I propose the term “Adversarial Infrastructure” to subvert the conventional understanding of bridges as connectors, objects of linkage that function in opposition to walls and borders, the emblematic tools of territorial disconnection and delimination. In 2016, Mark Zuckerberg, CEO of Facebook, proclaimed:

“Instead of walls, we can help people build bridges”. 

This is not an invention of his PR team. Indeed, bridges exist wthin the social imagination as symbols of unity and peace, standing in contrast to the violent images of separation evoked by border walls.

Logistics is covering the violence it is inflicting more than efficiently by rendering itself invisible. Nevertheless, the problem is not only the reduction of flow to the destination point in the movement. When being analysed, logistics is too often understood as the “capacity to move goods” that aims for planetary flux and therefore exists in opposition to disruption. I present Adversarial Infrastructure as the logic of the contemporary mobility regimes⁠ — what Salter defines as a new form of logistics that does not guarantee the mobility of any actor but instead brings about its restriction (Salter 2013). It is a structure in which certain actors are rendered (im)mobile, and their capacity to function outside of the programmed pattern is “removed from the political realm and treated as either technical or economic questions” (Salter 2013, 9).

The term “adversarial” comes from its use in machine learning methods. Adversarial machine learning is based on the idea that algorithms can learn via competition. Generative adversarial networks (GANs) consist of two neural networks that are designed to be antagonistic to one another. Even though their functions are programmed to oppose each other, their entrapment in a looped contest strengthens the neural network as a whole.

This same capacity defines Adversarial Infrastructure: while divergent functions may appear to create friction and technical problems for one another, when taken together they actually strengthen the infrastructure’s capacity to inflict harm. In this way, 

Adversarial Infrastructure aims not only to facilitate flow but also to block and disrupt. Such heterogeneity of its functions⁠ gives an opportunity to mask one behind the other whenever required.

Therefore, the semblance of antagonistic friction of these functions enables the forceful realisation of their weaponised nature. Adversarial Infrastructure realises a “capacity to contain and connect” simultaneously against the same Enemy of Empire (Cowen 2017). For instance, “Roads to Nowhere” in the US have been used to fuel racial segregation, and the Russian Crimean Bridge has allowed for extractivism in occupied land and economic blockade against Ukraine (Miller 2018 quoted in Lambert 2018).

An analysis of the Crimean Bridge as Adversarial Infrastructure reveals the counter-intuitive notion about how a bridge is simultaneously a border that facilitates and disrupts flow.

Follow the model to see the different modes of mobility which Adversarial Infrastructure aims to control.