Human erasure

Fine print can have coarse consequences. When a married couple went to Disney World’s Ragland Road Irish Pub, they asked if any of the menu items could be made allergy-free. One of them, Kanokporn Tangsuan, had a severe allergy to nuts and dairy. Assured it was possible, they ordered and ate. A little later, while walking around Planet Hollywood, Tangsuan collapsed. She jabbed herself with an EpiPen, but it didn’t help and she died. Tangsuan’s husband sued Disney, whose reflexive response was telling: the fine print of his Disney+ subscription limited recourse to arbitration, and so the wrongful-death case should be dismissed. Though Disney dropped its motion to toss the case after public backlash, that Mickey Mouse could even try shows how powerless we are against the illegibility of modern life.

If I tried to understand all of the fine print in my life, I suspect I would suffer from delirium, like a prisoner stuck in solitary confinement for too long. Maybe I’m being too optimistic. For Alex Hern, a British writer, reading all of his online-service terms and conditions made him want to die. Agreements can be as long as a novella. They’re often written in a language that means something only to people who can get into law school. Some agreements are even like Russian nesting dolls, embedding within terms and conditions links to other terms and conditions. At least understanding is possible, in theory. Negotiating any changes to unacceptable terms and conditions is not. A choice is presented, but the choice is draconian: accept fine print you don’t understand or take your business elsewhere.

The irony is that institutions are supposed to make complexity more legible. Rules save pedestrians crossing the road from having to develop a theory of mind for drivers and calculate the speed of moving traffic. Green means go. Red means stop. More complex situations require more complex institutions. An ancient-Greek merchant needed a legal and financial system to support a voyage across the sea. He couldn’t just trust mariners to take his money, source wares from a distant trade port, and return with everything in hand. The merchant needed bankers to raise funds. The sailors needed “bottomry” contracts to limit liability in the event of a shipwreck. Courts were needed to settle disputes, if any arose. To make such transactions work, complexity had to be expressed as guaranteed interest rates and drachmae. Revenue goes up. Risk goes down.

Institutions — laws, contracts, courts, banks, and payment networks — make it easier for strangers to transact, but they don’t eliminate all difficulties. There are devils among saints. They extend their hand to shake your own and smile as they shoot you from under the table with the other. Uncertainty is the most trigger-happy of all. The cooks at the Ragland Road Irish Pub weren’t conspiring to kill Tangsuan with nuts and dairy. At most, they were unaware of how lethally contaminated their ingredients were. Yet they looked and sounded exactly like saintly cooks who would have done better.

Even a grocer sourcing apples has to reckon with a dizzying list of them. Assessing whether the apples will be lethally contaminated is just the beginning. What variety of apples are we talking about? How sweet or tart will they be? Will they be crisp or soft? Will the shipment consist exclusively of big apples, or will small ones be mixed in? Will the skins be speckled with bruises or smooth with a waxy sheen? Is paying the apple supplier upfront better or worse? What happens if payment is made but the shipment isn’t delivered? What happens if it is, but late? What even motivates the apple supplier? Is it gathering intel so that it can start its own grocer and steal customers? Who compensates whom if there is funny business? Who intervenes if no one can agree to decide who compensates whom in the event of funny business? That’s just for apples. Imagine how much more dizzying the list is for iPhones, which are made of almost 3,000 parts, supplied by almost 200 companies, from almost 30 countries.

The money it takes to reckon with such questions are transaction costs. Conceptualized by economist Ronald Coase in 1937, transaction costs include those of gathering information about buyers and sellers, negotiating trades, and policing compliance with negotiated terms, as well as enforcing the terms if push comes to shove. Coase argued that markets are good at directing economic activity when the cost of transacting isn’t prohibitively expensive. When it is, alternative arrangements, such as the firm, emerge.

Silicon Valley has made its fortune by attacking transaction costs. Before Uber and Lyft, only the hard drug-fueled and other deranged would have waved down a random driver in the city and tried to negotiate a ride somewhere. Likewise, few drivers would have stopped at risk of being robbed or worse. Ride-sharing platforms changed all of that by translating complexity into a few seconds of screen-tapping. E-commerce platforms like Amazon did the same by giving people a way to shop around from the comfort of their homes. Getting dressed, and looking for your keys, and weaving in and out of traffic, and gesturing obscenely at the moron who is incapable of operating a vehicle, and rolling at a creeping pace to find a parking spot, and brushing against strangers in a crowded store before scoping the selection and prices may not cost money. But it costs peace of mind and time, which are money in disguise.

“Platform capitalism,” as it has come to be known, has a poor reputation. Some believe it is destroying everything. Liberal democracy. Our privacy and economic welfare. Mental health. The rights of workers. Culture and the arts. The news media. Our dignity and self-determination. Some of that reputation is earned, but the upside is real and too often ignored by critics. By attacking transaction costs, platform capitalism has been expanding the universe of trade, increasing what can be bought and sold, satisfying preferences that people didn’t even know they could have.

Scoffing at Silicon Valley’s obsessions with blockchain and AI is like experiencing art out of context. Memocoins may be scams. AI may fall short of AGI we were promised. But the techno-utopian vision of the future isn’t that it will be foretold by chatbots and paid for in memecoins. It’s that blockchains and AI will drive transaction costs to zero. Firms will cease to exist. Anything will function as money. Workers won’t be loyal to a single employer. They’ll be more like supercharged freelancers, managing a steady bundle of short-term gigs that demand their skills and pay their prices. There will be markets in everything, satisfying each and every conceivable preference, however weird and rare it may be. Inasmuch as techno-utopia sounds absurd, its absurdity is proportional to how much transaction costs gum up the gears of commerce today.

There is a joke that the global economy runs on Microsoft Excel, but the humour will be lost if the world reorganizes itself on blockchains. Luca Pacioli, an Italian mathematician from the 15th century, wrote about the power of double-entry bookkeeping, which conceptualized people as ledgers with incoming credits and outgoing debits. Whether scribbled on parchment in Renaissance Italy or typed in an Excel spreadsheet, ledgers have long recorded the scores between traders. Though blockchains can evoke images of unserious things like memecoins, they represent, fundamentally, an innovation in the ledger itself. That innovation is now being taken seriously. Even Jamie Dimon, CEO of JP Morgan Chase, is sold on the benefits of blockchain, and he once said buying bitcoin is no different from having a pet rock.

A blockchain is just a shared database that records and verifies the authenticity of transactions. What enthusiasts like about blockchain is that it gives anyone a way to transact with anyone else, subject to fine print. But the fine print doesn’t need to be in a contract outside the blockchain. The fine print can be programmed in the blockchain itself. These “smart contracts” facilitate cryptocurrency trading and speculation today. For example, suppose Bob needs a loan to buy bitcoin, but he only has his personal stash of bitcoin to pledge as collateral. Alice bought bitcoin when it was dirt cheap and now wants to put some of it to work. If Bob asked Alice for a loan on Whatsapp, she’d think it was a scam unless she already knew him. But smart contracts make it so that Alice doesn’t need to know Bob. She doesn’t even need to trust him. She needs to trust the smart contract, which would enforce the terms of the transaction in rigid if-then logic: the interest rate, the duration of the loan, the repayment, as well as the repossession of Bob’s collateral if Bob defaults. Smart contracts on blockchains can let complete strangers transact however they want, from wherever there is an Internet connection, so long as their wishes can be expressed as computer code.

Blockchains and the “smart” transactions they power can get just as complex as Wall Street’s financial operations. Warren Buffet may have called derivatives weapons of mass destruction, but they’re also tools of mass construction. Airlines trade derivatives to hedge against fuel-price volatility, allowing them to operate more predictably. When the market is on their side, farmers trade derivatives to sell future harvests at current prices, giving them the confidence to supply the world with food. Lenders and investors also trade derivatives to protect themselves against risks that would otherwise take the whole economy down. Giving people ways to understand and play the odds, the financial system lowers transaction costs and expands the universe of trade. Occasional destruction is to be expected when you play the odds because the odds also play you.

Smart contracts are no more intelligent than the ink-and-paper contracts that underwrite the economy today. Ambiguous contractual terms aren’t problems to be solved by computer code. They’re features that for centuries have helped humans make sense of an otherwise illegible world. A contract that defines and hard-codes every possible situation is a thing of fiction. Imagining one in any detail is impossible. What people don’t know could fill a university library. What they don’t know they don’t know could fill the universe. So people have agreed to use the next best thing: incomplete contracts, full of imprecise and fluid concepts, which can be interpreted and enforced and even revised by judges when things go awry.

The human brain is not made for the endless possibility of zero transaction costs. This should be self-evident to anyone who has spent hours on Netflix, browsing so intensely that by the time a worthwhile show is found the eyes can’t seem to stay open. That’s why people skip the burdensome browsing. They prefer to watch a limited series that ranks highly on Rotten Tomatoes, or an older film reviewed by Roger Ebert, or a sitcom launching its ninth season with a viral ad. There is hard evidence that when people are unsure of what their best choice is, they stop trying to make it. With more choice at their fingertips than ever before, people need curators to restrict choice more than they need institutions to increase it.

This is where AI is supposed to come to the rescue. Ancient Greek merchants didn’t just have a legal and financial system to make the world more legible. Complexity persisted because neither laws nor bank loans could tell the future. For what institutions lacked the Greeks could turn to oracles, vessels of the divine from whom they could seek counsel before a maritime voyage. They could climb Mount Parnassus and ask the gods, indirectly, whether their fortune would be plundered by pirates or sink to the bottom of the Mediterranean sea. Though AI is far from the superintelligence we were promised, it is still far superior to the fortune-telling priestesses who graced the Greek mountains.

The endless possibility of transaction costs may be too much for the human brain, but it isn’t too much for a superintelligent AI system — a system that can parse all information and tell Apple exactly where to source iPhone parts from, that can write a contract to protect a grocer from all the ways it could be let down by its apple supplier, and that can monitor supply chains so that dockworkers pilfering a shipping container are held liable, all without ever needing to explain how it arrived at such an unfathomably complex chain of conclusions. This world is still far away, but is it impossibly distant? Only fools and high priestesses feign a divine certainty. For all the slop, all the hallucinations, all the pseudo-intelligence, all the energy consumption, all the bubbly investment today’s AI systems are responsible for, they are just getting more impressive.

If the techno-utopian vision comes to pass, the hum of a nearby data centre may keep us awake at night. It won’t matter because we can sleep in the next day. AI systems will be constantly working with other AI systems to satisfy all human desire, the universe of which will have exploded like the big bang after the world reorganized itself on blockchains, making every theoretical transaction a feasible one. If AI systems are smart enough to eliminate transaction costs for us, it may not be long before they do the same for themselves. In the morning, while we’re still asleep and they’re still buzzing, will the hoarse hum of our snoring bother them?

Techno-utopia is a lonely and ignorant world, as ours will one day be under the laws of physics. The cosmos is like a bubble inflating at an accelerating rate, stretching out in all directions faster and faster, resulting in an ever growing rift between the galaxies within. With the passage of time, more and more of the cosmos will have crossed a horizon beyond which its light can no longer reach us. Far in the future, if there is any life left, it will look outside the Milky Way to see what’s out there, but it will see little. Telescopes will receive no light from beyond the horizon. Whoever or whatever is left standing will never know how full and beautiful the cosmos truly is.

As the universe of transactions grows, so does the space between us. Face to face becomes mainframe to mainframe. Handshakes become API calls. People may already be skipping the shops and restaurants more than they used to, preferring instead to shop online or order in. In techno-utopia, they won’t even get the chance to thank their delivery person, who will have been replaced by an autonomous drone. Most people won’t even order in because they will instead instruct their AI-powered kitchens to make their favourite recipes, after sourcing the best and cheapest ingredients on a blockchain from faceless farmers willing to trade.

In the mission to eliminate transaction costs — what has come to be known, more colloquially, as friction — techno-utopians have stuck their target on the right backs. Our own. Transactions are too complex for humans to perfect, like for Tangsuan who misplaced her faith in the Ragland Road Irish Pub. Humans, themselves, are the imperfections. Where the complexity of transacting with others is the greatest of all evils, the most just institutions will be the ones that preclude the need for humans to do the hard work of transacting in the first place. Such institutions could have saved Tangsuan from dying too soon by not letting her live. And that, buried deep in the fine print of the techno-utopian bargain, may be the coarsest consequence of all.