Chapter 4 Manifesting Connection

Substitution and Simulation

There are difficulties in attempting to survey the DH for the critical study of logistics. The most pressing of these is that there are simply not many digital projects that both explicitly address logistics and imagine themselves operating in DH. Jussi Parikka has asked: “Does digital humanities have things to say about, for example, postcolonial issues in digital culture and the strange planetary ties of digital infrastructures, whether those of satellite realities of visual culture, of supply chains of materials, of resource extraction, or of electronic waste as the residual level of dead media?” (Parikka, 451). Seeking projects that both address logistics and identify with DH, we are left with only a handful of examples. In some sense, this is not so surprising. While the first wave of work in DH was concerned with the application of digital methods to areas that lacked them, contemporary logistics was already a digital domain (Berry). Everything from purchasing to production scheduling was governed by enterprise resource planning (ERP) systems produced by companies like SAP, IBM, and Oracle. And conducting DH research inside this domain was difficult. Although examining proprietary ERP packages could provide valuable insights about their operation, such packages have largely inaccessible and opaque interfaces. Even if one could gain access or afford the licensing costs, these systems are meaningless outside their working implementations.

Nor is the data held in logistical systems trivial for researchers to access. Subsumed under an ever-expanding state security apparatus, a proliferation of legal designations, and the formalization of third-party logistics, data regarding traditional sites of material traffic has become inaccessible to the public. Ports, shipyards, and railways—once part of the city—now stand outside it. Warehouses and factories are no longer commercial establishments, but gigantic megastructures set in industrial geographies. Any intervention to render data about this infrastructure accessible must compete with systems that not only possess enhanced access but are materially responsible for the sequestration of this infrastructure for use in the network of global supply. In “the absence of access,” we are left with “a kind of substitute interface . . . to the software actually used in logistical industries” (Rossiter, “Materialities”).

But as Nicole Starosielski has noted, recognizing the “resource operation of digital media” means acknowledging the capacity of DH and allied approaches “both to reconfigure informational resources and to facilitate new extractions of meaning” (402). Some DH projects have thus attempted to offer such reconfigurations, gathering and visualizing data on contemporary port operations, warehouse traffic, and similar areas of internal logistical operation. The following section details distinct categories of projects that offer such substitutions for the critical study of logistics, providing interactive visualizations of logistical processes, new sources of logistical data, and spatial representations of supply chains and other networks.

Developed as part of the Transit-Labour research network, a collaborative project investigating changing patterns of labor and mobility in Asia, the Port Botany Visualisation software application is a representation of “wharf activity and truck turnover times at Port Botany, Australia” (Neilson, Ang, and Rossiter; also see Figure 4.1).2 Informed by a domain knowledge understanding of the “underlying dynamics of system operations” at ports, its goal is to study changes in port productivity contextualized in terms of their connection to “economic events, labour relations and technological change.” Indeed, upon launching the application, we are told that while “ports are logistical machines for creating productivity,” this productivity “rests on events and disturbances that elude integration” (Port Botany Visualisation; Neilson et al.). Users are introduced to a graphical table separated into land-side (truck) and wharf-side (ship) transportation. The data—from government reports and interviews with port authorities and cargo companies—is grouped by year and quarter. It is also partial and incomplete. In some respects, such abridgment is typical. Not only is logistics data—which might include both broad trade statistics and highly specific operation records—affected by changes in standards, it is often subject to varying levels of accessibility depending on the administrative authority overseeing it.

The display for each quarter provides both numeric data and graphic representations, with most of these representations showing observed data against a maximum value. The wharf-side data, for example, includes gauges for information like crane rate and the variability of the stevedores (dockworkers, longshoremen) responsible for unloading cargo. The display also includes graphics of stacked containers that visualize data such as throughput or the vessel’s working rate. The land-side data is similarly organized, and for years after 2008, it even shows a detailed breakdown of daily trucking slots. While the project includes documentation, the measures that it treats are quite technical. One would need to know, for example, that “ship rate” compares crane rate and crane intensity—the number of allocated hours divided by the labor time of workers from ship boarding to departure, minus delays. Still, there is value just in seeing a holistic comparison of productivity across quarters. Not only does the interface make visible the ebb and flow of cargo over time, but the side panel with quarterly news and information also provides connections that at least begin to integrate qualitative accounts of the surrounding economic, political, and logistical circumstances. Indeed, Ned Rossiter notes that while “the aesthetic logic of the visualization is not markedly different from the many visualizations developed in digital humanities,” its deployment here highlights the “multiplicity of logistical forces exerted upon labour.” As a substitute media form “that [makes] visible the pressures on labour within the shipping and transport industries,” it effects its own “algorithmic architecture,” which is entirely “distinct from those available through proprietary licenses” (Rossiter, “Materialities,” 222, 224, 233). The Port Botany Visualisation is not an application that coordinates supply chains, but “an analytical medium through which to register the interrelations between logistical infrastructure, algorithmic rules, labor practices, and supply chain assemblages” (Rossiter, Software, 69).

As an emblem for a class of digital—though not necessarily digital humanities—tools for which the primary purpose is indeed to make visible logistics data, the Port Botany Visualisation is distinguished by its humanistic interest in labor, infrastructure, and qualitative social context. But like all such tools, it is also characterized by its focused (but limited) quantitative data, a lack of actionable end points (these tools disassemble, rather than assemble, logistical operations), and a constrained range of visualization techniques. While geographic visualization may be a function of these tools, it is rarely the primary one. For example, Trase—a project by the Stockholm Environmental Institute and Global Canopy—provides a supply chain visualization allowing users to examine measures of deforestation, land use, and emissions for a select number of commodities and countries.3 While its historic data varies, and while Trase does include a map of producing regions, geographic visualization in the project is secondary to providing graphs of production over time, importing countries, ports, and so on. Perhaps Trase’s most valuable feature is a section connecting particular interest views to data for potentially related measures, commodities, and countries.

Trase, along with comparable projects like Visualizing Supply Chains for Eco-conscious Redesign (ViSER), primarily engage with logistics data from the perspectives of economics, industrial design, and environmental science (Bernstein et al.). With conflicting standards for representing supply chains, projects often adopt relatively flat and inflexible data models—though a few, like ViSER, employ relational data formats that can be used to generate graphlike visualizations of connections. ViSER goes so far as to present itself as a general-purpose dashboard with a variety of visualization panels and support for arbitrary datasets.4 But while this dashboard interface may be a common metaphor used in industry software, it is critical to recognize that this is not industry software. Even smaller industry packages and providers like Panopticon, InetSoft, and Log-hub, though lacking the operational capabilities of platforms like SAP, are not bound by the constraints of the substitutes offered by academic projects like ViSER or Trase.5 Panopticon, for example, has the potential to connect to live databases and subscribe to logistical event streaming feeds from systems like Apache’s Kafka—a functional impracticality for the scale of a DH project.

Rather than attempt to offer a substitute for industry software, some projects subvert them. Ian Cook et al.’s followthethings, for example, presents what Miriam Matthiessen and Anne Lee Steele describe as an “Amazon underbelly of sorts, designed to match the feel of online shopping but linking the visitor instead to scholarship, films, stories, reporting on a given product” (Matthiessen and Steele, 12; see also Cook).6 Other efforts have attempted directly to analyze their functions and effects. In 1996, for example, anarchist programmers affiliated with LabourNet cracked SAP’s software to identify lines of codes “that would affect workplace activity in detrimental ways,” and Ned Rossiter argues that, while less dramatic, similar “methods developed within the digital humanities . . . have an important role to play in the critique of logistical power” (Software, 55). Still, Miriam Posner is quick to note that, in many respects, “the term ‘software’ is misleading” when applied to industry systems. These are not singular applications, but complex “suite[s] of tools joined together through a shared database” with a modular nature that means that many “functions are ‘black-boxed.’” Posner writes, “In a company that makes use of an integrated SAP suite for supply chain planning, planners pass data along a chain,” moving, “with increasing granularity, down the line, until it reaches workers on the factory floor” (Posner, “Breakpoints”). This close reading nevertheless allows her to demonstrate how SAP’s specialization allows operations on different time scales (“time horizons”) simultaneously, and to show that, despite appearances, features like the “Supply-Chain Cockpit”—a graphic representation of the geography of factory locations, shipping routes, and warehouses—are “not dynamic,” but “a moment of the planning process, frozen in time” (Posner, “The Software”).

There are also those projects that attempt to provide access to the logistical data produced by corporations, nonprofit and activist organizations, and administrative government records by developing end points for new kinds of users to access that data.7 Although much of this data is, as Rossiter puts it, more “remodeled” than raw, there can be novel derived measures (Software, 62). Part of the work of the Helena Kennedy Centre, for example, involves cross-referencing import records and manufacturing data to determine likely pathways of forced labor (Helena Kennedy Centre); the University of Delaware’s project on the “Global Illicit Trade in Energy-Critical Materials” uses machine learning and remote sensing to map illegal extraction (Klinger et al.); The Prepared initiative’s Dong Xi project mined companies and locations from PDFs of corporate supplier reports (Wright). But despite the best efforts of projects like the Open Supply Hub, Wikirate, and others, the open-data ecosystem facilitating such study remains limited.8 While U.S. customs records, for example, are ostensibly public reports, they are buried behind administrative hurdles. Companies like Panjiva, ImportGenius, and MarineTraffic, which offer end points to logistics datasets, have made access increasingly restrictive or expensive over time, and projects like Importyeti and PortExaminer, which currently provide free queries for this kind of data but without well-defined application programming interfaces or direct access, have uncertain futures.9 But in all the projects creating substitutions or simulations of logistics information, there remains a disconnect—an abstraction—between the screen and the real people and places that comprise the landscape of logistics.

Spaces of Possibility

“Logistics space,” Deborah Cowen writes, is the network cartography defined by the global circuits of circulations, the deliberate result of “thinking and calculating space anew.” But “there is a dearth of scholarship on the representations of [this] space” (Cowen, “The Deadly Life,” 33, 48). Cartographic representations require a different kind of manifest, one that anchors abstract flows to material ones. As Johanna Drucker et al. note, maps require reconciling what might exist only in relative scale and space by organizing (or reorganizing) actors, events, and relationships: “forcing” this geographical representation, then, can represent a “profound, even violent, intervention” (Drucker et al., 57). But indeed, as I have already suggested, to map the supply chain is to respond to the violence of listing already embedded in its origins—to ground the network of the supply chain in those very geographies that it has already abstracted and de-territorialized. This may be why maps are rarely central to industry applications. Perhaps they are inherently “counter-logistical” pursuits (Bernes).

Supply-chain mapping projects employ a mix of commercial platforms and open-source software, including geographic information systems such as ArcGIS and QGIS, web mapping platforms such as Carto, Mapbox, and Leaflet, and spatial databases such as OpenStreetMap. Such projects tend toward two distinct areas of intervention: narrating accounts of commodities or representing landscapes of logistical infrastructures. The former is common to projects created by journalists and activists, whose stories gather insights by revealing the geographies traversed by commodities. Examples of academic projects in this area include mixed media initiatives such as Commodity Histories or the Charnley-Persky House Archaeological Project, which connects past consumer products to their wider logistical networks—listing, mapping, and historicizing the use and supply of household goods to “give voice to those who are not typically in the documentary record” (Charnley-Persky House Archaeological Project, “Archaeology”).10 But the direct connection to the manufacturer on the label of commodities represents only a partial perspective, and researchers cannot follow everything. Even in activist projects, high-profile companies like Apple and Nike must stand in for problems endemic to entire sectors, with articulations of those problems varying widely across numerous firms with sometimes divergent operational strategies. Often absent from these accounts are the ports, warehouses, and other logistical intermediaries, along with the suppliers and partners that may have contributed to their material inhabitance at locations such as Charnley House in Chicago. As a result, also absent are the pathways for the flows of the materials themselves.

The latter sort of intervention can be seen in projects such as the Empire Logistics research initiative, which addresses the representation of absent infrastructure directly, making the scale of network space commensurate to its real geographies to “articulate the infrastructure and [its] ‘externalized cost’” (Empire Logistics, “About”).11 Focused on warehouses and logistical infrastructure in the Inland Empire in southern California, the project’s “Supply Chain Infrastructure” map (Figure 4.2) includes information on railways, ports, and warehouses, identifying specific distribution centers for companies such as Amazon, Costco, and Walmart, and thus showing the broader context of technology, logistics, and finance.

Epitomized here is the power of listing: the logistical network of the Inland Empire gathered together and placed in a public geography of roads, airports, neighborhoods, and cities that makes manifest the spaces “where proletarian solidarity has the greatest possibility to spread up and down the chain” (“About”). But as a “counter-logistical” project (Bernes), Empire Logistics also uses its maps to open the choke points of the networks, thus making possible new—and given the sensitive nature of these choke points, its own potentially violent—forms of resistance. There are other examples of projects addressed to questions of logistical power, each with their own ethical or political stakes: the Abandoned Seafarer Map, which documents workers whom corporations have left behind at sea (Ader, Bolton, and Matthiessen); the Reducing Opportunities for Unlawful Transport of Endangered Species (ROUTES) Dashboard, which records wildlife trafficked through airports; and the Feral Trade project, which records transactions of a collective freight network, albeit one organized outside conventional logistical channels.12 Other investigations reveal unfamiliar logistical infrastructures to draw attention to their geopolitical complications, material fragility, and complex entanglements. This includes Nicole Starosielski, Erik Loyer, and Shane Brennan’s Surfacing project, which documents undersea fiber-optic communication cables, as well as work like Kate Crawford and Vladan Joler’s Anatomy of an AI System and Anna Tsing, Jennifer Deger, Alder Keleman Saxena, and Feifei Zhou’s Feral Atlas.13

Projects looking at earlier historical periods face different challenges. Not only does the reconstruction of historic trade networks depend on the availability of archival materials, but these networks exist within political landscapes and physical geographies that may be largely unrecognizable. Projects like the Viabundus Pre-modern Street Map, for example, attempt to make historical trade networks legible to contemporary viewers.14 Digitizing the routes recorded in Friedrich Bruns and Hugo Weczerka’s 1962 Hansische Handelsstraßen, the street map cleverly juxtaposes a modern interface for trip planning with directions from centuries past to make this trade network accessible in ways that a purely textual account cannot. Other projects, like Old Weather, recover maritime pathways from ships’ logs to enable new “perspectives on the movement patterns of early American capitalism” (Schmidt); and Trading Consequences reconstructs networks of trade, economy, and environment in the British Empire during the nineteenth century by mining over 200,000 documents for references to commodities or locations.15 The visualizations that resulted from this latter trade “search engine” allowed one to “correlate information extracted for one commodity with that of others,” and thus to analyze the flows of materials and configurations of networks across decades (Trading Consequences). But one must be cautious about such historical representations. As Ben Schmidt reflects, visualizations of historical data do not remove us from problem of grappling with the historical context, the underlying archive, and the constraints under which information about the past was collected. While such visualizations can be valuable tools, one must be cautious about reading current concerns into them. There were no supply chains for the Hanseatic League or for whalers in the nineteenth century, at least as we know them.

There are, of course, digital projects in other domains that intersect with the critical study of logistics. “Although they diverge in their object of study,” Starosielski writes, projects in ecological digital media such as Growing Blue, Aqueduct, Bodily Natures, and Protest are similarly “invested in digital technologies” and offer their own logistical insights (402). For more artistic inventions, we might consider projects like Georgina Voss and Wesley Goatley’s Familiars, an immersive installation embedding the viewer in a mapped representation of intercepted logistical signals broadcast via radio (Goatley); the interactive game Cargonauts; or Jesse LeCavalier’s architectural exploration Landscapes of Fulfillment.16 And there are countless other such projects to which we might look for inspiration.

Manifesto

This final section discusses the Manifest project (Figure 4.3), outlining how its technical and ethical framework has been informed by the earlier projects that I have described here, before concluding with future challenges for DH in the critical study of logistics.17 Manifest is based on some of the concepts behind Sourcemap, the web platform that I created to integrate social and environmental impact assessment (carbon footprinting, water, and energy usage) with geospatial visualizations of supply chains.18 While originally intended as a tool of journalistic or activist investigation and scholarly research, Sourcemap was ultimately commercialized in 2011 as an industry platform by other members of the project team. Prior to this, the project published over ten thousand user-submitted supply-chain documents.

Released in 2021, Manifest is “an investigative toolkit intended for researchers, journalists, students, and scholars interested in visualizing, analyzing, and documenting supply chains, production lines, and trade networks” (“About Manifest”). Its goals are to develop a common data format for work in the critical study of logistics that can incorporate information relevant to a range of research projects and to provide an extensible visualization framework for studying supply chains. It has support for both spatial and nonspatial representations, as well as an interface that allows for the aggregation of multiple concurrent datasets, such as multiple discrete supply chain documents, infrastructural data layers like port locations or railways, and live data sources such as container ships or trucks. Figure 4.3, for example, shows a series of screenshots from Manifest that illustrate both map and graph visualizations of several supply chains overlaid on multiple infrastructural data layers and live data sources. The project is supported by a Digital Humanities Advancement Grant from the National Endowment for the Humanities (NEH), which provides funding for both overall technical development and the publication of a series of Manifest “case studies” produced by external contributors (Hockenberry and Perold); for an example of a Manifest case study, see Schwartz et al.19

From a technical perspective, Manifest documents in simple JavaScript Object Notation (JSON) files represent trade networks, supply chains, and similar accounts; and the tool allows users to construct, visualize, and analyze multiple such documents concurrently. Each document contains a list of nodes, including information like a name, an address, coordinates, connections to other nodes, and user-defined text, categorizations, imagery, and impact measures, along with arbitrary key-value pairs (which could include citation information, data generated by other applications, or other links and references not directly useful to Manifest). For example, the Western Electric supply chain shown in Figure 4.3 includes an entry for “Iron,” with an address listed as “Hibbing, MN, USA,” coordinates of 47.4271546°N latitude and -92.937688700002°W longitude, a connection to Western Electric’s factory in Hawthorne, Illinois, a text entry noting that “the Hull-Rust-Mahoning Mine in Hibbing, Minnesota is one of the largest open pit iron mines in the world,” categorization as a “raw material,” and a measure that provides the approximate tonnage of iron extracted. In any visualization, these nodes can be filtered by category or keyword, or scaled to show the relative difference of the quantitative measures. The map visualization is customizable: colors, sizes, and icons can be set, and a variety of base tiles—the series of composited images that comprise the visual appearance of the map—like satellite imagery can be selected and layered with user-selectable supplemental datasets such as the shipping lanes, rail networks, and maritime traffic data previously described. Manifest provides extensible plug-in support for other visualizations as well, currently including a spreadsheet view, a chart view, the graph view shown in Figure 4.3, and flow and chord diagrams. It also has capabilities for sharing documents and generating static images of visualizations. While a public installation is available on the Manifest home page, anyone can set up an instance from the project’s GitHub repository.20

Manifest diverges from its predecessor, Sourcemap, in several critical ways. The map is still present. But although Manifest supports visualizing quantitative data, the calculator is less so. This is partly because while approximate information proved to be meaningful in geographic visualization, approximate calculations did not. This is one of the reasons why Sourcemap was reproduced as a tool for business rather than maintained as the messy means of journalistic and activist “muckraking” it was originally intended to be. As the projects detailed in this chapter demonstrate, logistics data is often partial and incomplete, with corporate actors the most likely ones to have detailed quantitative data (or the power to compel its collection).

Manifest is also less invested than Sourcemap in centralizing supply chain data or providing comprehensive accounts of supply chains. From a practical perspective, this is due to the constrained nature of most supply chain investigations. Indeed, one of the greatest obstacles in the classroom is a concern about limited data. Manifest was not designed to be a clearinghouse for complete accounts of logistical operation, but rather an interface for assembling multiple supply chains (or partial supply chains) to understand the relationships among them. In Manifest, a map of three nodes can be as meaningful as one with three hundred, since that map can be brought into alignment with an arbitrary number of other Manifest documents and data layers. This decentralization also speaks to the finite lifespan of digital projects in general (Meneses). Some of the projects that I have surveyed here, like Port Botany and ViSER, continue to be usable but are no longer under development; others, such as Trading Consequences and Empire Logistics, are all but inoperable; while still others, like ImportYeti, seem destined for increased commercialization. Even more fundamental than the ephemerality of digital projects, of course, is the fact that the nature of global supply itself is ever changing. Projects not intended to be historical work will soon become so. Port Botany, for example, makes visible the labor of workers who are no longer employed. Empire Logistics shows infrastructure that has come and gone. And while some projects leave their data accessible (Port Botany’s, for example, is available through the Australian Research Data Commons) or are committed to making regular updates (as in the case of the Abandoned Seafarer Map), many do not allow for sustainability. Fewer still share their code.

In Manifest, the data format is divorced from the interface. The data format not only privileges simple textual descriptions and clearly calculated measures, it supports arbitrary data that Manifest neither interprets nor visualizes. This leaves open the possibility that data produced for use in Manifest can be meaningful for other tools past or present and can be imported into general data science systems or even read in a basic text editor. Reliant on few external services, Manifest documents can be edited in a basic interface that requires users to geocode locations manually. Even the live map tiles can be exchanged for a static world map image.

But issues of functionality aside, the ethical framework that Manifest sets forth is important because of its endeavor to acknowledge the power of listing, working against a landscape of industry software best suited to those who have the ability not just to take, but to make, inventory. In this regard, Manifest does not just act counter to the prevailing logistical order but also as an inquiry into the fundamental structure that such order necessitates. Rejecting structures that seem to be collected and complete in favor of the distributed, the partial, and the temporary, Manifest advocates the creation of a scaffolding for collecting fragments of logistical networks. It is not a database, nor does it aspire to control or own its data—in part because this is exactly what has allowed for the abstraction of logistics space by enterprise software systems. The supply chain, after all, is a structure that “ravenously consumes other data” (Posner, “The Software”). Indeed, the very politics of supply chain transparency that Sourcemap once pioneered has become a central component of corporate social responsibility.

In the epigraph that starts this chapter, the manifest is described as an accounting of injuries. Despite our meager sympathies for Twain’s protagonist, I think that this perspective provides a better model for unraveling the global supply chain than transparency. Rather than allow transparency to remain as a form of corporate responsibility, with “mapping the supply chain” an exercise in corporate power, “making out its manifest” might now be the ideal that accounts for our value, as well as our injuries. Such an accounting records the places where labor has been exploited, the Earth has been plundered, waste overruns into rivers, and poison bleeds into the air. It is not a proclamation from on high, but rather an admonition from below—not an attempt at supply chain resilience but an opportunity for supply chain reconciliation. But this is not a singular endeavor. The manifest is a social document, and its listing is empowered by the shared understanding of those who read it. The future of DH for the critical study of logistics requires precisely such common cause. We must foster a community of open data and open-data standards, contribute and extend open-source frameworks for common visualization tasks, and imagine both the present moment and a collaborative afterlife for digital projects that are capable of contending with the logistical software systems of supply chain capitalism.

Notes

1. 1. See https://sourcemap.com/.

2. 2. For the dataset of the Port Botany Visualisation, see Neilson, Ang, and Rossiter. For a discussion of the project, see Nguyen, Zhang, and Simoff.

3. 3. See https://www.trase.earth/.

4. 4. This brings with it, of course, all the limitations of the dashboard. As Shannon Mattern writes: “The acknowledged partiality of the dashboard’s rendering might make us wonder what is bracketed out. Why, all the mud of course! All the dirty (un-‘cleaned’) data, the variables that have nothing to do with key performance (however it’s defined), the parts that don’t lend themselves to quantification and visualization.”

5. 5. See Panopticon, https://altair.com/panopticon; InetSoft, https://www.inetsoft.com; Log-hub, https://log-hub.com.

6. 6. See http://followthethings.com.

7. 7. End points are server locations, usually Universal Resource Locators (URLs), where an application programming interface (API) receives requests about a specific resource on its server. These are usually used to provide programmatic access to databases by external developers or internal developers working on a different software application.

8. 8. See Open Supply Hub, https://opensupplyhub.org/; Wikirate, https://wikirate.org/.

9. 9. See Panjiva Supply Chain Intelligence, https://panjiva.com/; ImportGenius, https://www.importgenius.com/; MarineTraffic, https://www.marinetraffic.com/en/ais/home/; ImportYeti, https://www.importyeti.com; Port Examiner, https://portexaminer.com/.

10. 10. See https://digitalchicagohistory.org/exhibits/show/charnley-persky-house/archaeology-at-the-charnley-pe.

11. 11. See https://www.empirelogistics.org/about/.

12. 12. See Abandoned Seafarer Map, https://abandonedseafarermap.cargo.site; ROUTES Dashboard, http://www.routesdashboard.org/; Feral Trade, https://feraltrade.org/.

13. 13. See Surfacing, https://surfacing.in/; Anatomy of an AI System, https://anatomyof.ai/; Feral Atlas, https://feralatlas.org/.

14. 14. See http://www.landesgeschichte.uni-goettingen.de/handelsstrassen/map.php.

15. 15. See Old Weather, https://www.oldweather.org/; Trading Consequences, https://web.archive.org/web/20220929115427/https://tradingconsequences.blogs.edina.ac.uk/. On the Trading Consequences project, see also Hinrichs et al.

16. 16. See Familiars, https://www.wesleygoatley.com/familiars/; Cargonauts, https://cargonauts.net/; Landscapes of Fulfillment, https://landscapes-of-fulfillment.org/.

17. 17. See https://manifest.supplystudies.com/.

18. 18. See https://sourcemap.com/. On Sourcemap, see also Bonanni et al.

19. 19. See https://manifest.supplystudies.com/data/.

20. 20. See https://manifest.supplystudies.com/; https://github.com/hock/Manifest.

Bibliography