How to Connect Sensors

IT COULD SEEM TO BE straightforward enough to buy a sensor, plug it in, and begin monitoring. But the situation is rarely as simple as that. The process of working with sensors includes struggles with upgrades, wonderings at data outputs on platforms, trials to test sensors in situ, and uneven comparisons with regulatory monitors. Through many practical tests of off-the-shelf citizen-sensing technologies undertaken as part of the Citizen Sense research project, it became apparent that as a user of these devices, one also became enrolled in contributing to their ongoing development by getting them to function and by contributing to (online) communities providing mutual instructions and tips for troubleshooting.

I should preface this more detailed account of how to make things with sensors by confessing that I am not, of all things, an inveterate maker. I have attempted to cobble together windmills and model cities, jelly rolls and button-down shirts, with each object bearing the sad signs of an absent hand–eye coordination. While others might have tinkered together amateur radio sets or carved out three-legged stools, for me the world of “making” has been—and remains—an ongoing but workable challenge to align bodies, forms, and functions. This how-to guide does not unfold as the advice of a seasoned expert to an audience of eager trainees (which itself is a highly gendered way in which the digital world churns as an ongoing performance of master-y). Rather, it is an account of dogged persistence and muddling along, of the just good enough and the bang-it-together, of the electrical tape and the makeshift arrangement. Luckily, what the hand does not or cannot make in all its supposed authenticity the computer can readily press out through a bit of code, CAD, and 3-D printing, along with numerous advice boards, collaborations, and conversations. In this sense, I engage with the world of DIY sensing as one is meant to: as an amateur connected to extended communities of practice. Indeed, it is exactly the DIY aspects of craft, making, and tinkering that can generate different experiences of embodiment, the everyday, and collective politics.[1]

In this way, I work through these faltering processes of getting sensors up and running in a citizen-sensing research group. By “up and running,” I am referring not just to making photoresistors blink and microcontrollers talk but also to the extended sociopolitical and environmental relations, from communities organizing to address pollution to participatory research practices that recast the usual contours of inquiry. But this account is still not a tale of salvation through making—digital or otherwise. Instead, it is a faltering if candid encounter with the promises of DIY. It questions the processes of making kits that are meant to empower while toppling prevailing power structures. By taking up tools—specifically, DIY sensors for monitoring environments—it is possible to describe more closely how kits are made. These practices are situated ways in which to understand better the call to “hands-on” action that is meant to be a remedy for contemporary malaise. This approach involves looking at the work-arounds used to make kit operational, whether for hobbyism or environmental activism. It also describes how these technologies enable particular political engagements and ways of being and becoming citizens. Along the way, this work bypasses Heideggerian hammers to rethink and rework what counts as making along the lines of what Elizabeth Povinelli has suggested can involve a probing of differential ways of being in the world.[2]

These accounts are a small selection of the many sensors reviewed, tested, and built. They work through the how-to, document experiences with following instructions, and elaborate on the potentials and pitfalls of working with citizen-sensing technologies. While these sensors were collaboratively assembled as part of a practice-based research process, it should also be said that some of these kits were tested and assembled in relation to pressing public events and participatory workshops in planning. In some cases, this testing and assembly process involved flying somewhat blindly into the world of sensors. These accounts re-create the steps of testing these devices after having fumbled through making and setup. In the process, I also argue that there is much to be said for using tool kits inappropriately and incorrectly. This is less a condition of embracing “error” as such and more a way of seeking out the practices that proliferate on the edges of straightforward instructions. As many feminist writers have noted in relation to their tool kits and survival guides, it is through these processes that other purportedly illegitimate or queer ways of being in the world are also forged or claimed.[3]

Making in the Imperative Mood

There are numerous guides for working with sensors, including such texts as Getting Started with Sensors and Environmental Monitoring with Arduino, among many other online manuals and tutorials. These texts could seem to be a good place to start, as they set out instructions to follow, sensors to make, and ways of toggling across making and essential concepts. The process of following one of these texts, however, yields unexpected processes of making, inquiry, and engagement. If you begin your voyage with sensors in this way, you will notice certain abiding themes for assembling technologies. Also of note with the how-to guide is that the you/your of the instructional refers to a maker who is brought into a technical relation. As I follow this instructional, I similarly traverse from first to second person to inhabit the you/your of the instructional and the imperative mood, and to work through the practice-based and participatory aspects of setting up DIY sensor technologies. Let’s begin.

See, connect, attach. Insert, double-click. Orient, insert, connect, push, fold, twist, insert. Grammarians would parse this as the imperative mood. Action verbs and commands distinctly characterize the how-to of DIY electronics. Words order action. Language exhorts. Do this and complete that. Command equals outcome. Make yourself into a model citizen, and a citizen able to make models. Along the way, there will be helpful tips and conversational sidebars to review key milestones and to assure you that the makerly relationship is more friendly than authoritarian as you progress toward your goal.

The how-to guide, then, is a genre of sorts, not only in the stylistic sense but also as a way of organizing anticipation. As Lauren Berlant has noted, “genres provide an affective expectation of watching something unfold, whether that thing is in life or in art.”[4] What is expected in the how-to guide, and how is it meant to unfold? One could say it is a genre of problem solving. Problems are identified that can be addressed through learning and sharing skills and procedures. Yet the very act of constituting problems is also a way of constituting worlds.[5] To identify the making of sensors, the writing of code, and the collection of data as key ways to skill up and undertake environmental monitoring is to commit to and become informed by a particular way of acting on the problem of pollution. The how-to guide is meant to make acquiring and executing these practices more do-able. It also creates practices that attach citizen makers and citizen sensors to distinct ways of engaging with worlds.

In the worlds of citizen sensing, instructions are found in the form of how-to guides for making kits from assorted electronic peripherals. They also assemble as step-by-step directions for setting up and plugging in an off-the-shelf sensor to an online platform in order to view data. There are instructions for following monitoring protocols, instructions for calibration, and instructions for installing sensors in polluted locations. In the O’Reilly published text Getting Started with Sensors, which I detail here as the first example of attempting to work with sensors, hypothetical maker-readers begin the project of assembling sensors, while also “bending technology to [their] will” to control computation and environments.[6] Here the instructional promises a curious mastery of technology, which, as one quickly discovers, can be a bit misleading.

According to this basic primer, the process of getting started with sensors first involves asking “what are sensors?” The text readily provides the answer that a sensor is an electrical input device that “evaluate[s] a particular stimulus within the environment.”[7] It would seem that any change or disturbance in an environment could be detected and transmitted into digital form. Sounding in a Simondonian register, the authors describe this as a process of “Transduction!” to explain how sensors control circuits and, by extension, environments.[8] Here environmental phenomena are undergoing conversions into electrical and digital outputs. The how prevails over the why—technology makes its own logic of execution.

Yet, when following instructions in a guide and tool kit such as this one, you will find that time for reflection is often cut short, since the point is to get on and make things. After asking the maker-to-be to reflect on what sensors are or might be, the authors of this instructional abruptly rejoin, “Enough discussion—it’s time to build!”[9] And so you will be off, testing batteries and breadboards and LEDs, switches and alarms. There are many sensors to build here, including infrared proximity sensors, rotation sensors, photoresistors, pressure sensors, temperature sensors, and ultrasonic sensors. It will take some time to get to the finer points of connecting up air quality sensors, which are not even covered in this basic text. However, as you work through the configurations of these many sensors, you will find that it is one thing to generate a reading from a temperature sensor and quite another to know whether the sensor is providing a verifiable measurement. In the course of setting up temperature sensors in our work space on the tenth floor of an aging office building without air conditioning, for instance, we found that according to our temperature sensors, indoor summer temperatures instantly leaped to 40 degrees Celsius. Was this due to the electrical wiring of our sensor circuit, was it due to the placement of our sensor near a window on the sunny side of the building, or were we really just about to perish from heat exhaustion? Sensors at this stage of assembly can give rise to extensive questions about the state of the surrounding environment.

The temperature sensor test that this guidebook provides is to move the sensor in and out of the refrigerator, placing it in room temperature and then cooling it down in an appliance. But if you also want to speed up the process, you can expose the sensor to ice cubes, the text suggests, to obtain a more instant result. The introduction of a stimulus is often used to see whether a sensor is reading. An air pollution sensor can be exposed to a lit match, cigarette, or vacuum cleaner to see a spike or dip in the data. The first stage of connecting a sensor, then, often involves working through these processes of setting up the sensor circuit configuration, loading a bit of copy-and-paste code to a microcontroller, and then testing whether the sensor detects the introduction of basic stimuli by generating detectable changes in the data.

Many of the kits that allow you to get started in testing sensors in this way are now available as defined maker kits with all the necessary parts to develop a basic plant watering system, gas sensor, or temperature sensor. Companies such as Seeed Studio sell an array of such kits that fit within the language of other assembly-based hobbies, such as model ships and airplanes. Parts to be assembled are included along with instructions, and the process of making is meant to generate new understandings of technology through doing. DIY practices on one level are meant to “challenge traditional hierarchies of authority and the existing status quo,” as Matthew Ratto and Megan Boler suggest, by decentering the usual sites and practices of making.[10] Yet DIY can also reinforce particular ways of engaging with technology, for instance, as a project of following instructions to bend technology to one’s will, or in other words, to gain technical mastery and to work on a universal if abstract problem. Mastery as a project has come under fire not just by philosophers of technology, such as Simondon, but also by postcolonial and decolonial thinkers, such as Julietta Singh, who suggests that a project of “unthinking mastery” can be a way to undo the estrangement that comes with these forms of relation.[11]

DIY practices can tend to be at once open-ended and instructional. What I am calling the imperative mood in this context is inevitably related to that better known theory developed by J. L. Austin of the “performative mood” in his study How to Do Things with Words.[12] The performative mood is a way of mobilizing actions, relations, and worlds through speech acts. Austin was interested in studying ways of doing things with words by considering the infelicities, misfires, and miscalculations that occur within the performative mood. His theory has in turn influenced thinkers like Judith Butler, who has further investigated how social constructs like gender are performed and materialized.[13] Karen Barad draws on and reworks Butler’s discussion of performativity by adding a material and posthuman dimension that shifts discursive statements to a field of multiagential possibilities (rather than an exclusively human utterance or action). In this way, performativity is less about language in abstraction and more about what Barad calls the “conditions of mattering.”[14]

Theorists of digital technologies and digital citizenship have also built on these theories of performativity to analyze how digital practices can constitute ways of performing digital citizenship.[15] While there is much more to say about performativity than space here allows, this proposal for an imperative mood attempts to work with and alongside this constructivist approach to words, actions, and materialities to investigate how such a mood might generate its own distinct configuration of instructions, relations, practices, technics, and milieus. Even more than drawing out the generative aspects that characterize theories of performativity, I would suggest that the misfires and miscalculations just as readily (if differently) come to the fore when engaging in the imperative mood. These misfires could also be a particular entry point into understanding how open-air instrumentalisms take hold, as swerving experiments with instruments. An instruction to insert a lead of a resistor into the same pin of a breadboard as an LED or sensor can easily end up back to front. A sensor-wire-battery configuration might connect up or become disconnected. A sensor output might be all but inexplicable and difficult to verify. Hence every list of imperative commands comes with its inevitable section on “Troubleshooting” to help makers figure out what has gone wrong along the way. But this process plays out in much more elaborate ways than simply faulty wiring or misaligned sensors. It can also extend to dodgy code and incorrect conversions as well as data platform errors and bungled sensor housing. Misfires and miscalculations can and often do extend to botched sensor installations, inscrutable data output, and indifferent responses to data gathered.

Many theorists have discussed the ways in which instruments generate more-than-descriptive engagements that enact worlds. In other words, instruments are world-making. They are constructive and performative of the worlds that they would detect, measure, and act upon.[16] But the imperative mood designates explicit actions along with observations that might be achieved. It constitutes the methods by which such constructions and performativity take place, or falter. The imperative mood constitutes the conditions, subjects, and environments in and through which a particular project is meant to occur and a particular outcome to be achieved. It tends to be normative in its register of address. When you load code onto a microcontroller, there is little sense that there are multiple ways of completing this task. Instead, you pursue the project in the seemingly correct way, which you are meant to master to move on to the next step. This configuration of technology, maker-citizen-subject, technical relations, and practical action is what the imperative mood designates.

This initial example of working with sensors vis-à-vis a standard maker text such as an O’Reilly guidebook is meant to demonstrate the particular entry point and process whereby sensor instruction occurs. Sensor ontologies quickly give way to flat-pack cosmologies, where component parts join up to create electrical arrangements of action and reaction. The progression through a guidebook, and formation of your own tool kit, can be delineated as the working through of devices: from LEDs on to temperature and pressure sensors. In other words, guidebooks can tend to direct makers to a particular notion of technical mastery without considering the open air where devices would not only be installed but also shapeshift in their broader arrangements. When making citizen-sensing tool kits and following instructions for assembling sensors, it is important to consider how these instructions are organizing a particular way of encountering the problem of monitoring environments as well as establishing technopolitical relations.

Simply Connect

Once you’ve attempted to assemble sensors from their basic component parts, you might next decide to test a sensor that is off-the-shelf and does not require an intricate process of assembly with breadboards and jumper cables. As noted, an increasing array of sensor objects and off-the-shelf products can be procured through Kickstarter pledges and online shopping. The Air Quality Egg, which I detail here as the second example of attempting to work with sensors, is perhaps one of the most iconic of these citizen-sensing devices. Having been prototyped through a series of hacker events from 2010 to 2012 (which I describe in Program Earth[17]), the Egg moved from prototype to saleable product by 2013, when the Citizen Sense research project was under way. Because the Egg was an air pollution sensor that could be purchased as a complete product, it did not require soldering or microcontroller setup. Such an off-the-shelf sensor would seem to allow for more attention to be given to environmental monitoring, data collection, and public engagement, and so our research group began to investigate the capacities of this device.

We placed an order for our own Egg in the middle of July 2013. The sensor arrived at our offices in London in late July, sent from Wicked Device, based in Ithaca, New York. The neatly packaged device promised on its outer label to make more engaged citizens of us all. It declared,

Problem Solved. Do you ever think about the air you breathe? It affects us in ways we can see and also in ways we can’t. The Air Quality Egg is a project working to make the air we breathe more “visible.” Simply hang it in your home, office or outside your window to start collecting your personal air quality data. The Air Quality Egg connects you with a global community of concerned citizens participating in the ongoing air quality conversation.

The strangely daunting prospect of simply plugging in and connecting to a global community of concerned citizens able to solve an environmental problem as intractable as air pollution meant the Egg actually sat on our shelf for another week. Opening the kit seemed to be a ceremonial event for which we best waited and prepared.

So when, in early August, we set aside time to unbox and install the Egg, we made a wager as to how much time would pass before we were successfully gathering air quality sensor data. Estimates spanned from having the device up and running within the afternoon to a few days. A more skeptical researcher estimated that it could take months, if ever, until sensor data were coursing through this plastic Egg and transforming into environmental solutions. You might find yourself engaged in such speculation as to what the final setup and output from your off-the-shelf sensors could be. This process is central to the ways in which citizen-sensing practices take form as ongoing contingencies of devices, environments, and engagement.

Once we had unpackaged the Air Quality Egg and scanned the different components of the kit, we next read through the seemingly straightforward how-to instructions. We began the setup by entering the device’s serial number (or MAC address) into the Air Quality Egg Google map platform, where we also located and named our Egg.[18] With this, we were able to see the Citizen Sense Egg in Southeast London, situated within a wider and global community of sensing citizens, albeit one that numbered around 250 in population worldwide. Here was an apparently eager if niche community of Egg owners, ready to solve the problem of global air pollution. Once our Egg was on the map, we turned to setting up the device and posting data. Yet you might find, as did we, that putting your sensor on a map is often the most basic and straightforward of the technical challenges that you will encounter with the Egg.

The Air Quality Egg is formed of a pair of translucent white plastic Eggs: a “base” station Egg that at the time of testing posted data to the Xively platform and a “remote” Nanode Egg that does the job of sensing air pollutants and gathering data via a shield outfitted with metal oxide nitrogen dioxide and carbon monoxide sensors and with temperature and humidity sensors attached to an Arduino microcontroller. Although there have been updates and a new version of the Egg since the time of this testing, with data now posted to a different platform and updates made in sensor setup, this was the device arrangement with which we worked at the time. If you refer to points 5 and 7 above, you’ll find that the Air Quality Egg presents numerous dilemmas in the form of upgrades and fixes. These events are not exceptional to the Egg, because not only are these relatively new and unstable devices but also they inevitably succumb to the rapid rates of obsolescence that are characteristic of electronics.

During Egg setup, we next discovered that the power plug was configured for U.S. electrics. As we did not have a power converter, we had to set the project aside until we sourced an adapter for the U.K. context. This was a simple enough problem, but we found that this was just the beginning of several stages at which we realized additional kit would be needed to make the Egg function. Once we eventually powered the device, we found that it did not perform the correct color sequence to indicate that it was collecting and posting data. Flashing color sequences were the means by which we were to be made aware of air pollution through the successful posting of environmental data to the platform. But the sequence of our flashing colors did not follow the same sequence outlined in the setup instructions. We spent some time trying to determine what the color conversions were indicating, exactly, when we discovered that the color issues were due to a bug, discovered several months earlier in January 2013 but still impacting devices shipped as late as ours, where the Eggs were no longer talking to the Xively platform for displaying data. An elaborate process then ensued of attempting to reprogram the Egg base and remote Nanode, following the official Air Quality Egg Google forum and FAQ instructions, which linked us to a seemingly straightforward video indicating how to fork a repository of code from GitHub to reprogram the Arduino microcontroller in the Air Quality Egg.

And yet, after completing the process of reprogramming, the Eggs were still not producing “data,” neither in the form of flashing color leading to awareness nor in the form of line graphs on Xively or cryptic bar charts on the Google Maps Air Quality Egg page. Eventually, through multiple waves of turning the device on and off and reloading and verifying code, we managed to obtain blips of data, strange right angles in line graphs, and numbers apparently in parts per billion of nitrogen dioxide or carbon monoxide but generally remaining inexplicable in what measurement they presented, exactly. We had a modest assurance that the temperature and humidity readings might be somewhat correct, because these agreed with other sensors we had in operation in the same space, but converting the nitrogen dioxide and carbon monoxide measurements to a legible figure was a rather more difficult matter, as we did not have other verifiable sensors in operation for comparison.[19] Our closest point of reference was with the official London Air Quality Network station, and here the measurements were not in units that could be easily cross-referenced.

By getting the Egg up and running, we had experienced what one of our developer-collaborators called the “‘Hello World’ of IoT.” Getting an air pollution sensor setup to post data was a basic achievement along the pathway of the Internet of Things. Yet, in the process of connecting the Egg, we were more intent on asking this IoT “world” a whole host of other questions about how these devices were sensing air pollutants. What were these sensors sensing, exactly? How could we find out more about the hardware and software setup and the extent to which this could influence the data outputs? How did the “color-equals-awareness” engagement with environmental data work, exactly, especially when color did not signal pollution levels?[20] At the same time, what sorts of data were these, in terms of their accuracy, legibility, and legitimacy, when even getting a device to work, whatever the readings, seemed to be an achievement? And now that our device was operational, what were the capacities of the network of concerned citizens to which we were connected? From our use of the Air Quality Egg forums, it seemed that the communities we connected to had more interest in hobby electronics and computer tinkering than mobilizing their data to influence air quality policy or enforcement.[21] Indeed, the Egg was now circulating in the world in ways where maker communities effectively contributed free R&D by testing the device, while finding fixes and improvements through necessary troubleshooting.[22]

The universal citizen sensor, embedded in a community of global citizens, then comes down to earth by having to work with the specificities of particular infrastructural configurations. The seamless plug-and-play logic and practices that such devices would promise continually meet with simple obstacles and more complex malfunctions. The imperative mood here might instruct you to plug in your Air Quality Egg and connect to other global and instrumental citizens. But your inability to complete the command or follow the instructions can multiply into a whole set of other practices, infrastructures, and relations that you will likely need to call on to fulfill seemingly straightforward instructions. In a similar way, Lucy Suchman has pointed out that even a task as obvious as pushing a green button on a photocopier machine can give rise to confusion, adjustment practices, technical communities, and modified technological artifacts that materialize when simple button-pushing does not yield the expected results.[23] Instructions seem to guide a technical encounter, but they do not determine it. Instead, the “sense” made of and through instructions materializes through the actual undertaking of a technical practice.[24] Practice-based research in this way is an approach that surfaces the many adjustments and deviations that can arise when working with technologies in lived situations. Instruments and instrumentalities frequently deviate from simple action and outcome. These are the misfires of the imperative mood, which generate the open-air instrumentalisms of citizen-sensing technologies.

You might then find that off-the-shelf sensors promise that you can “simply connect”—and by extension also connect to greater air pollution awareness and a community of global and instrumental citizens—and yet they do not unfold in such a straightforward or liberatory fashion. Instead, they generate open-air instrumentalisms that deviate from the process of following instructions and setting up tool kits. Air quality sensors do not always immediately function, either technically or in terms of their broader sociopolitical and environmental effects, but in the malfunction of devices and the reconstitution of instructions, other worlds-in-the-making are generated along with technological subjects. The misfires that percolate through the imperative mood occur in part through anomalous technical arrangements that come into being, as detailed here, and in part through the ways in which these devices circulate and are taken up to address environmental problems. While these sensors proved to be anything but off-the-shelf, there is the possibility (if not the danger) that the promise of such modular and ready-to-use devices also could begin to inhabit the space of politics, encounters, and relations, for instance, in the form of off-the-shelf politics, off-the-shelf citizenship, and off-the-shelf public engagement. This is why it is important to ask what sorts of instrumentalisms are mobilized with apparently ready-made sensing technologies and sensing practices.

At the time of this writing in 2019, the Air Quality Egg has undergone many updates and is now available for sale in newer versions.[25] The website notes that “big improvements” have been made over the version 1 Egg that we had tested. Indeed, its supplier, Wicked Device, no longer supports version 1, and Xively, the platform host, has terminated the data service that the version 1 Egg used. To post data, the Egg would need to be reconfigured to post to a different data platform. As the Wicked Device announcement states about this option, “That’s a fair amount of work, and will require that you recompile and re-load your Egg with the new service destination.”[26] The easier route is to purchase the version 2 Air Quality Egg, which guarantees an even more seamless plug-and-play experience. A new device replaces the defunct one, and the promise of technical action becoming democratic action is refreshed.

Whereas the Air Quality Egg on one level seems to promise that air quality monitoring can become a relatively effortless affair, many plug-and-play sensors require considerable effort to become operational. Updated and upgraded versions will still require ongoing maintenance and fixes as well as skilling up to learn about technical configurations. Even with all of this effort, there are still many more questions about the verifiability of the data that plug-and-play devices generate as well as the protocols and practices that are used when monitoring environments. While this is in no way meant to deter you from testing out citizen-sensing technologies, this how-to setup that involves working across standard instructions as well as actual practices undertaken is meant to demonstrate the misfires and miscalculations that proliferate when inhabiting the imperative mood and when working with the genre of the how-to and the tool kit.

Around the time we had undertaken our own provisional setup with the Egg, I began to be asked by representatives from local governments, environmental NGOs, and even air quality officers in small nations whether they could replace their expensive air quality monitoring instruments and networks with Eggs. My cautious reply was, not unless you’d like to spend considerable time dealing with misfires and miscalculations. While many of these off-the-shelf devices can be and are used in interesting ways that add to the scope of DIY practices, they are tetchy gadgets that produce a variable range of data that currently do not transfer well to the spaces of air quality regulation. While a modest achievement can be made in getting a sensor device such as an Egg up and running, its flickering displays and data outputs do not necessarily sync well with the expanded technical, social, political, and environmental requirements of air quality governance in its usual sense. For this, you might need to engage with even more versions of the how-to, including points 7 to 10 given earlier, which indicate how the technoscientific configuration of an air quality sensor and the data it generates depend upon extended infrastructures to make sense.[27] These are infrastructures not just of technical capability but also of stabilizing data-as-evidence to address the experience and event of air pollution.

The Reluctant Prototype

Parallel to, and perhaps even in advance of, working with the Air Quality Egg, the Citizen Sense research group was in the process of developing prototype air quality sensing kits of our own making, which I detail here as the third example of attempting to work with sensors. These were provisional groupings of nitrogen dioxide and particulate matter sensors, which we had used in a pilot walk in the New Cross area in London in early July of that same year.[28] You might find that as you progress from following O’Reilly and Instructable tutorials, your own devices assemble neither as makerly stuff nor as off-the-shelf kit but as particular prototypes that are cobbled together in a cut-and-paste and makeshift way. A sensor configuration that works in one setup can be morphed over to another expanded kit, and code passes along on these various iterations or is drawn from libraries to create a new workable concoction.

In just this way, we were attempting to put together a possible prototype citizen-sensing kit that might be used as a tool of engagement while undertaking fieldwork in the United States, as we were researching fracking in Pennsylvania.[29] Multiple citizen-monitoring activities had already emerged to address pollution and public health concerns taking place in Pennsylvania. In this context, we wondered what role a prototype kit could play in engaging people to ask questions about environmental data, how the data are generated, and their effects in addressing the problem of air pollution. How might data be shared and collectivized? How might they travel differently than the current if complex modes of reporting on well locations and pollution levels? In the process of building a kit comprising multiple air quality sensors, including nitrogen dioxide, carbon monoxide, temperature, humidity, and particulate matter 2.5 sensors, we found that as many questions were raised about the validity of data that might be generated from such a kit and the possible promises and expectations that could be raised by circulating environmental technologies to communities affected by pollution from fracking. We were beginning to engage with sensors in the open air, not just by moving them to actual sites of pollution detection, but also by collaboratively testing them with communities knowledgeable about documenting pollution through environmental data collection and analysis. In this context, open-air instrumentalisms multiplied and abounded even further.

This tool kit in the making then demanded that we think through the instructions and how-to pointers that we might provide so as to make the kit legible and usable. Questions that came up when thinking about how these kits might be used in the field included, Would there be a manual with instructions for use? Would each sensor be explained in relation to what it senses and how chemical detection optimally works? How long would the sensors need to operate to collect usable data? How long should the sample rates and duration of monitoring be? Would the sensors work only if stationary, and should instructions indicate to keep the kit stable during use? If we are recording data, would these be made available to individual participants, or would they be shared collectively on a web platform? Should these data be given locational information or be made anonymous? What instructions might participants need to analyze the data in order for them to be meaningful? Would the sensor housing skew the readings in any way? What are the base readings of the sensors, and are we sure they are properly calibrated? Would the sensors or pollutants interfere with each other? Could we be sure they are sensing exactly what they are meant to sense? Would the kit be damaged in shipping from London to Pennsylvania, and what adjustments might need to be made in the field?

These questions were also informed by an attempt to ensure that multiple participants’ engagement with the kit might be collaborative and experimental from the beginning and not only a functional end application. At this point in the development of sensors, we queried the notion that by collecting data—accurate, skewed, or otherwise—environmental politics would be more readily democratized or facilitated. This was an attempt to critically engage through practice with a seemingly more instrumental–functionalist agenda. Yet while we sought to critically examine the role of environmental-monitoring technologies in forming practices and politics, rather than simply becoming advocates of this approach, we found that we also had to take seriously the instrumental logics of these devices and the citizen-sensing practices they activated and organized. These reworkings of instrumentality became part of the way we experimented with making alternative citizen-sensing tool kits that could engage with the practices and concerns of participants (who were engaged in preexisting monitoring projects), and generate usable environmental data, but that would also open into other engagements with environmental problems.

So, with all of this in mind, in autumn 2013, we began the process of making prototypes to test the ways in which sensors generate, influence, and operationalize environmental data. Our version 1 Citizen Sense Kit initially consisted of two primary devices, one a sensor shield pulled from the Air Quality Egg, which included nitrogen dioxide, carbon monoxide, temperature, and humidity sensors that we attached to a combination of a Grove Board to add a real-time clock and a Raspberry Pi microcontroller. The version 2 Citizen Sense Kit comprised stand-alone sensors (rather than a pluggable sensor shield), including carbon monoxide, nitrogen dioxide, particulate matter 2.5, temperature, and humidity sensors, where we added our own resistance configurations that we found considerably affected the readings in comparison to the Egg shield.

With both of the preliminary Citizen Sense Kits, the first intention was first to get the devices up and running, because in the process of making the kit, even more questions emerged about the how of the how-to. The second intention was to disassemble and reverse-engineer more black-boxed technologies, such as the Egg, which on one level required all sorts of capacities and resources to make function and on another level had rather unclear information about how the hardware and software were put together, how the sensors were configured, what resistance was used and how this affected data outputs, and how continual changes of the data platform “back end” (from Pachube to Cosmo to Xively) could affect the data’s form and analysis. You might find yourself asking similar questions if you attempt (or have attempted) to make sensor tool kits using devices sourced and assembled through multiple configurations. Repurposing and questioning become key techniques in assembling these kits.

In this way, the Citizen Sense Kits were built through information from multiple forums, as there was no official single forum from which we might obtain guidance on how to make the monitoring technology “work.” We developed even more iterations of the Citizen Sense Kit, including a version 3 that included a Speck particulate matter sensor, an analog BTEX (or benzene, toluene, ethylbenzene, and xylene badge), a Frackbox (for monitoring volatile organic compounds and nitrogen oxides), a monitoring logbook, and a data platform. This version of the kit was later used to monitor pollutants from fracking at thirty community locations from autumn 2014 to early summer 2015.[30] Along the way, we foraged for diagrams and work-arounds; forked code from GitHub; and shoveled piles of breadboards, resistors, and cables across desktops. By rebuilding kits, and gakining another perspective of the hardware and sensor configurations, we were also able to observe along the way what technical resources, capacities, and infrastructures these technologies require as well as the decisions that were made or elided to make the monitoring kits in these particular ways, and the domains inhabited to generate and circulate environmental data through these contraptions.

By working with air pollution–sensing tool kits, we tested how environmental sensing technologies enable certain types of monitoring and generate questions about the limits and possibilities of each of these monitoring practices for addressing environmental and political problems. The practice of making prototypes and setting up off-the-shelf sensors becomes a way to work through the instructions, promises, functions, and malfunctions of these devices. It also generates open-air instrumentalisms. In the process of procuring air quality sensor guides, making kits, following instructions, and installing devices in the open air, a number of splintering pathways came into view by deviating from a straightforward approach to these devices. Online forums read as tales of ongoing struggles to set up sensors, to maintain their operations, and to update and adjust when upgrades are available. FAQ sections are brimming with queries about connections, data, and modifications. Platforms bear the traces of half-finished efforts in running sensors, where maps of monitoring locations click out to nonexistent line graphs or inexplicable charts. These open-air instrumentalisms began to take on a more-than-technical quality as sensors were readied for installation and use, where an initial success at connection splintered into multiple concerns about the use or relevance of these devices.

But this is not to say that devices never arrive at a condition of organized use or implementation. Instead, it is to signal the ways in which setting up citizen-sensing technologies is an ongoing trial, a back-and-forth effort of testing and tweaking. At the same time, despite the democratic selling points, many devices remain tied to practices focused primarily on technology and “making” and so can become somewhat self-referential in their pursuits, thereby missing the promise to address—and even improve—environmental problems. Yet if, as Dewey has suggested, the “invention of new agencies and instruments create[s] new ends,” then how do these new instruments “create new consequences” and “stir” us to “form new purposes?”[31] This is a question about instruments and instrumentality, which the next section considers in a more reflective key.