We asked seven different maps for navigation in between a 15-mile radius. Only one map provided directions. The destination? Gaza.
Source: Google Maps
About a month ago, I visited Gaza virtually. Intrigued by its proximity to Egypt, I entered street view via Google Maps, although there were few locations where one could do so. I then asked for directions from an arbitrary nearby landmark, the Rafah Central Hospital in Egypt. The distance from the hospital to a beachfront neighbourhood called Al Mawasi, near Khan Yunis, in southern Gaza, is roughly twelve miles, but becomes about twenty car miles due to the roads. In order to traverse the stretch, one must pass though the Rafah crossing. All in all the journey takes around 40 minutes.
A few weeks ago, I googled the same directions using the same two points. This time, it could not be done. Neither walking nor driving directions for this itinerary exist on Google Maps anymore, and if you ask for them it simply states that directions cannot be calculated.
Source: Google Maps
Rafah Hospital, Egypt, to Kerem Shalom, Israel. Source: Google Maps.
Instead, I chose a new destination; nearby kibbutz and border crossing village Kerem Shalom just a handful of miles away in Israel. This time, Google Maps generated directions without issue. The same procedure was attempted with several other map software platforms, including Apple Maps, Waze, AA Route Planner, MapQuest, Via Michelin, and Bing.
Bing—though it neglected to acknowledge Al Mawasi as an address and instead directed to Khan Yunis—was the only platform that navigated in or out of Gaza. Every other map either produced a blank page or a statement along the lines of the directions could not be calculated.
Why won’t most maps show directions to Gaza anymore?
Source: Apple Maps
The impossibility of being able to navigate to or from Gaza gives it a detached sense of placelessness and calls into question its legitimacy and existence. Its presence on the map, or lack thereof, has long since been deteriorating, so the sudden incapacity to even get directions to it feels like the next step in a kind of cartographical inevitability, but not one that has been openly acknowledged.
Directions from Rafah Central Hospital, Egypt, to Khan Yunis, Gaza. Source: Bing.
Maps can ontologically and epistemologically inform other visual and representational modes of knowledge, and most map readers will not be able to differentiate the Cartesian distinctions between the map and the territory it claims to represent. For most people, maps are just maps. Maps are also compelling visual images with symbolic rhetorical power. They are images of domination, segregation, and hostility.
A plain reading of the drawing and reading of maps as a kind of information exchange of universal truth, or a uniformly topographic survey of the surface of the planet, would ignore the complex politics of maps, their interpretive investigation, and their relationship to international law. A debate continues around the practice of map generalisation, and whether the science itself is a science at all, if mapping is “holistic or fragmentary, stochastic or regular, invariant or contingent, natural or cultural, objective or subjective, functional or symbolic, and so on.”3rethinking_maps_introduction.pdf
Map making is in fact one of the most primitive incarnations of AI and contemporary rendering of maps employs various forms of automation, algorithmic efficiency, and visualisation technology in addition to human interaction. Online mapping follows the basic mechanisms of AI and the premise of its technology as a product of processes surveillance – as does any kind of surveillance that relies on the extraction of data. As opposed to AI being used as a surveillance strategy in its most plain sense; like CCTV, surveillance drones, or biometric regulation, navigation mapping surveils in another way; through its analysis of user data and real time map engagement.
The ensuing questions that arise (regarding territory, movement, and nation state borders) come down to one basic question; do maps exist as a navigation tool, a surveillance tool, or something else? One could argue that live mapping falls under the umbrella of surveillance capitalism, especially as more corporations appropriate a kind of global citizenship, no longer observing national boundaries.
Either way, the greater inconsistency of maps represents a bigger issue, which is the breakdown of trust of any kind of information that falls under the category of digital or artificial intelligence. The validity of knowledge seems to be increasingly unreliable, as is the distinction between what is a fact and what is a theory, or what is a place and what is not.
Even static maps, which have been widely appropriated, have been proven not to convey the topological reality. For instance, the graphic below depicts the substantial and inaccurate inflation of numerous regions on the standard world map, including the USA, Russia, the European Union, Greendland, and Australia to name a few.
The subjective sizing of first-world countries demonstrates a god-like authority bestowed upon the map maker and the world as it is interpreted, understood, and imagined by the readers of maps. Other more artistic approaches to mapping that openly favour the privileged user are compatible with this interpretation, if not to make light of it. The currently inactive site hoodmaps conveys open-source mapping that allows residents in various areas to write their own borders, descriptions of neighbourhoods, and assessments of gentrification.
Though democratic, these maps are simply not accurate, and not reliable. One could never use them to navigate. A more nomothetic approach, which emphasises laws and denies such idiosyncratic differences, also falls short of functioning, as it relies heavily on generalisations and an unrealistic investment in logical universalisation of disputed spheres.
At the core of mapping debates lies the map reader’s agency. The acknowledgement and social contextualisation of the map reader represents a cultural turn in academic geography and mapping, a turn away from production and towards consumption. Subjugation of maps, or the apparent suppression of navigational intelligence, demonstrates a phenomenon described by digital mapping consultant Steven Feldman as “agnostic cartography,”4https://knowwhereconsulting.co.uk/blog/when-politics-meet-maps-there-is-no-right/ a kind of geopolitical discretion that sullies the proverbial lines of demarcation.
Google Maps in particular has been characterised by its agnostic cartography since decades. Instead of maintaining a definitive world map, Google Maps reflects a constantly changing, and dare I say unstable, interpretation of geography that has frequently entangled it in numerous ongoing geopolitical conflicts, including a decades long deliberation over its designation of Palestinian land.5https://www.npr.org/sections/talk/2010/07/12/128466278/more-than-a-map
What this means is that instead of Google Maps, or any other interactive mapping tool, managing its data with scientific diligence, or collecting its information from the topography of territories, the internet itself and its users have as much influence on the map as the physical earth. It is not a representation of the world, it is a representation of the perception of the world inhabited by the most influential majority.
What the map says is more indicative of social attitudes of human beings than anything encyclopedic, whose intelligence is to some extent fixed. In this way, agnostic cartography is mapped using a knowledge framework whose parameters are similar to what artist Hito Steyerl has described as compositional methodologies in her recently published essay, Mean Images, in New Left Review.
Steyerl’s essay refers to “stochastic discrimination,” or a predictably unpredictable tendency to make prejudicial distinction for one direction from another. At the heart of this revelation is an inherently political aspect of AI, sheerly for the fact that its initial programmers are always living, breathing, biased human beings. It is therefor safe to assert that behind the function of any AI system is an anthropoid agenda, with all its problems, prepossessions, and preferences.
Steyerl discusses the generation of images by AI and the criticality of the reference pool from which images are generated. Increasingly, models are being trained on their own output, meaning that the data, whether factual or not, is based off of “likeliness” rather than “likeness”. The AI then generates what is again used as data in an invariable feedback loop that increasingly excludes that which is not currently and aggressively contributing.
Can Steyerl’s observations about composite aesthetics and mean images be applied to composite cartography? Are the maps mean too?
In context, “mean” refers to a way of describing something akin to the median. In another and more subjective interpretation of “mean,” the word not only refers to nastiness. Not only is the unavailability of AI tools or maps themselves nasty, but the maps themselves are nasty too. Nasty maps don’t depend on places but on hallucinations of parameters, and like the mean images, follow a system of social patterns, vector coordinates, and a persistent subjectivity to the colonial narrative.
Like the images, the maps also no longer portray reality, but rather a demographic abstraction; If the images are dreams without sleep then these maps are destinations without journeys. They are standardised renderings drawn from collective ideas and aspirations.
The means of production of mean images and mean maps is similar in that it is decidedly and strategically unclear regarding who the producers are and how they are made. Steyerl explains that the reason she refers to renderings as “mean images” is because they “rely on an averaging of stolen data” as well as the tradition and history of statistics; something that often “crosses over into the field of eugenics, biopolitics, population regulation” and the defining norms for human beings as established by the governing power – in this case the governing power would not only be in control of the data, but would do so by inherently also being in control of the surveillance, how it is used, and the building of its infrastructure.
The video below portrays an Israeli soldier entering a school in Gaza. Upon encountering numerous maps of Palestine, he proceeds to redraw the maps, remapping the land in real time.
The mainstream media’s skewed portrayal of Israel’s invasion of Gaza is as palpable through the channel of AI as anywhere else, which begs the question of its subjectivity, misuse, and regulation. While employment of propaganda and misinformation is as old as the hills, the advent of AI presents an unprecedented obligation for validating information, and perhaps a new procedure altogether. The most productive AI feature for maps is surely the notion of traffic, and this was also the first asset to be withdrawn from Palestinian maps.
The decline of internet connectivity parallels the decline of information in general, as well as the validity of Palestine IP addresses as an internet presence or cartographical reference point. In fact, the most relevant and up to date map at the current moment is one that has been written by the Israeli Defence Force and distributed via qr codes printed onto leaflets. The sad irony of the map being inaccessible to most because of the lack of internet connectivity may also be read as a blessing in disguise, since the ultimate purpose of the online map is to be accessed, at which point it ushers civilians to deserted areas void of sanitation, shelter, or any chance to survive.
The map segments the area on the Gaza strip into squares, indicating with different colours which areas are safe and which are not. This map, though unaccessible to most people (seeing as they are without electricity as well as internet) is arguably the most important map to anyone currently located in Gaza. In many ways it is the only map that matters, and it is entirely at the discretion of the IDF.
Detail from one of the grid evacuation warnings posted by Lt Col Avichay Adraee of the IDF on social media. Photograph: @AvichayAdraee/X
The map is nothing more than another AI model, its parameters defined by a framework that comprises prejudices of the mainstream, the majority, and the users. If the instrumentality of user influence is as critical as Steyerl asserts, then A to B process of something like Stable Diffusion’s would loosely mimic the A to B journey on a map’s itinerary. Steyerl explains that the mechanisms are not just excluding based on bias, but that they are making the exclusion productive by projecting it forward. This is a much more active process than any basic bias.
This points to a poststructuralist way of knowing and mapping the world. Instead of a belief in absolute space, or a socially constructed world, technological circumstances become the determining factors of space and the marking of barriers. Alternative ways of understanding mapping have emphasised relativity and contingency in a world where notions of reality have been replaced by simulation.
A 2010 article by John Gravois, auspiciously removed by its news outlet, explained that the indecisive nature of Google Mapping is a result of business interests and the logic of its software, that logic being one that favours engagement over actual data by way of catering to the beliefs of one nation over another.
All AI are not created alike, and the system will either run off of pre-existing data, or data that it continues to acquire the more it is used. For every day that Gaza, and Palestinian voices, are offline, the less they contribute to the pool of data collected and used by AI to generate anything, from a statement to an image.
Source: Google Maps
We still don’t know why navigation in or out of Gaza is virtually impossible (literally). Perhaps the reason does come down to offline servers…but that would not explain the Bing map working. Moreover, it would not explain why navigation remains fully functional within destinations inside of Gaza. When asked to navigate from Al Mawasi to Khan Yunis, Google Maps provides both driving and walking directions, again without a hitch. What it will no longer do, it seems, is assist in the navigation into or outof the area. Yet, it continues to support movement within the area. It sounds not unlike how we might describe a guinea pig enclosure.
To say that would be obscene, right? But the map doesn’t lie.
Without permission of Naked Snail Magazine, no one may copy, reproduce, or otherwise use the content of this website. Please read the website before commenting. If you have any questions please send an email.
