make even more money on his own because he had no material attachments. “It wasn’t going to cut into his lifestyle, because he didn’t have a lifestyle," as one former Jane Street trader put it.

Caroline was the second effective altruist to quit Jane Street’s New York office in a matter of months, by telling them, on her way out the door, that she was leaving to maximize her expected value. This time they were ready for it. Her manager, a Jane Street partner, pulled her into his office. “He was pissed off," she said. “Just cold."

He then proceeded to challenge her deepest beliefs. Effective altruism made no sense, he said, and then laid out the many senseless things about it: There was no accurate way to measure the consequences in the distant future of your present actions; if such measurement did exist, it would likely be done best by the market; no one would pay you as much as Jane Street paid you, ergo, your highest value was at Jane Street. And so on.

This was a first: a Wall Street trading firm whose business was premised on its ability to hire the brightest mathematical minds was now compelled to make arguments about the limits of math in life. “It was this hour-long conversation with the partner where he tried to convince me to stay because utilitarianism was flawed," recalled Caroline. “I thought, this isn’t something we’re going to solve in an hour."

She was surprised by the Jane Street partner’s lack of sentiment after his argument hadn’t landed as he imagined it might. “Once he saw I wasn’t going to change my mind about utilitarianism, my stuff is in a box," she said. No one on the trading floor offered her so much as a hug. With her box of stuff, Caroline Ellison walked out onto the streets of lower Manhattan. In March of 2018 she moved to the Bay Area and started working for Sam Bankman-Fried.

The situation inside Alameda Research wasn’t anything like Sam had led Caroline to expect. He’d recruited 20 or so effective altruists, most of them in their 20s, all but one without experience trading in financial markets. Most neither knew nor cared about crypto; they had just bought into Sam’s argument that it was this insanely inefficient market in which they might use his Jane Street-like approach to trading to extract billions. They were now all living in Sam’s world, and they weren’t hiding their unhappiness.

“He was demanding and expecting everyone to work 18-hour days and give up anything like a normal life, while he would not show up for meetings, not shower for weeks, have a mess all around him with old food everywhere, and fall asleep at his desk," said Tara Mac Aulay, a young Australian mathematician who was, in theory, running the company with Sam. “He did zero management and thought that if people had any questions, they should just ask him. Then in his one-on-ones with people, he’d play videogames."

The firm’s finances were already in a state of chaos. They’d started small a few months before, with the half-million left over after taxes from Sam’s Jane Street bonus, but within a few months they’d persuaded other, richer effective altruists to lend them $170 million to trade crypto. They’d lost millions of it already, though how many millions no one could say for sure. In February their trading system had lost half a million dollars per day.

Amid the turmoil, Sam created another, supposedly better system. Modelbot, it was called. Modelbot had been programmed to scour the world’s crypto exchanges for inefficiencies to exploit. If for even a few seconds it was possible to buy bitcoin on some Singaporean exchange for $7,900 and sell it for $7,920 on an exchange in Japan, Modelbot would do it over and over again, thousands of times per second.

But that example made Modelbot sound simpler than it was. Modelbot was programmed to trade roughly 500 different crypto coins on 30 or so different crypto exchanges, most of them in Asia, all of them basically unregulated. The tulip-bulb-like explosion in crypto over the previous year had encouraged the creation of hundreds of new coins. Modelbot made no distinction between the better-known coins with deep markets, like bitcoin and ether, the Ethereum blockchain’s token, and the so-called shitcoins that hardly traded at all, like Sexcoin and PUTinCoin and Hot Potato Coin. Modelbot just hunted for any coin it could buy at one price in one place and sell in another at a higher price.

Modelbot was maybe the biggest point of disagreement between Sam and his management team. Sam’s Release-the-Kraken fantasy was to hit a button and let Modelbot burn and churn through crypto markets 24 hours a day, seven days a week. He had not been able to let Modelbot rip the way he’d liked—because just about every other human being inside Alameda Research was doing whatever they could to stop him. “It was entirely within the realm of possibility that we could lose all our money in an hour," said one. One evening, Tara argued heatedly with Sam until he caved and agreed to what she thought was a reasonable compromise: He could turn on Modelbot so long as he and at least one other person were present to watch it, but should turn it off if it started losing money. “I said, ‘OK, I’m going home to go to sleep,’ and as soon as I left, Sam turned it on and fell asleep," recalled Tara. From that moment the entire management team gave up on ever trusting Sam.

Once she’d started, Caroline heard the gory details from Sam’s disaffected partners. At the end of Caroline’s second week, they called a meeting to announce that they had persuaded the rich effective altruists who had lent them the $170 million to demand its return—which meant that in a few weeks Alameda Research would have no money to trade with. Caroline did not know whom to believe. She felt deceived that Sam had not warned her of how poorly Alameda Research was doing before she quit her job at Jane Street. And yet she didn’t know any of these other people. She thought she knew Sam, but she also thought that if the entire management was talking about quitting in protest, and the investors were taking their money back, someone must know things about Sam that she did not. It was at this point that Caroline called her mother and cried.

On top of the trading losses, $4 million worth of XRP, a cryptocurrency issued by the exchange RippleNet, had simply vanished from Alameda’s accounts. Sam suspected that it had been sent from an exchange in the U.S. to one in South Korea, and that the South Korean exchange was just dragging its feet in crediting it to Alameda’s account. The other members of the management team were unconvinced. They insisted that Sam stop trading so that they could figure out where their Ripple had gone.

At length Sam agreed. He stopped trading for two weeks. The other members of the management team confirmed that millions of dollars’ worth of Ripple was indeed missing. At which point everyone except Sam, and perhaps one other senior manager, became upset. “We thought we needed to tell investors and tell employees so they could reconsider their options, but Sam hated that idea," said one of the firm’s managers. Sam continued to insist that the missing Ripple was no big deal. He told his fellow managers that in his estimation there was an 80% chance that it would eventually turn up. Thus they should count themselves as still having 80% of it. To which one of his fellow managers replied: After the fact, if we never get any of the Ripple back, no one is going to say it is reasonable for us to have said we have 80% of the Ripple. Everyone is just going to say we lied to them. We’ll be accused by our investors of fraud.

That sort of argument just bugged the hell out of Sam. He hated the way inherently probabilistic situations would be interpreted, after the fact, as having been black and white, or good and bad, or right and wrong. So much of what made his approach to life different from most people’s was his willingness to assign probabilities and act on them, and his refusal to be swayed by any after-the-fact illusion that the world had been more knowable than it actually was.

By early April, the other executives at Alameda Research had grown to fear how little Sam worried about where exactly their money was. They were making 250,000 trades a day, and their system had somehow lost, or failed to record, some large number of them. Among the many problems their shoddy record-keeping caused was the difficulty of filing an honest tax return. “How are we going to pass an audit if we’re missing 10% of our transactions?" asked Tara.

The missing Ripple was the final straw. Apart from a shared alarm at his recklessness, the members of the management team were not perfectly unified in their opinions of Sam. Tara had long since decided that he was dishonest and manipulative. Ben West, another manager, still thought him well-intentioned—but terrible at his job. But all felt themselves on a suicide mission. “I had a conversation with Tara and Peter [McIntyre, another executive]," recalled Ben, “and we were talking about how to help Sam and the conversation changed to: How do we get rid of Sam?"

In the end, for Sam to leave he had to want to leave, and Sam did not really want to leave. And so, on April 9, 2018, his entire management team, along with half of his employees, walked out the door, with somewhere between $1 and $2 million in severance. The investors had no real idea whom or what to believe, or even how to figure out whom or what to believe. “There may be ways in which I shouldn’t trust Sam, but it felt nuanced," said one. They’d all made their money in startups; they all knew that startups were chaotic. Now they had to decide: Was Sam a reckless, phony effective altruist who was going to steal or lose all their money, or were these other people simply unsuited to working in a startup hedge fund?

It was one or the other, an either/or question, to which they responded probabilistically. Just about all of them kept money invested in Alameda, but just about all of them reduced the size of their investments. The capital at Sam’s disposal plunged from $170 million to $40 million. He wouldn’t be able to trade as much as he had before, but he could still trade.

What happened next, in retrospect, seems faintly incredible. With no one left to argue with him, Sam threw the switch and let Modelbot rip, and it began to make lots of money.

Then Alameda finally found the $4 million worth of missing Ripple. First they figured out its travel itinerary: it had been sent from Kraken, a U.S. crypto exchange, to an exchange in South Korea called Bithumb. Then they figured out that the computer languages used by the two exchanges were not perfectly compatible. Inside Bithumb, the employees saw huge amounts of Ripple piling up without any indication of who it belonged to. Once Sam figured out where the missing coin was meant to be, he telephoned the Bithumb exchange directly. The call was transferred around the company about three times before a voice finally came on the line and said, “Are you the fucker who sent us like 20 million Ripple tokens? How the fuck are you only calling us now?" In the background Sam heard someone shouting, “Holy fuck, we found them!"

They’d even paid their taxes. And they resumed making millions of dollars a month in trading profits. Alameda wasn’t the same company, however. They were no longer a random assortment of effective altruists. They were a small team who had endured an alarming drama and now trusted Sam. He’d been right all along! To those who remained—and even to some who had quit—Sam went from someone they weren’t quite sure about to a leader to be followed even if they didn’t completely understand what he was doing, or why.

An odd company from the start had just become even more odd. The people inside were those most able to tailor their thoughts and feelings to those of its creator.

“To Sam: Sorry I wrote all of this in the third person," wrote Caroline, in late 2018. “I didn’t decide to send it to you until the end."

The turmoil had been stressful for Caroline. But she’d left her job at Jane Street and had no obvious place to go, and so, even though she wasn’t sure whom or what to believe, she’d stuck it out. The dust had settled on what was being called The Schism, and only Sam was still ruminating on the meaning of it all. Alameda Research had righted itself and was consistently profitable. All was not right with Caroline, however. Which was why she was now writing to her boss.

“What is the problem?" she asked, in what at first glance appeared to be a business memo.

I have pretty strong romantic feelings for Sam.

Not a business memo! It was merely written in a businesslike manner.

Why is this a problem?

These feelings consume a ton of my brainspace

This prevents me from thinking about other important things and takes up a lot of my time

These feelings can often be positive or feel good but are on net unpleasant

They affect my ability to work

—Mainly by amplifying work-related feelings in unhelpful ways e.g. “I did a bad job on this thing"-> “now Sam will hate me"-> sad

The memo went on for four pages. She clearly wanted to be heard, and equally clearly sensed that her intended audience was likely to be playing a videogame as he half-listened. She pressed on anyway, sounding logical, or at least reasonable, about what were at bottom emotional needs.

For Sam’s benefit, she imagined what he might be thinking about her. (“The last thing he told me was that he feels ‘conflicted.’ I imagine the conflicting desires here are wanting to sleep with me, etc, vs. worry about negative professional consequences.") She listed the things Sam had done that bothered her, among them giving off “confusing signals, e.g. telling me that he felt conflicted about having sex with me, then having sex with me, then ignoring me for a few months." She attempted an expected value analysis of their clandestine relationship, such as it was, starting with the costs:

• scandal if it comes out

• conflict of interest stuff

• work-related tensions

She ended by wondering if she might not be better off quitting her job at Alameda Research and cutting off all contact with Sam. On the other hand…

• It would be ideal if Sam and I could have a discussion to understand each other’s feelings and reach a conclusion about what to do.

Ideal, perhaps. However, unlikely under the best of circumstances, and circumstances were soon less than best. Not long before Caroline was moved to write her memo to herself and then share it with Sam, Sam had left for what was meant to be a brief trip to Hong Kong. After he read her memo, he called the 15 or so people still working for him in downtown Berkeley to say he wasn’t coming back.

Michael Lewis is the best-selling author of “Moneyball," “Flash Boys," “The Big Short" and other books. This article is adapted from “Going Infinite: The Rise and Fall of a New Tycoon," published this week by W.W. Norton.

Milestone Alert!
Livemint tops charts as the fastest growing news website in the world 🌏 Click here to know more.