Issue 18 / Pivot

December 21, 2022
Rectangles and circle pattern

Ghost Ships

Miriam Posner

What happens when ships become data?

It seemed like half of Los Angeles had turned out for boat tours at the Port of Long Beach: parents corralling toddlers, couples on dates, even dog owners in line for pet-friendly tours. The port offers free guided tours to the public once a year, a sort of goodwill gesture to the community that has suffered decades of pollution as a result of its activity, and that of the adjoining Port of Los Angeles. After two hours of waiting, I filed onto an erstwhile whale-watching tour boat, where I took in the port’s enormous container ships. Like my fellow tourists, I was excited to get a glimpse of the scale of operations necessary to keep the nation supplied with toilet paper, plastic toys, and every other conceivable good.

Squinting against the sun, I tried to imagine the ships another way: as numbers on a screen, cells in a spreadsheet, dots on a grid. I’d been reading about the information transfer that accompanies the movement of these vessels, and I knew that the scale of this data is nearly as impressive as the ships’ sheer size. Ships like those docked at Long Beach are vital links in the global supply chain, but they’re also floating “data terminals,” as the global maritime industry consultancy Lloyd’s Register put it in 2015. Increasingly, these vessels receive and transmit an enormous amount of information: about their position, of course, but also about weather, traffic, temperature, maintenance, staffing, ocean conditions, and much more. The streams of information are so complex that they threaten to exceed humans’ ability to interpret them. That’s partly why many newer vessels—“smart ships,” in industry parlance—use complex algorithms (some of them devised by Google and Microsoft) to chart their courses. Within the next decade, carriers hope to launch fleets of automated or remote-controlled vessels—“ghost ships,” as they’re sometimes called.

Further away from the port, in office blocks and operations centers, fleet management centers house another tranche of data: information about which containers hold which goods, which ships carry which containers, where those ships are headed, and who paid for what. Elsewhere, “quants” with PhDs in astrophysics collate historical data with information about geography, weather, stock prices, and ship movements, searching for opportunities to place stock market bets on global trade.

At the port, I marveled at the feat of coordination represented by all these containers, all of this loading and unloading. But I also knew something strange about the shipping industry: despite all its technology, global shipping is still infamously paper-heavy. Important documents, like bills of lading and letters of credit, tend to pass physically from person to person, from driver to dockworker to engineer to trucker to warehouse supervisor. While a container moves across the ocean, its accompanying paperwork might literally be flown across the world to meet it.

That’s slowly changing: in the cloistered world of global shipping, a gold rush is taking shape, as companies vie to connect and commodify shipping data. Advanced real-time data about shipping promises to improve the speed and reliability of global logistics. But it could also have other, stranger implications. As ships edge toward automation, the prospect of rich shipping data makes it increasingly possible to imagine a future in which shipping is controlled by machines. And because shipping and banking are so deeply intertwined, better data could attune the movement of cargo near-seamlessly with the priorities of financial markets.

An Industry Built on Paper

Shipping may be unusually dependent on paper, but it’s not for any lack of data. And there are lots of different kinds, including data about ship locations, ships themselves, what ships are carrying, and the buyers and sellers of ship cargo. Ship locations are relatively easy to come by: anyone with an internet connection can monitor the movement of large ships. The UN’s International Maritime Organization requires large vessels to transmit their location information using a VHF-radio protocol called the automated identification system, or AIS. A Google search will yield numerous portals where you can view ship movements in near-real time.

Data about what’s happening on ships is increasingly sophisticated, but it’s usually proprietary, held mainly by the ship operators. The newest shipping vessels are covered with sensors of all kinds, monitoring everything from cargo temperatures to fire hazards to hull conditions. On the bow of a ship’s mast, an anemometer might measure wind speed and direction. On the bottom of the hull, an echosounder can detect the depth of the seabed. In the engine room, meters measure the flow of fuel, monitoring the engines’ efficiency and condition. Increasingly, the data is reported back to shore in near real-time: 5G technology and low-Earth orbit satellites have increased the practicability of worldwide connectivity. This is important, since technologists in the shipping industry envision a near future in which one captain controls a fleet of crewless ships from an onshore computer. But detailed as this shipboard information is, it’s generally not available to anyone outside the ship operators.

Things get even trickier if you want to get ahold of information about what ships are carrying. That data exists, but it tends to be stored within the software systems of shipping companies and amongst the paperwork of customs agencies—there’s no central clearinghouse for information about cargo. AIS data can tell you where ships are, but it can’t tell you what they’re carrying. In the absence of real-time cargo data, companies like CargoMetrics use historical data and proxy information (about which types of vessels are at sea and where they’re headed) in an attempt to derive information about which cargo is in motion. Figuring out who’s selling to whom is even more complicated. There’s no one organization that keeps track of this kind of transaction information, and supply chains are rife with small outfits that spring up and then vanish from sight at a confounding pace.

So the data exists, but the paper persists. The various stores of information are in separate, locked-down databases, kept apart by competing business interests, incompatible data models, and differing legal frameworks. To make things more complicated, when we talk about the “supply chain,” we’re not really talking about one industry; instead, we’re talking about a stunning variety of disparate players, all engaged in moving stuff: freight forwarders, charterers, drayage companies, container lines, truckers, terminal operators, and chassis providers, to name just a few. Each has its own data, but as often as not, information moves between operators on paper. “It is not uncommon for a shipping document package to contain fifty sheets of paper that must, in some cases, be exchanged between thirty different stakeholders,” reports the Digital Container Shipping Association (DCSA). In the absence of a unified mode of communication, paper patches over the gaps between systems.

The bill of lading, for example, a critical document that describes a shipment and acts as a bill of title, is almost always paper: the DCSA reports that only 1.2 percent of bills of lading were digitally transmitted in 2021. At each stop on the cargo’s journey, the physical document must pass from hand to hand until it reaches its destination. In the UK and some other countries, there was until recently no legal provision for the possession of an electronic title; anyone seeking to prove ownership had to have that ownership on paper. Different countries have different laws, different industries need different information, and different carriers each have their own requirements. During the height of the Covid-19 pandemic, when many flights were grounded, some containers were stuck at ports because the paper documents necessary to release them couldn’t be couriered via airfreight. The global supply chain is a stunning feat of communication and technology, but surprisingly little of that communication is accomplished digitally at the moment.

So the issue is not that movement isn’t tracked and computerized; it’s that it is tracked too many different ways, in scores of different proprietary systems, most of them locked down. Individual carriers might have granular detail about their vessels’ condition and cargo, but in the absence of a network that stitches all of that information together, the big picture remains incomplete. “You have to connect data from multiple data sources, and to be able to do that, you have to be able to access data that resides within the systems of different carriers, different freight forwarders, different customs brokers,” logistics expert Inna Kuznetsova told the Journal of Commerce in 2019. Paper documents like bills of lading may be unwieldy, but they’re still more convenient to exchange than unintelligible or inadmissible digital records.

Data standards would make a difference: if information could be exchanged in a predictable format, companies could share it electronically. These standards do exist, but there are also significant barriers to their adoption. In 2015, the United Nations Centre for Trade Facilitation and Electronic Business (UN/CEFACT) published a set of data models that lay out definitions and specifications for trade documents, designed to be encoded in data exchange formats such as JSON or XML. If a trading partner knows how information will be formatted, it can design a computer system to interpret that data without human intervention. Data standards aren’t unlike the shipping container itself: once everyone agreed on the box’s dimensions, containers could be swapped in and out without any need for discussion. Likewise, if everyone agrees to use the same data standards, information can be handed over as easily as a shipping container moves from cargo hold to truck chassis.

The existing guidelines are complex, however—they have to be, in order to account for every possible use—and require a fairly high level of technical proficiency to implement. There’s no single organization or governing body with enough heft to impose standards on the entire industry. Some countries, industry groups, and regulators are concerned about electronic fraud. In some jurisdictions, laws need to be revised to permit electronic documentation. And some carriers and container lines are reluctant to make their data available to competitors because, while standardization could benefit the industry as a whole, individual companies’ data represents a potentially valuable asset. Standards, in other words, are inflexible, and the world of global trade is a chaos of variables.

Covid has only heightened the urgency of efforts to network shipping data. “Covid was the spark that ignited everything,” Thomas Bagge, CEO of the DCSA, told the Global Trade Review in 2021. “There is a significant number of stakeholders now motivated to change the opaque and antiquated processes that are still common in international trade.” In March of 2022, the Biden administration announced the launch of Freight Logistics Optimization Works, an advisory group tasked with developing a “proof-of-concept information exchange” designed to ease supply-chain congestion. The Port of Long Beach, an advisory group member, pledged to work with defense contractor UNCOMN to create the “Supply Chain Information Highway,” designed to “liberate actionable data across supply chain nodes and organizations.”

Dematerialize the Bill of Lading

Despite all the complications, shipping’s lingering dependence on paper makes it an irresistible target for tech companies. Numerous startups seek to “dematerialize” the bill of lading and other documents, while others offer platforms that promise to connect disparate streams of data, often with the use of AI. So far, though, the result of all this activity has been an even knottier tangle of data, with no single standard or company emerging as the clear consensus choice. As separate companies seek to optimize logistics sectors like container movement, fuel demand, shipping volume, and charter prices, it gets even more difficult to turn all of this data into a coherent whole. “I’ve got 99 problems, but data’s the biggest one,” writes logistics journalist Eric Johnson. “It’s like the faucet turned on, and no one has been able to slow down the flow.” Data is flowing from every direction, but no one seems to be able to turn all of this information into a unified stream of global activity.

Yet the reward for integrating supply-chain data could be immense. To understand why, we need to take a brief tour of shipping finance—a sector that’s as complicated as you might expect in a system that spans the globe. One area of it, called trade finance, addresses the lag time between the issuing of an order (for goods or commodities) and the delivery of that order to the buyer. Sellers often need money to purchase supplies and labor to procure or assemble the goods. Buyers, however, usually resist paying for goods until they’re delivered. To plug that gap, lenders extend loans to either buyers or sellers, charging interest or a fee in return. There are countless variations, of varying degrees of complexity, but the purpose is generally to keep the wheels of commerce moving smoothly by advancing cash to suppliers.

It’s time-consuming, by modern standards, to arrange these loans. Lenders need to know that their loans will be repaid, so they want to know about the borrower’s finances, market conditions, transport risks, and every other piece of information they can get their hands on. Documents need to be exchanged. People need to sign off. It can take months. But what if all that information could be networked, so that it could be accessed and evaluated in an instant? What if loan decisions could be made algorithmically? What if—irresistibly!—these algorithmic transactions could be executed by “smart contracts” on the blockchain? These transactions could take place instantaneously, but you’d need to have really accurate data about a company’s solvency, its loan history, and the market for whatever’s being transported. You’d need, that is, a lot of data: about what’s being exchanged, which shipments are already underway, about the market in general, and about who’s doing the buying and selling. As we’ve seen, that data exists, but it has yet to be networked.

An enormous amount of money is at stake. In 2020, global trade finance had a value of $5.2 trillion, according to McKinsey & Company, and the market is only expected to grow. Moreover, a more “transparent,” legible market in trade finance could herald the entrance of a new set of market players. Maritime finance has traditionally been the province of large financial institutions and shipping insiders. Shipping has long had a reputation for deals conducted in smoky backrooms by insiders steeped in arcana, such as bunker rates, freight-forwarding agreements, and value maintenance clauses. Few non-expert investors have felt comfortable wading into this environment. But what if there was a kind of internet of cargo—an easily legible map of what’s traveling where?

Shipping finance has already absorbed a significant amount of private equity and hedge fund investment, and has spawned IPOs and even NFTs. (A company called Infinity Maritime allows investors to purchase tradable “MetaUnits” that represent a share in a shipping fleet.) If an internet of cargo existed, investments could be much more data-driven, making trade finance an even more alluring target for investors.

The Shadow Life of Grain

What’s more, where there’s debt, there’s a sellable asset. Banks are talking increasingly about securitizing debt from trade finance, meaning they want to package these loans and sell them on to investors on a public market. A container’s worth of grain, for example, might be paid for with a loan that is then sold as a security, or part of a security. As a ship carries the grain across the ocean, the loan could change hands multiple times as market conditions change. Grain prices drop and an optimistic investor steps in to purchase the loan, betting that prices will rise before the ship makes landfall. They do, and the lucky investor sells the loan on, pocketing the price difference.

“Given that global demand for trade finance already outstrips supply by about US$1.5tn a year, we see huge potential for a thriving secondary market to stimulate trade in goods and services,” Surath Sengupta, HSBC’s global head of trade portfolio management, told Global Trade Review in 2019. In this way, grain becomes valuable not only for its existence as grain, but for its shadow life as part of a financial instrument. As the physical grain moves across the ocean, ownership of its debt could ping-pong around the globe.

Accurate cargo and commodity data could also provide an invaluable window into the flow of commerce. Imagine if you knew at all times which commodities were headed where. You could determine whether a market is about to be flooded or whether a shortage of, say, wheat means the cost of bread is about to skyrocket. “Shipping holds the no-shit, honest truth of what the economy is doing,” angel investor Doug Doan told Institutional Investor in 2016. Traders with special insight into shipping data could identify mispricing in the stock market—perhaps GE’s stock is too high, given the flood of cheap washing machines headed our way. They could then exploit that information gap to pocket the difference in price. Currently, data analysts can make estimates about ocean freight based on vessel type and economic indicators, but these are just educated guesses. If a database was constantly updated as transactions took place, this real-time data would make it possible for those with access to place more aggressive, potentially more lucrative bets.

Given the potential value of this information, the question of who will control or have access to the network of comprehensive shipping data is of urgent importance to many players within the worlds of logistics and finance. Currently, the major shipping lines, like Maersk and Hapag-Lloyd, have privileged access to a great deal of detailed information, because data about their customers and cargo is locked into their technology platforms. National agencies, such as US Customs and Border Protection, hold another large swath of data, but it’s not linked with other datasets. A growing number of startups is attempting to harvest and synthesize various sources of data, but to obtain proprietary data would require the participation of ports and shipping companies.

Here, China has an advantage: its government-sponsored National Public Information Platform for Transportation and Logistics, or Logink, collects data at a number of participating ports in Asia and Europe. Because Logink is part of ports’ information systems, it absorbs key documents like manifests, bills of lading, and customs documentation, and then integrates these documents with countless other streams of information, such as AIS data and information about historical trends. Logink is such an impressive feat of engineering that its growth is making parts of the US government nervous: Michael Wessel, a commissioner of the United States–China Economic and Security Review Commission, a federal agency, told the Wall Street Journal in December 2021 that Logink represents “a treasure trove of intelligence of national security and economic interest.”

The future of shipping could, in investors’ fantasies, be an almost pure expression of financial markets, with automated ships following routes dictated by live market data. Consider if the price of grain falls below a threshold necessary to make a profit from a carrier’s voyage. There’s nothing to prevent a (possibly autonomous!) ship from changing its route and setting course for a more lucrative market. In a world of perfect information, what would prevent an algorithm from revising market and fuel costs mid-voyage, rerouting a ship on the fly, and sending its cargo to a different destination entirely? Efficient though this dynamic system might be, it hardly seems likely to result in more predictable supply chains. On the contrary, events like the stock market crash of 2008 or the “flash crash” of 2010 suggest that global shipping’s deeper integration with financial markets could introduce another layer of instability.

If one is inclined to doubt carriers’ willingness to cast aside human welfare in the pursuit of short-term profits, the growing problem of crew abandonment offers a corrective. When a ship’s debts are worth more than the ship itself, shipowners have increasingly opted to simply stop paying crew members’ wages and abandon the ship where it is. In 2021, at least ninety-one ships were abandoned by their owners, their crew members still aboard. Seafarers might be abandoned in Mogadishu, as one crew was in 2021, or Umm Al Quwain in the UAE, as another crew was for forty-three months starting in 2017. Stranded onboard without visas or the means to make their way home, seafarers’ ordeals can last for months or years. One crew was abandoned off the port of Ajman in the UAE for eighteen months. These instances of abandonment, cruel though they may be, are simple calculations on the owners’ parts: when value x exceeds value y, perform action z. Full and accurate data would allow for these kinds of calculations on every load a ship carries: algorithmic apportionment of goods, powered by increasingly granular data.

Flesh into Fractions

In the postmortems of the worst days of the pandemic’s supply-chain crisis, technology experts were quick to blame antiquated communication and information systems. “Supply chain management is facing its own pandemic of outdated processes,” EPSNews reported in 2021. But modern shipping’s abandoned seafarers are proof that more precise data doesn’t always work in favor of the common good. In a way, the seafarers were abandoned because of efficient access to information: to the workers’ detriment, their wages outweighed the current value of the ship. It’s just math. And this kind of math, in which humans are measured against market values and found wanting, seems to be repeated over and over again in the history of global logistics.

Shipping, in fact, has a special relationship with data. In some accounts, the growth of international trade in the sixteenth century helped to create the kind of information we now call data. Maritime trade generally involves the conveyance of objects over long distances, to unseen trading partners. But how to represent a product’s qualities in mutually intelligible terms, if the buyer can’t judge it with their own eyes? Seafaring merchants had to devise standardized categories that would stand in for the characteristics of particular goods: Tellicherry pepper, for example, or long-grain rice. With this act of categorization, individual peppercorns can be aggregated, made interchangeable, and exchanged for an agreed-upon rate. It’s this kind of quotidian bookkeeping that produces data: particular entities get grouped under one name and added up on a spreadsheet.

But sixteenth-century seafarers weren’t just trading in peppers and rice. Historians tell us that the transatlantic slave trade was the engine behind the explosion in global shipping from the sixteenth through nineteenth centuries. And even as merchants were codifying standards for coffee and tea, they were keeping detailed ledgers of human beings, carefully categorizing them by physical characteristics and market value. “Compilers of slave ship manifests,” writes the historian Jessica Marie Johnson, “participated in the transmutation of black flesh into integers and fractions.”

Logistics, enslavement, and the advent of data are all tangled up in the historical record. And race continues to structure the global supply chain. Many of today’s abandoned seafarers are from southeast Asia, from countries left impoverished by a legacy of colonization. The warehouse workers who handle the goods onshore are disproportionately people of color. The very undersea cables that deliver data between continents retrace maritime routes first mapped by the transatlantic slave trade. And of course, we wouldn’t really have global supply chains, at least not to the extent we do, if the market didn’t award a substantially lower value to the labor of people from the Global South.

The scholars Stefano Harney and Fred Moten write that logistics is a way to assert ownership. Atop a roiling ocean, rocky terrain, or noisy metropolis, logistics imposes a grid of straight lines. “Logistics aims to straighten us out, untangle us, and open us to its usufruct, its improving use,” they write. “Such access to us, in turn, improves the flow line, the straight line.” The internet of cargo, the supply chain information highway, and similar efforts promise to use data to pave over the world’s contingencies—restive workers, ecological disasters, unpredictable demand. But does logistical efficiency equate with the common good?

For as long as logistics has existed, it has viewed humans and their environment as friction to be eliminated. The enmeshment of logistical data with high-speed trading simply completes a circuit initiated with the inscription of human beings into slave-ship manifests: it submits all that the supply chain touches to the disciplining logic of the marketplace. In its single-minded quest for ownership, logistics has the goal, write Harney and Moten, “quite simply and starkly of preventing us from taking care of one another, from looking out for one another.”

It should make us wonder: how sure are we that a speculative market in shipping data will not result in a more unjust world? Are we certain that we want to assign responsibility for distributing goods to algorithms we don’t understand? We have the proven ability to codify logic that demands the performance of atrocity—particularly for those judged subhuman as a result of their race. Perhaps, then, when the shipping and finance industries laud technology’s power to make the supply chain “more agile” and “responsive to market forces,” as the DCSA put it in 2022, we should question what that actually means. Perfect data could allow cargo to move more efficiently, but it would also be more likely to move in lockstep with the inscrutable whims of financial markets.

Into the Sea

It’s hard not to get a little carried away imagining a future conjured by all of the coming technological changes the industry touts: the world’s oceans transformed into a computer-operated game of Battleship, with faceless autonomous vessels tracing routes dictated by algorithmic finance. In reality, these changes will be piecemeal, probably painfully slow, with a lot of haggling over policy and standards. Parts of global shipping may never digitize. The forces pushing toward a future of data-driven shipping, however, are strong.

Back in Long Beach, our tour boat slid neatly into its berth and I followed my fellow passengers back onto land, dodging toddlers and wishing I’d worn sunblock. I started the walk back to my car, thinking about something I’d learned recently, in Christina Dunbar-Hester’s Oil Beach. In the 1950s, the land around the Port of Long Beach began to slide into the ocean, the result of an intensive program of extraction that pumped billions of barrels of oil from the nearby Wilmington Oil Field. Officials managed to stave off the subsidence by pumping seawater underground, re-pressurizing the eroding land. So all of us here in the port parking lot were treading on ground being held in place by the sea. Logistics had drained the land of its resources and logistics had reinflated it, forcing the ground into Harney and Moten’s “straight line.”

Whether that equilibrium will hold, however, is an open question, since rising sea levels threaten to inundate parts of the port by 2050. The current plan to defend the port entails shoring up infrastructure, hardening the port’s defenses against rising tides and extreme weather. I wondered, though, about the wisdom of this. Instead of doubling down on the system that brought us here, I thought, what if we stopped to ask whether the logistical infrastructure we’ve built is the one we actually want?

Miriam Posner is an assistant professor at the University of California, Los Angeles, and is writing a book on supply chains and data for Yale University Press.

This piece appears in Logic's issue 18, "Pivot". To order the issue, head on over to our store. To receive future issues, subscribe.