NAVIGATIONEN L IE F E R N . L O G IST IK E N , D A T E N U N D P O L IT IK E N AUTOMATED DELIVERY: AMAZON’S URBAN STACK A R M I N B E V E R U N G E N AMAZON’S URBAN STACK The city today is filled to the brim with logistics and platform companies vying for the chance to deliver goods to us. For example, the rise of dark retail and ghost kitchens during the COVID-19 pandemic has meant that as restaurants and shops closed shut, their operations were relocated from the high street to basements, industrial estates, and other dark sites not welcoming to customers. This »compen- satory convenience«, as Aaron Shapiro has called it,1 has exacerbated a trend in which delivery has in many ways displaced retail.2 Delivery promises a particular kind of convenience: where even a visit to the local convenience store for conven- ience food is a hurdle, companies like UberEats offer fast meal delivery and others such as Getyr promise delivery of everyday basics within a 10-minute timeframe. Amazon is inscribed in this broader landscape of logistical convenience oriented around delivery, and stands out both for its push towards speed in offering next- day or same-day delivery, for the breadth of goods on offer in its everything store, and for the spread and coverage of its logistical operations.3 The expansion of Amazon’s last-mile delivery network means that it is only matched by the infrastructures of national postal services, making it minimally reli- ant on these or other logistics companies such as DHL or UPS.4 In Germany, for example, since 2017 it has set up more than 70 delivery stations, where parcels are sorted and prepared for last mile delivery.5 In the USA, there are hundreds of de- livery stations. These are serviced by fulfillment centers and sort centers – the larger elements in the topology of Amazon’s logistical infrastructure – and they pro- vide the entry point to the last mile. These delivery stations move further into urban areas, although they are not located, like dark stores, directly in densely populated areas but often found in established industrial areas. From there, mostly drivers in vans take over the delivery of parcels to customers’ doorsteps. It is from here on 1 Shapiro: »Platform Urbanism in a Pandemic«. 2 Rodrigue: »The Distribution Network of Amazon and the Footprint of Freight Digitaliza- tion«. 3 Beverungen: »Collectivizing Convenience? From Delivery to Logisticality«. 4 Nonetheless, especially at the perimeters of its networks, Amazon fundamentally relies on postal services to expand its reach. See e.g. O’Donovan/Bogage: »A Rural Post Office Was Told to Prioritize Amazon Packages. Chaos Ensued«. 5 As here, the contribution broadly draws on the extensive research findings conducted as part of the »Automating the Logistical City« project. See https://logistical.city/. ARMIN BEVERUNGEN NAVIGATIONEN 48 L IE F E R N . L O G IS T IK E N , D A T E N U N D P O L IT IK E N that further elements are enrolled in Amazon’s logistical operations – constituting an urban stack of last mile delivery which is increasingly automating the logistical city. Speaking of the urban stack is one way to account for how different elements of material, technological, and human infrastructures come together in last mile delivery. The term stack is widely used in computing to refer to hardware and soft- ware assemblages that are ordered through different layers, and it has been em- ployed in media theory to give an account of planetary networked infrastructures and to raise questions of sovereignty in relation to it.6 Focusing on the city, Shannon Mattern has used the term urban stack to describe the computing infrastructure of smart cities, from urban dashboards to network protocols. Mattern, in discussing interfaces such as urban dashboards, raises the following questions: »What is the ›city‹ they propose to put us in relation with, and how deep into the stack does that relation go?«7 An equivalent question is addressed here in regard to Amazon’s ur- ban stack for last mile delivery: What kind of city does it produce and put us in relation with, and how deep are we drawn into that relation? Shapiro develops the notion of the urban stack further, inviting us to inquire beyond interfaces into the assemblages of materialities, technologies, and humans that make up the urban stack. In his two case studies exploring data infrastructures – public wifi-infrastructures and platform labor apps – he notes how value produc- tion »hinges on urban technology producers’ ability to enroll heterogeneous ele- ments into a hierarchical flow of information.«8 In the following I will also draw on this notion of enrolment, which Shapiro takes from John Law and Annemarie Mol,9 and focus on flows of information. I understand Amazon’s urban stack to be con- stituted by diverse logistical media10 variably enrolled in the logistical operations of Amazon’s urban stack. While these elements are several, I will analyze three in particular: the door/porch, the locker, and the van. Exploring these as elements of Amazon’s urban stack for last mile delivery shows how existing elements of urban infrastructure are enrolled in Amazon’s logistical operations. For each of these three elements, I focus first on how they are enrolled in Amazon’s urban stack and how this enrollment implies a reconfiguration of urban space and topology and an integration into the standards and protocols of Amazon’s logistical operations. Furthermore, I explore how automation recruits both last mile delivery labor as much as consumers in its logics of automation, which extend be- yond machinic automation to include the (attempted) automation of human behav- ior through scripts and habits. Finally, I demonstrate how automation involves data 6 Bratton: The Stack. 7 Mattern: »Interfacing Urban Intelligence«, n.p. 8 Shapiro: »The Urban Stack«, p. 62. 9 Law/Mol: »Situating Technoscience«. 10 On the notion of logistical media, see Peters: »Calendar, Clock, Tower«; Hocken- berry/Starosielski/Zieger: Assembly Codes. AUTOMATED DELIVERY NAVIGATIONEN L IE F E R N . L O G IST IK E N , D A T E N U N D P O L IT IK E N 49 extraction and machine learning, which turns the city into a site of continuous test- ing and speculation. The contribution concludes with a consideration of the limits of automation in Amazon’s urban stack for last mile delivery. THE DOOR/PORCH Bernhard Siegert reflects on the door as architectural media: »Doors are architec- tural media as an elementary cultural technique because they process the guiding difference of architecture, the difference between inside and outside.«11 The oper- ation of opening and closing is central to the door as a cultural technique. Doors can also be considered spatial and logistical media, since they mediate spaces and flows: as »operators of symbolic, epistemic, and social processes ›they‹ generate spheres of law, secrecy, and privacy and thereby articulate space in such a way that it becomes a carrier of cultural codes.«12 Bruno Latour famously took up the door of his office building as a prime example of the missing masses in sociology and described how the door in its assembly with hinges and grooms operates certain scripts and serves as a delegate regulating entry to a building. He also understood the door as constituting order and as an information processor: »The hinged door allows a selection of what gets in and what gets out so as to locally increase order, or information.«13 In somewhat of a side-note, Latour also observes how the programmed scripts of the door can discriminate: »They discriminate against furniture removers and in general everyone with packages, which usually means, in our late capitalist society, working- or lower-middle-class employees.«14 Even though Latour refers here spe- cifically to the challenge of opening a door with one’s hands full of packages, this discrimination against everyone with packages in the liminal space of the door in- dexes the challenge at the center of Amazon’s efforts to recode the door. After all, it is the front door of their homes where the brown packages with the smile are deposited and where convenience is delivered.15 Delivering the goods through the front door, in particular when the customer is not at home, constitutes the final challenge of last mile delivery. The assemblage of the front door includes elements in addition to the hinges and the locks Latour describes: they also usually include a doorbell and a letterbox, and in front of them lies the porch as a further threshold in need of recoding. First of all, however, in order to know where to deliver its packages, Amazon must deal with the computational problem of addressability, in this case the need 11 Siegert: »Doors«, p. 8. 12 Siegert: »Doors«, p. 9. 13 Latour: »Where Are the Missing Masses?«, p. 228. 14 Latour: »Where Are the Missing Masses?«, p. 234, emphasis added. 15 West: Buy Now. ARMIN BEVERUNGEN NAVIGATIONEN 50 L IE F E R N . L O G IS T IK E N , D A T E N U N D P O L IT IK E N to corroborate the correspondence between geolocation and delivery address.16 To achieve this, Amazon enrolls an urban stack emerging from a history of address, postal services, and logistics, and therefore a long history of the mediation of cit- ies.17 In addition, Amazon has established its own location services, which are also available through Amazon Web Services. Here the key element in terms of address- ability is the unique place identifier (»PlaceID«), which is assigned to each location and tied to customer accounts and myriad other data, e.g., whether a dog awaits the delivery driver. In this way they are more precise and also richer than street addresses. However, for the optimization of Amazon’s routing, the driver needs to know exactly where to find the front door, a piece of information not provided by the delivery address. Towards this end, Amazon uses GPS to track when drivers acknowledge delivery, and it uses this GPS data to calculate the exact location of the front door. In a recent Amazon Science blog post, a principal applied scientist recounts the challenges of fixating the coordinates of the door.18 Where existing algorithmic approaches make the mistake of assuming that the door must be lo- cated somewhere in the middle of the many recorded GPS coordinates, the newly implemented machine learning technique (»learning-to-rank«) focuses on identify- ing the edge of GPS coordinates where the door is most likely located. In this way, as the door becomes addressable, it becomes enrolled in Amazon’s urban stack of last mile delivery in an operation of ground truthing. However, the door remains a stumbling block, which – even when found – too often remains shut. Particularly during the COVID-19 pandemic, when human contact was largely to be avoided, customers would keep their doors closed to delivery drivers. Not only because of this incentive, Amazon developed a workaround to a closed door: depositing parcels on the porch. Social distancing requires that customers no longer need to personally receive the parcel; this is achieved by enrolling the porch in Amazon’s urban stack and reconfiguring it as a site of temporary storage. However, this requires another way of recording delivery, since the parcel is not handed di- rectly to the customer. This is solved through further mediation. As Matthew Hockenberry details, confirmation occurs through operational images, which make human contact redundant. The delivery drivers simply take a picture of a parcel on the porch: »As the system does not intend for the consumer to see the worker, the worker need not see the consumer. It is the camera – the system – that sees.«19 Thus, the door is sidestepped and the porch becomes a site of temporary storage, complemented by the driver’s smartphone and the Amazon Flex app, which docu- ment the delivery with a picture. This operational image potentially circulates 16 Dhaliwal: »On Addressability, or What Even Is Computation?« 17 Mattern: Code + Clay... Data + Dirt. Kittler: »The City Is a Medium«. 18 Forman: »Using Learning-to-Rank to Precisely Locate Where to Deliver Packages«. Many thanks to Ulf Treger for pointing me to this case and for discussing it within our research group. 19 Hockenberry: »Cellular Capitalism«, p. 273. AUTOMATED DELIVERY NAVIGATIONEN L IE F E R N . L O G IST IK E N , D A T E N U N D P O L IT IK E N 51 beyond its immediate context, becoming part of operational databases enabling a much wider »platform seeing.«20 In addition, Amazon offers another way to engage and recode the door: its Ring devices. Ring doorbells, cameras and Amazon Keys can be connected to front doors and give Amazon access, so that drivers can open doors and deposit parcels safely inside homes. The door in this way becomes differently enrolled in Amazon’s urban stack: it can now be controlled, its operation of opening and closing auto- mated. At the same time this recoding of the door through Amazon devices repro- grams the script of the door – who has a key, who is allowed in and when, etc. – as well as the threshold of the door as a boundary between public and private space. Furthermore, as Lauren Bridges has noted, since Amazon cooperates with police, at least in the USA, by sharing video footage for policing, the Ring devices enroll the door, through what Bridges calls »infrastructural obfuscation«, in carceral regimes – so not just in Amazon’s urban stack, but a wider urban stack also serving policing and incarceration.21 THE LOCKER The front door, with the development of postal services, would soon also often incorporate a mail slot or be supplemented by a mailbox to receive letters.22 Yet these rather thin slots, which provide an opening in the door or into the compart- ment safely storing the delivery, are made for letters or postcards, hardly for par- cels. While Amazon provides some envelopes which do fit into these slots, most do not, considering that many of the sortables that Amazon delivers through its logistical network require larger packaging, even if they can still be carried by hand. Between the house (and its door) and the container, which has received much at- tention in media and logistics studies,23 the letterbox can be considered another form of container or housing.24 Since for last mile delivery the typical letterbox doesn’t work for Amazon, we can consider the locker a kind of displaced letterbox: although some of the lockers are placed within residential blocks, most are situated conveniently in locations constituting nodes in existing logistical flows. And they serve not one addressee but many through an address that becomes accessible to all Amazon customers in that area. The locker is Amazon’s own addition to its urban stack. Amazon introduced its lockers in 2011 as an alternative endpoint for its last mile delivery. Since then, it has vastly expanded its network both in the USA and globally. While in the USA it 20 MacKenzie/Munster: »Platform Seeing«. See also Beverungen: »The Invisualities of Cap- ture in Amazon’s Logistical Operations«. 21 Bridges: »Infrastructural Obfuscation«. 22 Siegert: Relays, p. 109-119. 23 Levinson: The Box; Klose: The Container Principle. 24 On the letterbox as container, see Fingerhut: »Übertragen und Speichern«. ARMIN BEVERUNGEN NAVIGATIONEN 52 L IE F E R N . L O G IS T IK E N , D A T E N U N D P O L IT IK E N provides the largest network of lockers,25 elsewhere other logistics and postal ser- vices companies provide alternative networks. In Germany, for example, there were more than 13,000 so-called Packstationen operated by DHL in 2023.26 This is in contrast to 1000 Amazon Lockers at the end of 2019,27 although that number would be much higher today. Lockers are part of a wider shift in last mile delivery, discussed in the transport logistics literature in terms of alternative delivery loca- tions, which also include parcel pick-up in locations such as postal offices or click- and-collect locations in retail.28 Lockers are supposed to offer an alternative con- venience to delivery to the front door: there is more security, as parcels are locked away safely instead of lying on the porch for anyone to take away, and the lockers usually provide a 24/7 availability for self-service pick-up. Image 1: Amazon Locker »pya«, located opposite the Amazon Sort Center in Witten (pic- ture by author, taken on April 20, 2023) 25 Schaefer/Figliozzi: »Spatial Accessibility and Equity Analysis of Amazon Parcel Lockers Fa- cilities«, p. 1. 26 https://www.dhl.de/de/privatkunden/pakete-empfangen/an-einem-abholort- empfangen/packstation.html, 24.03.2024. 27 https://www.aboutamazon.de/news/logistik-und-zustellung/1-000-locker-bieten-ama- zon-kunden-in-deutschland-eine-flexible-paketabholung, 24.03.2024. 28 Kim/Wang: »The adoption of alternative delivery locations in New York City«. AUTOMATED DELIVERY NAVIGATIONEN L IE F E R N . L O G IST IK E N , D A T E N U N D P O L IT IK E N 53 Amazon keeps the criteria determining the selection of locations for lockers secret. However, since it solicits applications for hosts of lockers, asking for upfront infor- mation, some of the criteria for selection become visible, such as: the general avail- ability of space for particular sizes of lockers, accessibility to the public, and existing customer traffic – not to mention geolocation. Amazon presumably draws on a plethora of existing data-driven spatial and algorithmic techniques to determine lo- cations. Through what can be read as reverse engineering of Amazon’s algorithm for locating lockers, literature in transport geography has sought to map the distri- bution of Amazon lockers and to understand their distribution. This highlights fur- ther criteria for site selection, such as: population density, income distribution, and actual delivery data.29 The analysis also points to some of the inequities produced in terms of accessibility once these algorithmic logics of devising network topolo- gies are deployed. As Randi Heinrichs has argued in relation to Amazon Prime, this use of what she calls »data neighborhoods« potentially reproduces and exacerbates existing forms of segregation.30 Amazon, through its lockers, enrolls existing socialities and spatialities, for ex- ample when its Amazon Hub Apartment Locker program invites large residential sites to place lockers in their shared spaces, thus recoding these spaces as tempo- rary sites of storage exclusive to (relations with) Amazon. It also enrolls existing infrastructures of mobility and convenience, for example when it enters coopera- tion agreements with petrol stations such as Shell or retail chains such as 7-Eleven to host its lockers. The network of lockers thus comes to mirror those of conven- ience stores and petrol stations, and seeks to draw on the mobilities and relations already established through these. Amazon certainly isn’t the only provider of lock- ers, as mentioned above, and apart from operators such as logistics companies and postal services, there are also many manufacturers, such as Bloq.it, Cleveron, Keva, HiveBox or Variocube. While the lockers of these different companies share similar functionality, offering modularity in terms of overall size and individual box size and providing access either through touch screens or through a bluetooth-connected app, Amazon’s lockers stand out in specific ways. Amazon’s lockers can only be used for delivery and returns from and to Amazon (and its Delivery by Amazon partners), and box sizes are aligned with Amazon packaging standards. Perhaps most importantly, however, Amazon’s lockers integrate fully into Amazon’s logis- tical operations, not just its urban stack for last mile delivery. Drivers connect to the locker via their Flex apps, and consumers do so via their Amazon apps, there- fore providing a continuous flow of information which feeds both the optimization of last mile delivery and the predictive analytics which feed Amazon’s recom- mender system.31 29 See in particular Schaefer/Figliozzi: »Spatial Accessibility and Equity Analysis of Amazon Parcel Lockers Facilities«; Fang/Giuliano/Wu: The Spatial Dynamics of Amazon Lockers in Los Angeles County. 30 Heinrichs: »Where Do the Data Live?«. 31 Pöchhacker/Nyckel: »Logistics of Probability«. ARMIN BEVERUNGEN NAVIGATIONEN 54 L IE F E R N . L O G IS T IK E N , D A T E N U N D P O L IT IK E N Lockers can also be understood as a new architectural element in the logistical city, one which at first may seem very mundane, considering that through its sleek aesthetics particular to its logistical function as well as its branding it blends right into an urban environment populated by similar elements. It certainly reconfigures the boundaries of public and private space, turns spaces of encounter into spaces of logistics and of storage, and reshapes our movements through the city through minimal architectural intervention. Following an argument of Claire Lyster, who looks at developments such as Amazon Go stores and its delivery robots, lockers partake in the promise of automated retail, which in her view nowadays is »part vehicular system, part information network and part architectural interior.«32 The locker creates an architectural interior (the differently sized containers in which the packages can be deposited) as much as it partakes in vehicular systems (those of the delivery driver as much as our own) and information networks (connected through the Flex app of the drivers and the consumer’s Amazon app required for pickup). For Lyster, this »implies that urbanity is now extended beyond architec- tural form to new hybrid species of space,« with the city »composed of smart, dy- namic, responsive entities developed equally across multiple design disciplines, from information to industrial and from product to graphic design.« Where for Lys- ter this opens up »the design of the city to a range of creative stakeholders and, by extension, to an array of new spatial typologies and their corresponding effects,«33 in the case of the Amazon locker it demonstrates the tight hold Amazon keeps on elements of its urban stack as they integrate with its wider logistical operations. THE VAN The ubiquitous delivery vans roaming our cities for last mile delivery today stand at the end of a long lineage of transport vehicles for the movement of goods. This lineage can be traced back at least to the first mail coaches of the 18th century, emerging alongside the first extensive road networks for the regular, daily move- ment of people and goods prior to the railway age.34 It also passes through the first attempts by what would later become United Parcel Service to assemble what would be called a package car out of a 1913 Model T Ford,35 and subsequent at- tempts to optimize petrol-fuelled cars for last mile delivery. More recently, electric delivery vehicles such as the Deutsche Post / DHL Streetscooter seek to move this category of delivery vehicle away from petro-mobility and to further optimize its design for logistical labor and its flows.36 While Amazon around the world relies on a fleet of hundreds of thousands of delivery vans, most of these are standardized 32 Lyster: »Disciplinary Hybrids«, p. 105. 33 Ibid. 34 Bagwell: The Transport Revolution, p. 29-37. 35 Niemann: Big Brown, p. 49. 36 Kampker/Gerdes/Schuh: Think Big, Start Small. AUTOMATED DELIVERY NAVIGATIONEN L IE F E R N . L O G IST IK E N , D A T E N U N D P O L IT IK E N 55 delivery vans, usually driven by subcontractors – so-called delivery service partners – or self-employed drivers enrolled through Amazon Flex. It has, however, recently invested in e-mobility, for example establishing an infrastructure of charging sta- tions at most of its delivery stations in Germany and providing electric delivery vans to delivery service partners through rental agreements. The infrastructure of delivery stations, now frequently coupled with parking for overnight charging of electric vans, provides the inroads for a further element of Amazon’s urban stack, which is recoded to match Amazon’s logistical operations. In 2019, Amazon announced an exclusive cooperation with the car manufacturer Rivian for 100,000 electric delivery vans to be put on the road by 2030. These were subsequently developed in close collaboration with Amazon, and the first vans were put into action in Amazon’s last mile in 2022 in the USA and in 2023 in Germany. The Rivian van is equipped with a lot of technologies assisting drivers, which makes it an autonomous vehicle, following a definition by Florian Sprenger, and like other autonomous vehicles it must be understood as more than simply a mode of transport: »They are not only engine, but at the same time media, computer, inter- faces, adaptive system, data-processing machine and context-sensitive environ- mental technology, but also objects of financing and debt.«37 Amazon strikingly adapts these features in order to enroll and integrate the vans into its urban stack for last mile delivery as well as its logistical operations more widely. In terms of financing and debt, Amazon’s delivery service partners usually enter a rental agree- ment in which they rent the Rivians (previously other vans) in order to become part of the program, with partners usually also tied to a maintenance agreement, pro- ducing a lock-in in which payment for delivery also pays for van rental. In this way, Amazon enrolls subcontracted labor in its urban stack. Some of the technology that makes the Rivian an autonomous car is deployed in the same way as in other autonomous cars, principally to assist the driver, for example automatic emergency braking, adaptive cruise control, or collision warn- ings. These technologies usually imply that a driver has to pay less attention to the environment than in non-autonomous vehicles. Yet some of the Rivian’s features specifically address its character as a delivery van: for example, its 360-degree cam- era system, which complements the very large front windshield, enables safer ma- neuvering during delivery. Amazon in particular has been critiqued for the reckless- ness of its drivers in last mile delivery – with vans often double parked or blocking bike lanes and its time-pressured drivers endangering pedestrians, cyclists, and other motorists. The van’s design, in response, allows more attention to be paid to the environment. These features more broadly respond specifically to the way the van relates to a city that is generally car-friendly but not designed for large delivery vans. Amazon and Rivian offer two sizes of the vehicle (the EDV 500 and 700), with one slightly shorter and narrower, to minimally adapt to different urban environ- ments. Amazon is also experimenting with other modes of delivery with mixed 37 Sprenger: »Autonome Automobilität«, p. 18, my translation. ARMIN BEVERUNGEN NAVIGATIONEN 56 L IE F E R N . L O G IS T IK E N , D A T E N U N D P O L IT IK E N success: its program for delivery robots called Scout launched in 2019 but was dis- continued in 2022;38 its drone delivery program Prime Air is proceeding very slug- gishly;39 and delivery by bike in North America and Europe is restricted to a few cities like New York and London. The Rivian is by far the central Amazon-specific element of its urban stack, connecting delivery stations to front doors and lockers and reinforcing automobility in all its social, spatial, and environmental effects. The Rivian van’s capacity for computing is most clearly enrolled in Amazon’s logistical operations when it comes to monitoring delivery labor. Previously, Ama- zon relied on an app called Mentor, which drivers had to install on their phones, or the Driveri camera system installed in Amazon vans. Drivers were monitored in the name of safety, and penalized for not conforming to specific rules, for example when braking harshly, speeding, or not using a seatbelt.40 With the Rivian, some of that functionality is integrated into the vehicle management system, which collects large amounts of data about how the vehicle moves as well as the driver’s interac- tions with it. This means that some of the techniques of algorithmic management that Amazon uses in last mile delivery can now be justified as an integral part of using an autonomous vehicle – even in Europe, where data protection regulation may have prohibited the use of Mentor or Driveri.41 Further, the connectivity of the Rivian gives Amazon the ability to monitor the progress of its drivers and deliv- eries without relying entirely upon the Amazon Flex app that drivers use for routing and for recording deliveries. During delivery, partners and Amazon staff located at delivery stations closely monitor drivers through a dashboard42 in order to be able to respond to delays or problems in delivery. The Rivian as an element of Amazon’s urban stack in this way integrates further into Amazon’s logistical operations, with data available for further optimization. In some ways, the Rivian van is very much unlike contemporary autonomous cars. As Sprenger notes, with cars becoming more autonomous, they require less attention from drivers, exacerbating a trend in which, as John Urry has put it earlier, the car becomes a »privatized, cocooned, moving capsule,«43 increasingly discon- necting passengers from its environment. These capsules’ designs, for example in terms of sound and ergonomics, »promise to turn the car into a world of its own, 38 Vincent: »Amazon Stops Field Tests of Its Delivery Robot Scout«. 39 Tarasov: »Amazon’s 100 Drone Deliveries Puts Prime Air Far Behind Alphabet’s Wing and Walmart Partner Zipline«. 40 Palmer: »Amazon Uses an App Called Mentor to Track and Discipline Delivery Drivers«. 41 In Germany, Amazon justifies its close monitoring of staff in fulfilment centers through barcode scans with the necessity of these scans for its operations, effectively circumvent- ing data protection regulations. Presumably, a similar way to introduce close monitoring could be at work here. On the court case regarding surveillance in fulfilment centers, see: Lenz/Kausche: »Nach Urteil zur Mitarbeiterkontrolle bei Amazon«. 42 For an impression of this monitoring, see: What an Amazon Dispatcher Sees While You’re Delivering, https://youtu.be/8SJYPf73iJI, 24.03.2024. 43 Urry: »Inhabiting the Car«, p. 20. AUTOMATED DELIVERY NAVIGATIONEN L IE F E R N . L O G IST IK E N , D A T E N U N D P O L IT IK E N 57 sealed off from the outside and allowing its safe passage, without being contami- nated by the world.«44 This encapsulation is anathema to the delivery van, which must be amenable to logistical flows and thus interactions with the outside world. Therefore, a different design is required. For the Rivian, this starts with the driver seat, which doesn’t swivel but sits on a raised pedestal so that drivers, as they turn towards the back, can literally hit the ground running on the flat floor. When drivers put the van into parking gear, the bulkhead door to the back of the van opens au- tomatically. The curbside sliding door allows the driver to park close to the curb – or rather the hurb as a »pop-up package hub«45 enrolled in Amazon’s urban stack – and the key fob locks the van automatically as the drivers move away and opens the van up again once they return. In these and many other ways, the Rivian van is designed to enable logistical flows of last mile delivery. The design includes an endless number of scripts for delivery labor, promoted as ergonomic and safe and conveniently enabling smooth bodily movements. The van here mirrors the protocols of other logistical spaces Amazon has designed in terms of optimized workflows for logistical labor. It also integrates materially into these, for example when the roller shutter at the back of the van allows for easy access to place the bags with parcels which the drivers pick up at delivery stations, or with the shelving in the back sized to fit the bags and foldable so that larger parcels can also be placed inside. In this way, Amazon in- scribes the logics of its logistical operations – the scripts imposed by the design of the van on delivery labor and the habits it wants to encourage; the forms of moni- toring and control that the sensors, interfaces, and algorithms make possible; the workflows and its materialities in terms of parcels, bags, and so on – into its urban stack for last mile delivery, beyond its delivery stations and fulfillment centers. Now that Amazon’s deal with Rivian is no longer exclusive, Rivian sells delivery vans nearly identical to the Delivery 500 and 700 to other clients, making Amazon’s pro- tocols standard in last mile delivery.46 CONCLUSION The door/porch, the locker and the van are just three elements of Amazon’s urban stack for last mile delivery, with the van connecting both the door/porch and the locker to its vast delivery network in addressing them as the end points of delivery. 44 Sprenger: »Autonome Automobilität«, p. 53-54, my translation. 45 Bonanos: »The Hurb Is the Pop-Up Package Hub We’re Stuck With (For Now)«. Credit goes to Matthew Hockenberry for this reference; his work on this element of Amazon’s urban stack is forthcoming. 46 That the van developed by the company Arrival for and with UPS includes many of the same features in this regard and looks cunningly similar demonstrates how such co-oper- ations between electric vehicle manufacturers and logistics companies explicate the pro- tocols and logics of last mile delivery in design. See https://arrival.com/topic/van, 24.03.2024. ARMIN BEVERUNGEN NAVIGATIONEN 58 L IE F E R N . L O G IS T IK E N , D A T E N U N D P O L IT IK E N These three elements most directly demonstrate how Amazon is remaking cities, both in terms of reasserting automobility as a prime organizing logic of cities today and how Amazon imagines its relation to customers to which convenience is deliv- ered. Their exploration has already shown how entangled they are with other ele- ments, such as the Flex app for drivers or the Amazon app for consumers. It has also shown how other existing elements of the urban stack are enrolled, for exam- ple, existing infrastructures of mobility and convenience. The three elements demonstrate how Amazon draws on existing materialities and socialities while re- programming other elements and the kinds of scripts they impose in particular on delivery labor. This reprogramming not only enrolls elements in Amazon’s urban stack but also its logistical operations more widely, both materially, for example when the size of Amazon’s packages and bags determine the dimensions of shelves and other aspects along the operational chain, and in terms of a data economy geared towards data extractivism and machine learning. As Ned Rossiter and Brett Neilson have argued, automation makes futures, needs data, intensifies extraction, and adapts environments.47 What Amazon’s automation of the logistical city demonstrates is how much the optimization of logistical operations requires exten- sive data extraction fuelling machine learning. The urban futures devised and put in place here, while adaptive for example to environmental or labor market condi- tions, also impose a particular vision of logistical urbanism traversed and infused with the protocols and logics of Amazon’s logistical operations. We don’t need to look as far as Amazon’s speculative patents, for example for flying warehouses or extensive drone delivery, to identify an »Amazon urbanism.«48 Attention to the elements of Amazon’s urban stack also highlights how auto- mation here is not simply machinic, as in autonomous cars and robots, but also geared towards including humans through scripts and habits, as described above with regards to how Amazon invites us to engage with its lockers or in terms of the logics of »buy-now« and »click-and-collect.« This is most obviously the case when looking at how the van and other environments designed by Amazon, such as its delivery stations and fulfillment centers, carefully script bodily movements and op- erations and draw on extensive data extraction to optimize these actions, for ex- ample when it comes to driving and parking or the ergonomics of package han- dling.49 The extent of these logics of automation becomes clear when looking at how customers are also enrolled – through the scripts and habits associated with handling the Amazon app and its lockers, but even more so through its online store and its gadgets such as Echos, which bring voice interactions with Amazon to our smart homes. David Hill has noted how »Amazon is a good example of what Wendy Chun (2016) describes as the disappearing from consciousness of ›habitual media‹ 47 Neilson/Rossiter: »Theses on Automation and Labour«. 48 Stewart: »Amazon Urbanism«. 49 For more, particularly with regards to fulfillment centers, see Beverungen: »Automatis- iertes Verhalten: Regierungskünste bei Amazon«. AUTOMATED DELIVERY NAVIGATIONEN L IE F E R N . L O G IST IK E N , D A T E N U N D P O L IT IK E N 59 […] – a process of withholding from awareness what ought to be at the centre [sic] of our attention.« Emily West has, in a similar vein, noted a shift at Amazon away from »a choosing subject to served self,« wherein shopping with Amazon, »embed- ded in daily habits and ways of thinking,« becomes unthinking, which she considers a primary cost for Amazon’s convenience.50 Yet where Amazon seeks to automate the logistical city, automation also breaks down. David Bissell suggests that these breakdowns and failures of automa- tion, and the associated glitches, can constitute kinds of »deautomation« in an on- demand platform capitalism which so thoroughly relies on it.51 Such a politics of deautomation would also need to direct its attention at the elements of Amazon’s urban stack for last mile delivery if its scripts and associated habits, not to mention the forms of algorithmic management, are to be challenged. ACKNOWLEDGMENTS Research for this contribution was funded by the Lower Saxony Ministry for Science and Culture and the Volkswagen Foundation as part of the zukunft.niedersachsen funding line through the project »Automating the Logistical City: Space, Algorithms, Speculation.« Thanks in particular to Ilia Antenucci, Maja-Lee Voigt, Ulf Treger and Klara Friese for the collaborative research out of which this contribution emerges. REFERENCES Bagwell, Philip Sidney: The Transport Revolution: 1770-1985, New York, 2014. Beverungen, Armin: »Automatisiertes Verhalten: Regierungskünste bei Amazon«, in: Gräfe, Sophia/ Toepfer, Georg (eds.): Wissensgeschichte des Verhaltens. Interdisziplinäre Perspektiven, Berlin, 2024. Beverungen, Armin: »Collectivizing Convenience? From Delivery to Logisticality«, in: Neves, Joshua/ Steinberg, Marc (eds.): In/Convenience: Inhabiting the Lo- gistical Surround, Amsterdam, 2024. Beverungen, Armin: »The Invisualities of Capture in Amazon’s Logistical«, in: Digital Culture & Society 7, No. 2, 2022, pp. 185-202. Bonanos, Christopher: »The Hurb Is the Pop-Up Package Hub We’re Stuck With (For Now)«. Curbed, 12 May 2023, https://www.curbed.com/2023/05/hurb- package-delivery-curb-hub-amazon-truck-curb.html, 24.03.2024. Bratton, Benjamin H.: The Stack: On Software and Sovereignty, Cambridge, MA, 2015. 50 West: Buy Now, p. 111. 51 Bissell: »The Politics of Deautomation«. ARMIN BEVERUNGEN NAVIGATIONEN 60 L IE F E R N . L O G IS T IK E N , D A T E N U N D P O L IT IK E N Bridges, Lauren: »Infrastructural Obfuscation: Unpacking the Carceral Logics of the Ring Surveillant Assemblage«, in: Information, Communication & Society 24, No. 6, 2021, pp. 830-849. Dhaliwal, Ranjodh Singh: »On Addressability, or What Even Is Computation?«, in: Critical Inquiry 49, No. 1, 2022, pp. 1-27. Fang, Jiawen et al.: The Spatial Dynamics of Amazon Lockers in Los Angeles County, Los Angeles, 2019. Fingerhut, Elena: »Übertragen und Speichern: Zum Verhältnis von Adressen und medialen Gehäusen«, in: Neubert, Christoph et al. (eds.): Gehäuse: Mediale Einkapselungen, Paderborn, 2017, pp. 343-361. Forman, George: »Using Learning-to-Rank to Precisely Locate Where to Deliver Packages«, Amazon Science (blog), 15 September 2021, https://www.ama- zon.science/blog/using-learning-to-rank-to-precisely-locate-where-to-de- liver-packages, 24.03.2024. Heinrichs, Randi: »Where Do the Data Live? Anonymity and Neighborhood Net- works« in: Anon Collective (ed.): Book of Anonymity, Santa Barbara, 2021, pp. 226-254. Hockenberry, Matthew: »Cellular Capitalism: Life and Labor at the End of the Dig- ital Supply Chain«, in: Graham, Mark/Ferrari, Fabian (eds.): Digital Work in the Planetary Market, Cambridge, MA, 2022, pp. 263-280. Hockenberry, Matthew/Starosielski, Nicole/Zieger, Susan (eds.): Assembly Codes: The Logistics of Media, Durham, NC, 2021. Kampker, Achim et al. (eds.): Think Big, Start Small: StreetScooter die E-Mobile Erfolgsstory: Innovationsprozesse Radikal Effizienter. Berlin/Heidelberg, 2017. Kim, Woojung/Wang, Xiaokun Cara: »The Adoption of Alternative Delivery Loca- tions in New York City: Who and How Far?«, in: Transportation Research Part A: Policy and Practice 158, 2022, pp. 127-140. Kittler, Friedrich: »The City Is a Medium«, in: New Literary History 27, No. 4, 1996, pp. 717-729, trans. Matthew Griffin. Klose, Alexander: The Container Principle: How a Box Changes the Way We Think, Cambridge, MA, 2015. Latour, Bruno: »Where Are the Missing Masses? The Sociology of a Few Mundane Artifacts«, in: Bijker, Wiebe E./Law, John (eds.): In Shaping Technology/Build- ing Society: Studies in Sociotechnical Change, Cambridge, MA, 1992, pp. 225- 258. Law, John/Mol, Annemarie: »Situating Technoscience: An Inquiry into Spatialities«, in: Environment and Planning D: Society and Space 19, No. 5, 2001, pp. 609- 621. Lenz, Miriam/Kausche, Katharina: »Nach Urteil zur Mitarbeiterkontrolle bei Ama- zon: Datenschutzbeauftragte legt Berufung ein«, correctiv.org, 28 April 2023, AUTOMATED DELIVERY NAVIGATIONEN L IE F E R N . L O G IST IK E N , D A T E N U N D P O L IT IK E N 61 https://correctiv.org/aktuelles/ungerechte-arbeit/2023/04/28/nach-urteil-mit- arbeiterkontrolle-bei-amazon-datenschutzbeauftragte-berufung/, 24.03.2024. Levinson, Marc: The Box: How the Shipping Container Made the World Smaller and the World Economy Bigger, Princeton, NJ, 2016. Lyster, Clare: »Disciplinary Hybrids: Retail Landscapes of the Post-Human City«, in: Architectural Design 89, No. 1, 2019, pp. 100-105. MacKenzie, Adrian/Munster, Anna: »Platform Seeing: Image Ensembles and Their Invisualities«, in: Theory, Culture & Society 36, No. 5, 2019, pp. 3-22. Mattern, Shannon: Code + Clay... Data + Dirt: Five Thousand Years of Urban Media. Minneapolis, MN, 2017. Mattern, Shannon: »Interfacing Urban Intelligence«, in: Places Journal, April, 2014, https://placesjournal.org/article/interfacing-urban-intelligence/, 24.03.2024. Neilson, Brett/Rossiter, Ned: »Theses on Automation and Labour«, in: Bigo, Didier et al. (eds.): Data Politics: Worlds, Subjects, Rights, London, 2019, pp. 187- 206. Niemann, Greg: Big Brown: The Untold Story of UPS. San Francisco, 2007. O’Donovan, Caroline/Bogage, Jacob: »A Rural Post Office Was Told to Prioritize Amazon Packages. Chaos Ensued«, in: Washington Post, 28 November 2023, sec. Technology, https://www.washingtonpost.com/technology/2023/11/28 /amazon-missed-mail-rural-towns/, 24.03.2024. Palmer, Annie: »Amazon Uses an App Called Mentor to Track and Discipline De- livery Drivers«, in: CNBC, 12 February 2021, sec. Technology, https://www.cnbc.com/2021/02/12/amazon-mentor-app-tracks-and-disci- plines-delivery-drivers.html, 24.03.2024 Peters, John Durham: »Calendar, Clock, Tower«, in: Stolow, Jeremy (ed.): Deus in Machina, New York, NY, 2012, pp. 24-42. Pöchhacker, Nikolaus/Nyckel, Eva-Maria: »Logistics of Probability: Anticipatory Shipping and the Production of Markets«, in: Burkhardt, Marcus u.a. (eds.) Ex- plorations in Digital Cultures, Lüneburg, 2020. Rodrigue, Jean-Paul: »The Distribution Network of Amazon and the Footprint of Freight Digitalization«, in: Journal of Transport Geography 88, 1 October, 2020., art.-no.102825. Schaefer, Jaclyn S./Figliozzi, Miguel A.: »Spatial Accessibility and Equity Analysis of Amazon Parcel Lockers Facilities«, in: Journal of Transport Geography 97, 1 December, 2021, art.-no. 103212. Shapiro, Aaron: »Platform Urbanism in a Pandemic: Dark Stores, Ghost Kitchens, and the Logistical-Urban Frontier«, in: Journal of Consumer Culture 23, No. 1, 2023, pp. 168-187. Siegert, Bernhard: »Doors: On the Materiality of the Symbolic«, in: Grey Room 47, April, 2012, pp. 6-23, trans. Peters, John Durham. ARMIN BEVERUNGEN NAVIGATIONEN 62 L IE F E R N . L O G IS T IK E N , D A T E N U N D P O L IT IK E N Siegert, Bernard: Relays: Literature as an Epoch of the Postal System, trans. Repp, Kevin, Stanford, CA, 1999. Sprenger, Florian: »Autonome Automobilität: Eine medien- und kulturwissen- schaftliche Einführung«, in: Sprenger, Florian (ed.): Autonome Autos: Medien- und kulturwissenschaftliche Perspektiven auf die Zukunft der Mobilität, Biele- feld, 2021, pp. 9-82. Stewart, Matthew: »Amazon Urbanism: Patents and The Totalizing World of Big Tech Futures«, in: Failed Architecture (blog), 23 May 2018, https://failedarchi- tecture.com/amazon-urbanism-patents-and-the-totalizing-world-of-big-tech- futures/, 24.03.2024. Tarasov, Katie: »Amazon’s 100 Drone Deliveries Puts Prime Air Far Behind Alpha- bet’s Wing and Walmart Partner Zipline«, CNBC, 18 May 2023, https://www.cnbc.com/2023/05/18/amazons-100-drone-deliveries-puts- prime-air-behind-google-and-walmart.html, 24.03.2024. Urry, John: »Inhabiting the Car«, in: The Sociological Review 54, No. 1_suppl, 2006, pp. 17-31. Vincent, James: »Amazon Stops Field Tests of Its Delivery Robot Scout«, The Verge, 7 October 2022, https://www.theverge.com/2022/10/7/23392360/amazon- disbands-delivery-robot-scout-development, 24.03.2024. West, Emily: Buy Now: How Amazon Branded Convenience and Normalized Mo- nopoly, Cambridge, MA, 2022. Cover Titelblatt INHALT INTRODUCTION : WHEN DELIVERY MEETS LOGISTICS: ON INFRASTRUCTURES, AESTHETICS, AND LABOR / MIGLĖ BAREIKYTĖ AND JULIA BEE MAGICAL DELIVERY? LOGISTICS, DELIVERY, AND INFRASTRUCTURES DELIVERY AND MEDIA STUDIES AESTHETICS LABOR REFERENCES INFRASTRUKTUREN WELTEN LIEFERN. CHOREOGRAPHISCHE KONTROLLE UND PRODUKTION IM LOGISTISCHEN KAPITALISMUS / GERKO EGERT CHOREOGRAPHIE I: BEWEGUNGSPRODUKTION CHOREOGRAPHIE II: DIE DATAFIZIERUNG DER BEWEGUNG VON DER BEWEGUNG ZU BEWEGUNGSSYSTEMEN ÖKOLOGISCHE KONTROLLE PRODUKTION VON UMWELTEN VOM LIEFERN ZUM LOGISTISCHEN KAPITALISMUS LITERATURVERZEICHNIS AUTOMATED DELIVERY: AMAZON’S URBAN STACK / ARMIN BEVERUNGEN AMAZON’S URBAN STACK THE DOOR/PORCH THE LOCKER THE VAN CONCLUSION ACKNOWLEDGMENTS REFERENCES DER WEG DES BUCHES. BIBLIOTHEKEN ZWISCHEN PROZESSARCHITEKTUR, FLOW CHARTS UND SERENDIPITY / HANNAH WIEMER PROZESSARCHITEKTUR DES 19. JAHRHUNDERTS: GEBÄUDE ALS WEGE WEG DES BUCHES UND WEGE DER LESER:INNEN WEGE PLANEN: FLOW CHARTS, SOLL-MODELLE UND DIE BIBLIOTHEKARISCHE ARBEIT LITERATURVERZEICHNIS ABBILDUNGSVERZEICHNIS FLATTENING THE MAP: HOW HUMAN MOVEMENT IS TURNED INTO A LOGISTICAL PROBLEM; THE CASES OF ASYLUM AND HUMANITARIAN RELIEF / MICHELLE PFEIFER AND PATRICIA WARD INTRODUCTION REGULATING HUMAN MOBILITY FROM MANAGEMENT TO LOGISTICS MAPPING TECHNIQUES AS LOGISTICAL TRANSFORMATION CONCLUSION REFERENCES LIST OF FIGURES ÄSTHETIK AUSGELIEFERT. LIEFERN UND CARE-ARBEIT ALS FILMISCHE INFRASTRUKTURKRITIK IN SORRY WE MISSED YOU / FELIX HASEBRINK UND MAXIMILIAN RÜNKER 1. MEETING THE MOMENT 2. INFRASTRUKTURFORSCHUNG UND LIEFERLOGISTIK 3. INFRASTRUKTURÄSTHETIK IM FILM 4. INFRASTRUKTURSZENEN DES LIEFERNS 5. INFRASTRUKTURSZENEN VON CARE-ARBEIT 6. FILMISCHE INFRASTRUKTURKRITIK LITERATURVERZEICHNIS THE WORKER AS FUTURIST PROJECT: INTERVIEW WITH MAX HAIVEN ABOUT WRITING WITH AMAZON WORKERS AS PRACTICE OF RADICAL IMAGINATION / MAX HAIVEN, MIGLĖ BAREIKYTĖ AND JULIA BEE THE WORKER AS FUTURIST – A PROJECT IN SOLIDARITY WITH AMAZON RANK-AND-FILE WORKERS AMAZON AS DYSTOPIAN SCI-FI »ONE PLATFORM TO RULE THEM ALL« NEW NARRATIVES WITH SPECULATIVE FICTION: WORKERS LITERATURE AND EXPERIMENTAL METHODOLOGIES FROM WRITING TO VISION AND SOLIDARITY THE »NOT SO HIDDEN« WORK OF FOOD DELIVERY / MATHIAS DENECKE INTRODUCTION DELIVERY WORKERS AND ALGORITHMIC MANAGEMENT WORKERS’ RESISTANCE AND THE PUBLIC CUSTOMERS CONCLUSION ACKNOWLEDGMENTS REFERENCES ARBEIT PIZZALIEFERN IST POLITISCH. ZUR KOMMODIFIZIERUNG UND FINANZIALISIERUNG SOZIALER REPRODUKTION / ANKE STRÜVER 1. EINLEITUNG 2. GIG-PLATTFORMEN FÜR ESSENSSCHNELLLIEFERUNGEN 3. DIE PLATTFORMISIERUNG VON SOZIALER REPRODUKTION 4. AUSBLICK: REPOLITISIERUNG VON SOZIALER REPRODUKTION DURCH FINANZIALISIERUNG? LITERATURVERZEICHNIS AUSGELIEFERT? EIN GESPRÄCH ÜBER WIDERSTANDSFORMEN IM LIEFERKAPITALISMUS / SEBASTIAN RANDERATH UND SEMIH YALCIN LIEFERN AM LIMIT? APP-BASIERT UND ATOMISIERT? ZWISCHEN INFORMELLEN WIDERSTÄNDEN UND GEWERKSCHAFTLICHER INTERESSENVERTRETUNG WIDERSTÄNDE ZWISCHEN KONFLIKT UND SOLIDARITÄT LITERATURVERZEICHNIS MAKING TIME WORK: ETHNOGRAPHIC EXPLORATIONS OF TEMPORAL EXPERIENCES IN DELIVERY WORK / ŽIVILĖ MIEŽYTĖ INTRODUCTION TEMPORAL STRUCTURE OF DELIVERY WORK HARDCORE WORK NEGOTIATING TEMPORAL STRUCTURE TEMPORAL AGENCY AND HARD WORK MAKING TIME WORK REFERENCES NACHRUF AUF JOCHEN VENUS ABSTRACTS BEVERUNGEN, ARMIN; AUTOMATED DELIVERY: AMAZON’S URBAN STACK DENECKE, MATHIAS; THE »NOT SO HIDDEN« WORK OF FOOD DELIVERY EGERT, GERKO; WELTEN LIEFERN. CHOREOGRAPHISCHE KONTROLLE UND PRODUKTION IM LOGISTISCHEN KAPITALISMUS HAIVEN, MAX; BAREIKYTĖ, MIGLĖ; BEE, JULIA; THE WORKER AS FUTURIST PROJECT: INTERVIEW WITH MAX HAIVEN ABOUT WRITING WITH AMAZON WORKERS AS PRACTICE OF RADICAL IMAGINATION HASEBRINK, FELIX; RÜNKER, MAXIMILIAN; AUSGELIEFERT. LIEFERN UND CAREARBEIT ALS FILMISCHE INFRASTRUKTURKRITIK IN SORRY WE MISSED YOU MIEŽYTĖ, ŽIVILĖ; MAKING TIME WORK: ETHNOGRAPHIC EXPLORATIONS OF TEMPORAL EXPERIENCES IN DELIVERY WORK PFEIFER, MICHELLE; WARD, PATRICIA; FLATTENING THE MAP: HOW HUMAN MOVEMENT IS TURNED INTO A LOGISTICAL PROBLEM, THE CASES OF ASYLUM AND HUMANITARIAN RELIEF RANDERATH, SEBASTIAN; YALCIN, SEMIH; AUSGELIEFERT? EIN GESPRÄCH ÜBER WIDERSTANDSFORMEN IM LIEFERKAPITALISMUS STRÜVER, ANKE; PIZZALIEFERN IST POLITISCH – ZUR KOMMODIFIZIERUNG UND FINANZIALISIERUNG SOZIALER REPRODUKTION WIEMER, HANNAH; DER WEG DES BUCHES. BIBLIOTHEKEN ZWISCHEN PROZESSARCHITEKTUR, FLOW CHARTS UND SERENDIPITY KURZBIOGRAFIEN LIEFERBARE HEFTE