AAA 2022: Infrastructural Intimacies – Scales of Infrastructure
DRAFT - DO NOT CIRCULATE OR CITE
Sensing the System: Inside/Outside the Black Box
Dean Chahim (University of Texas at El Paso)
While there is now little doubt that embodied senses play a crucial role in operating, maintaining, and even designing infrastructure, what is less clear from this work are the politics of sensing infrastructure, particularly around the question of state power.[1] I therefore want to talk today about how states demarcate what can and cannot be sensed, and how this affects movements seeking to hold the state accountable for environmental injustice. I’m going to focus on the case of Mexico City’s vast urban drainage system.
Floodwaters in Mexico City are not exactly mysterious things. Anyone passing by can see - and often hear and even smell floodwaters, especially given that they are almost always contaminated with sewage and industrial waste in this metropolis of 22 million. For those unlucky enough to be caught in them, floodwaters can be felt and, if you fall in, even tasted. Floodwaters are, then, immediately available to the senses.
And yet, if we consider flooding as the emergence of water “out of place,” to borrow from Mary Douglas’ classic notion of “pollution,” its causes are often far less clear.[2] Floods in Mexico City often begin with water simply backing up into the street and homes of residents, particularly those living in marginalized areas of the urban periphery. Or in other cases, rivers and canals simply overflow with wastewater. But this water carries no indexical trace; nothing about the water in and of itself makes clear where it came from, or why.
I’ve written elsewhere, but briefly: through the manipulation of pumps and floodgates, engineers exercise tremendous control over where wastewaters flow under and through the metropolis, and therefore where, when, and how they emerge from the system as floodwaters (or how stormwater and sewage that would have simply drained away become floodwaters).
Nevertheless, while floodwaters can be sensed, the flows of these wastewaters are almost completely inaccessible to outside observation - to direct sensing by humans not employed by the water utility. This inaccessibility is itself infrastructural: much of the flow of wastewater is now channeled through vast underground tunnels redirected through floodgates and pump stations that are themselves ensconced in intimidatingly tall walls.
You cannot just wander over and peer down a shaft and drop a rope - as the city workers themselves do - to “feel” for the depth of water or see the position of floodgates. Residents are thus excluded from the ability to sense a huge part of the infrastructural system - they cannot deploy a straightforward community monitoring program, as is common in other environmental justice struggles around contaminated air or water.
This exclusion is not, I want to emphasize, necessarily fully intentional or malicious. But this structural exclusivity - whatever the motivation - has real political effects. It means that there are very real boundaries between what aspects of the flows of water in the system can be sensed - and therefore, on some level verified - by external observers and what aspects cannot. In essence, this creates what we might think of as a “sensory black box” for community members: they can sense what comes out (flooding) of the drainage system and have a vague sense of what goes in (rain - although quantitative, high spatial and temporal resolution precipitation data is also controlled by the state and is not always easy to access),[3] but they do not know what happens inside the system.
Why does this matter? Because without knowing the movements of water in the system - and, by extension, what operations engineers took - it is extremely difficult for victims to prove that a given flood was the result of deliberate state actions, rather than a natural disaster – and therefore, that they are owed some compensation from the state.
So what options are left for flood victims seeking justice? The most obvious is to demand access to the black box, to sense the floodwaters for yourself. But of course, there are spatial and temporal obstacles to directly sensing flooding after the fact: one can neither go back in time (to the crucial moments before a flood event) nor be everywhere at once.[4] And unlike other forms of “accidents” or disasters (such as the recent collapse of a portion of the Golden Line of the Mexico City Metro), there are no material traces left behind by the flow of wastewaters through the system. The engineers and technicians might let the residents through the gates to peer down into the tunnels and they would see nothing that would help them understand what happened in the hours or days prior; wastewater is constantly being flushed out of the system in order to prevent the city from returning to its watery origins.
Logging the Black Box
There is one alternative: flood victims could demand access to the internal logs (bitácoras) - the spreadsheets created by field operators and engineers in the water utility that purport to represent a faithful material inscription of water levels, pumping rates, and floodgate positions across the system over time. These are an attempt to overcome the very spatial and temporal obstacles I mentioned previously: they are continuously collected at regular intervals every hour of the year and are compiled in both field offices and engineers’ command centers.
These logs are of tremendous importance to engineers. In their command center and field offices, where I spent dozens of rainy nights during my fieldwork, these logs appear like the “immutable mobiles” of Latour’s imagination, stacks of documents compiled hour after hour and day after day into thick binders that appear to make these seemingly “insignificant people working only with papers and signs become the most powerful of all.”[5] Unsurprisingly, engineers frequently gloated about which field office had a more complete set of logs, or complained when logs did not arrive in time or in complete forms.
If these documents are so central to the power of engineers, then it would stand to reason that flood victims should seek to obtain them, and once obtained, they would - at least in theory - be able to understand why a given flood happened and even prove the state’s responsibility for flooding. There is certainly some truth to this. And indeed, the government does often closely guard these logs, often finding ways to deny them to those who request them through government transparency portals.
Nevertheless, the logs may not provide the smoking guns - clear evidence of discriminatory state actions or inactions - that activists so desire and need. Part of the problem is that the logs are themselves often unfaithful mediators. Let me explain why.
To begin with, the logs are based primarily on direct human observation of water levels through manual measurement tools, given the unreliability and exorbitant expense of digital sensors. (Commercially available water flow sensors simply are not designed to deal with the extraordinarily toxic atmospheres, gusts of gas, and violent surges water that prevail in the system.)
Thus, even when compiled in visually authoritative forms - like a smartphone app installed on the smartphone of the director of the water utility, the data still “depends on a damn guy [pinche güey] taking [manual] measurements,” as one engineer jokingly remarked. The engineer was referring to the preferred practice of measuring water levels with a stick suspended via a knotted rope, lowered slowly into the rushing water in the drainage tunnels below. As I learned from operators experienced in the technique, it was very much an individual art to figure out when the stick had touched the water, and therefore when to stop lowering the rope and count the knots - and thus measure the depth to the water below. This dependence on human sensing meant that substantial error was inevitable, and the numbers inscribed in the operator’s logbook - communicated via radio back to the command center - were best understood as crude estimates, rather than definitive measurements, a fact engineers were well aware of.
Worse still, engineers and operators alike admitted that they did not always log everything they did, nor do everything they logged. While occasionally this was chalked up to negligence by higher-ranking engineers (who were quite cognizant of this fact), it became clear that such subtle deviations had a strategic purpose. Whether a floodgate open a few more minutes than reported or a pump turned off a little sooner than planned, these deviations were based on the intimate tacit knowledge and sensorial information available to field operators and engineers who were trying desperately to avoid a flood in the particular segment of the system under their command or doing it as a favor to a friend elsewhere in the system.
Moreover, far from being the “immutable mobiles” Latour imagined, the logs are extraordinarily mutable.[6] As engineers themselves explained to me had occurred on numerous occasions, an operation understood in retrospect to be an error could be easily “erased” from the logs, even if that meant forcing an operator to rewrite (by hand!) the entire logbook. The logbook could then be presented, as was often done, as “definitive” evidence of the water utility’s correct and judicious operation of the system.
I was surprised by this: given that it is an interconnected hydraulic system, any such modification should be, in theory, detectable through a careful analysis of the data in a hydraulic model; something would not “add up.”[7] But engineers waved away my concern: the level of error inherent in the measurements to begin with meant there was already so much noise in the system’s data that a small change could likely go undetected.
Separating the Signal from the Noise
This all raises the question: if the logs are such (potentially) unfaithful inscriptions of water onto paper, then what purpose do they serve for engineers? Why do they make such a point of collecting, centralizing, and combining these numbers? They are clearly not useless.
My point is this: the logs are made into useful mediators through a host of contextual information - much of it sensorial in nature - that engineers gather alongside the logs themselves. In the command center, they are not simply sitting quietly and running calculations based on the spreadsheets they are given. Rather they are constantly calling, texting, and radioing with field engineers and operators, often asking them to go outside and tell them what they see, hear, feel (by touch), and smell. Field engineers will often be sent off in their trucks to make the rounds of different installations across their jurisdiction, confirming water levels, the positions of floodgates, and the operation of pumps. Field operators also call in with qualitative assessments of the rain, wind, and flooding on the streets. Some of this is logged, but most is not.
The engineers even have an entire fleet of motorcycle riders - known as the Unidad Tormenta (lit. “Storm Unit,” in an apparent military allusion).[8] Decked out in motorcycle armor branded with the water utility logo, one of the Unidad Tormenta riders sitting in the command center once told me that “we are their eyes,” referring to the engineers sitting around the wide conference table who would regularly check their phones for images sent from his colleagues in the field. He was waiting his turn to be dispatched to weave through traffic in the middle of rainstorms (not a safe job!) to visually verify whatever was of interest to the engineers in the city.[9] This could mean anything from taking a photo of a flooded intersection to wading into the dark floodwaters directly to feel for a blocked drain.
The contextual information conveyed by these human observers, with their deeply embodied sensing of the flows of water in the system are what allows engineers verify whether the numbers they are seeing in the logs “make sense.” Thinking with Michel Serres, these human observers are the “parasites” who stand in between engineers and the logs, allowing the engineers to separate the “signal” from the “noise” in the logs. Put another way, these sensorial, contextual clues are what make the logs meaningful, making communication - between the addresser (the flows of water in the system described in the logs) and the addressee (the engineer) possible.[10]
These sensorial clues, are of course, mediated not only by a vast network of humans but also another infrastructural apparatus of communication - phones, radios, and so on. But almost none of the information gathered in this way is inscribed in any durable form after the flood event passes. The messages over the radio and through the phone are ephemeral - even though texts and photos may have a digital life, they are left on personal devices (no one has official phones) that are not subject to archiving.
Being immersed in this sea of contextual clues from dinging phones, squawking radios, and fleeting images on the screen in the moment of a storm are what make it possible for engineers to use the logs effectively - if not infallibly - to operate the system. But it is precisely the impossibility of such immersion after the fact that makes understanding what happened in the system to cause a flood occurs so difficult for someone - such as a community activist - who only has the logs to go on. In and of themselves, separated from this vast, yet ephemeral sensorial apparatus, the logs open, rather than close interpretive possibilities.[11] They allow unfaithful inscriptions to live on unchallenged, frustrating even the attempts of engineers themselves to understand what happened on any given night after the fact.
Engineers often chastised me for asking what happened two or three weeks prior when a flood happened, claiming they could not remember - I needed to be there, amid the radio and telephone chatter of the command center, or of the numerous field installations, to understand what happened. And even then, there were some things they admitted they would never know and never understand.
Conclusion
What does this all mean? Are these logs essential documents one minute and useless piles of rubbish the next? Are they still worth demanding and analyzing? The short answer is yes, if only because there is no alternative. They are the only view inside the sensory black box that is available to activists. And there still might be something, if only partial, that could be recovered from them.
In the wake of a devastating recent flood in 2021, I partnered with local environmental activists and an investigative journalist to request a series of logs via the Mexican government’s transparency portal. They returned hundreds of pages of manually redacted logs, as shown below. I am still going through these, but I am not hopeful that we can discover much. And yet, there is a useful crumb of data in the redactions themselves: the government has something to hide. Identifying exactly what that is - and proving it - is a very different story, although I have certainly tried to piece together the evidence.[12]
Nevertheless, I am not convinced that the principal struggle is to uncover these moments of state violence or even to conduct citizen audits of the drainage operations. Instead, the long-term struggle is to find ways to radically reorganize the city’s infrastructure to make such inequitable distributions of environmental suffering structurally impossible, in other words, to abolish the infrastructures that make such discretional flooding both necessary and possible.
[1] Alejandro De Coss-Corzo, “The Sensorial Management of Water in Mexico City,” Roadsides, no. 6, accessed March 18, 2022, https://roadsides.net/decosscorzo-006/; Ashley Carse, Beyond the Big Ditch: Politics, Ecology, and Infrastructure at the Panama Canal (Cambridge, MA: The MIT Press, 2014).
[2] Mary Douglas, Purity and Danger: An Analysis of Concepts of Pollution and Taboo (London; New York: Routledge, 2015).
[3] Precipitation in Mexico City is highly uneven given the vast spatial extent and complex orography. Thus it is not enough to know it is raining somewhere, but rather to make any kind of meaningful assessment of total runoff, one must know the spatial and temporal patterns of the rain.
[4] One can similarly not know in advance that a flood will occur - although a constant citizen’s watch of every installation in the system - if an enormously costly and complex undertaking - might guarantee you’d be there when it happens.
[5] Bruno Latour, “Visualization and Cognition,” in Knowledge and Society Studies in the Sociology of Culture Past and Present, ed. H Kuklick, vol. 6, 1986, 30, http://hci.ucsd.edu/10/readings/Latour(1986).pdf.
[6] Latour, “Visualization and Cognition.”
[7] One of the simplest problems would ostensibly be that of continuity: large, unexplained jumps or drops in water levels, which should vary only gradually unless there is a floodgate closure or opening, or a pump turned on or off.
[8] One of the higher-level staff who oversaw the team even called himself Capitan Tormenta, “Captain Storm,” and was subsequently the frequent butt of engineers’ jokes.
[9] These sensorial accounts are still primarily ocular, but hardly exclusively. The smell of sewer gas is of great importance, indicating a potentially higher water level or a backup causing a lack of ventilation.
[10] Michel Serres, The Parasite (Baltimore: Johns Hopkins University Press, 1982); I am particular indebted to Bernhard Siegert’s interpretion, see Bernhard Siegert, Cultural Techniques: Grids, Filters, Doors, and Other Articulations of the Real (New York: Fordham University Press, 2015).
[11] On a similar note, see Annemarie Mol, The Body Multiple: Ontology in Medical Practice (Durham: Duke University Press, 2002).
[12] Dean Chahim, “La Tragedia de La Inundación En Tula Fue Una Decisión Política,” Washington Post, September 20, 2021, sec. Opinion, https://www.washingtonpost.com/es/post-opinion/2021/09/20/tula-inundaciones-rio-causas-hidalgo-mexico/.