Sensory Labor as Un/knowing in a Waste Composition Study: Identifying a Chain of Translation

Collecting samples: Translating loads of waste into as-reliable-as-possible samples

The first phase in the chain of translation we identified was sample collection. In this phase, a load of waste collected from different neighbourhoods by a waste truck needed to be translated into as-reliable-as-possible samples from which the average composition of household waste could be concluded. The shift from a load of waste into a sample was the first transformation (see Latour, 1999) the waste went through during the composition study.

First, the waste truck deposited a 1,000-kilogram load of household waste outside the tent in which the sorting part of the composition study was conducted. This load was spread evenly across the ground so that we were able to identify and collect individual waste bags. First, large and heavy items, such as tires, mattresses, wooden furniture, metal items and carpets, were taken from the waste load and weighed separately, to avoid skewing the distribution of different materials. After this, we had to collect a 100-kilogram sample in a sample container, which was later brought inside the sorting tent. The container was first placed on a pallet truck equipped with a scale to weigh the sample. After that, the collection of the sample began. The waste bags had to be picked out from all around the area to ensure that the sample included bags from different households. We also had to be careful to collect more fine-grained waste particles, such as cat litter or soil, that were not in the waste bags, using a shovel to move them to the sample. Usually, there were two to four people collecting the sample, while others continued sorting previous samples in the tent or tidied up the area surrounding the sorting tables (Figure 1).

OPEN IN VIEWER

Figure 1.

A 100-kg sample of waste was gathered by shovelling waste and handpicking waste bags and items into a container.

To conduct the labor of translating the waste load into a sample that could later be sorted and measured, we used multiple senses. We had to use our vision to collect waste bags of different colors and sizes for the sample: Waste bags of the same color may have come from the same households. We were also often able to superficially assess what the bags contained by looking inside or assessing the shape of the bag. For example, we had to avoid collecting too many bags of garden apples, as apples are quite heavy. In addition to using vision when collecting the bags, the feeling of heaviness or lightness was crucial in collecting the sample. There were many ways to feel the waste bags: When we collected the sample, we walked among the bags and occasionally stepped on them, stuck them with our shovels or kicked them, which enabled an embodied assessment of the type of stuff the bag potentially contained.

In this sense, the sensory labor (Spackman & Lahne, 2019) of categorizing the items for the sample was conducted through co-configuration of the senses—vision operated together with tactile sensations of the heaviness or lightness of bags and other embodied assessments of the contents of each waste bag. Thus, in different phases during the chain of translation, different senses formed a system of perception that enabled the categorization of waste (both in the sense of recognizing waste objects from each other and choosing suitable objects for the sample).

Although the categorization of objects suitable for the sample happened through the co-configuration of vision and tactile sensing, there were also some general technical tips, provided by employees of the waste management company. We were instructed to divide the waste load into different sections by creating a visual ‘grid’ in our heads, and then to make sure that we took something from each section of the grid. When the sample started to come close to 100 kilograms, we began to yell out the scale numbers to each other, and this also influenced whether heavier or lighter bags were picked up from the load and added. Moreover, when collecting large items from the load, we negotiated over issues such as where to place the different items and whether certain items were large or heavy enough to be taken from the load.

While we were guided through the procedure of collecting the sample, the process varied each time, due to the heterogeneity of the waste material. This complicated the process of categorizing objects suitable for the sample, and also made the sensory labor more difficult, as we had to always re-orient the system of sensory perception to recognize different objects. The contents of each waste load were different, and even if the loads were equally heavy, they looked (and sometimes smelled) different. As was noted several times by multiple people during the composition study, handpicking waste for a 100-kg sample introduces the possibility of aberration.

I still think that creating representative samples is super hard with this type of stuff. If you just try to take regular waste bags, it isn’t representative. Because, for example, some textiles may be thrown out in their own bags if you clean out a wardrobe or something. So, there is probably quite a lot of variation in the results because of that. (Car ride discussion, 18 August 2023)

As the composition study proceeded through the weeks, our embodied, sensory knowledge about waste also started to affect our practices of collecting the sample. We came to understand that it was good to collect both heavy and light waste bags for the sample, but at the same time, we learned that the heavier bags that both looked and smelled damp—and felt that way when poked with the shovel—often contained something disgusting, such as diapers or rotten organic residue. Moreover, the colors of the waste bags started to give some information about their contents; waste bags from bathrooms were often pink, purple, or blue, while waste bags from kitchens were often reused plastic bags with grocery store logos. One of us recalled how picking up certain waste bags started to cause visceral effects of disgust and repulsion and how this created an internal negotiation: ‘Now that I touched it, I need to take it.’ This had to be done to avoid allowing these bodily reactions to ‘disturb’ the sample collection.

Thus, the embodied knowledge acquired through sensory labor sometimes created a temptation to avoid certain waste bags when collecting the sample, which could have shaped the chain of translation (see also Hardahl et al., 2019; Latour, 2004), and certainly created tensions. Embodied and other ways of knowing waste did not necessarily make the knowledge production process more reliable—sometimes quite the opposite (see also Alexander & O’Hare, 2023). The temptation not to know was thus constantly present in the knowledge production process: We did not always want to know what the disgusting, damp and smelly waste bags contained, especially after we had gained more experience about the potential contents of these bags—and still this knowledge was crucial for the study.

Determining categories: Translating masses into as-categorizable-as possible objects

After the sample was collected, the main part of the day-to-day work in the waste composition study was assigning a category to each piece of waste. This work of translating waste masses into as-categorizable-as possible objects constituted the second phase and transformation of waste in the chain of translation. It should be noted that ‘mixed waste’ is itself a product of categorization, a process that flattens the multiplicity of heterogeneous matter into one neat, symbolic, manageable category. Our task was to pick apart the category of ‘mixed waste’ and to classify it again into its parts, to know what the category included. This highlighted the material aspect of waste.

The sorting procedure started by placing a couple of waste bags on sorting tables from the sample container. The bags were opened either with knives or by hand. We had to be careful not to stick our hands directly into the bags since they may have contained, for example, sharp objects. Instead, we were advised to empty out the bag on the table or create a large hole in the bag to see its contents fully. The sorting tables had grates through which smaller particles fell to the bottom of the table so that they could be weighed separately after the whole 100-kilo sample was sorted (Figure 2).

OPEN IN VIEWER

Figure 2.

The waste was divided into categories on sorting tables with metal grates on top. Each category of waste had its own plastic bucket or container around the sorting tables.

Once the trash bag was opened on the table, each item from the bag had to be categorized before it could be removed from the table. The established waste categories were mostly based on the material components, such as metal, paper, and plastic. The waste bags that we opened on the table were usually crammed with plastic packaging from food products, paper towels, biowaste, maggots, ash, cat litter, indeterminable fluid, cotton pads, and many other items tangled up with each other. Formerly distinct objects had turned into anonymous masses in the bags. Our job was to make sense of this mess by categorizing the anonymous masses into objects that were, at least to some extent, recognizable (see also Lehtokunnas & Pyyhtinen, 2023). This meant that if, for example, a plastic package of minced meat was full of cotton pads and maggots, the pads had to be identified as separate objects and removed from the package (the maggots stayed with the matter to which they were stuck).

The categorization of waste masses into identifiable objects required that we used multiple senses to conduct the labor. For certain materials, there were tricks and cues for checking the material. For instance, to detect whether a yoghurt lid was made of plastic or metal, we were advised to use the ‘wrinkle test’; metal will stay wrinkly, while plastic straightens. Training the body to differentiate materials required knowing and observing the qualities of the materials. Determining whether an opaque package still had something in it was done by shaking and listening to the item as well as trying to estimate the weight of the product.

I noticed what I was doing with my body. I shake, I sway, I squeeze, I rattle, I crumple, I listen, I look attentively. I shake deodorant containers and other cosmetic containers. By shaking a container, one can determine surprisingly well if there is something inside or if it’s just the package. Probably somewhere from one’s spine arises an estimation. … The lids of yoghurt packages need to be crumpled to know whether they are plastic or metal. Medical packages need to be rattled and listened to to know whether they are empty. Multiple senses are needed to separate the materials. (Extract from Ulla-Maija’s field diary)

Thus, it was often not enough to visually identify a certain object; categorizing objects was done by creating a system of sensory perception through the simultaneous use of vision, touch, shaking and listening, and other actions. Mixed waste sometimes also included biowaste, which required specific sensory techniques from the sorters. Sometimes, biowaste needed to be smushed between one’s fingers before one could tell the difference between, for instance, a paper towel and a potato. This had to be done since, as all waste items get mixed and start to rot in the waste pickup process, they become less easily identifiable. However, even as we learned these multiple bodily strategies for translating masses into knowledge categories, a certain element of unknowing (see Alexander & O’Hare, 2023) was always present. For example, it was impossible to tell whether a piece of cling film came from product packaging or from a cling film roll that someone used at home. In this sense, sensory knowledge was not always enough to ensure completely reliable categorizations, but all the objects had to be classified regardless. Even though one would be able to use the senses to identify an object, its’ origins affect how it should be categorized, and these origins, in turn, may be impossible to detect in some cases.

A further element obstructing the fluent categorization of waste masses was the need to wear protective gear. We had to protect ourselves both from the sensory qualities of waste and from the bacteria that waste contains. We used gloves (three sets for some of us), which created a physical barrier to the sense of touch and covered the tips of our fingers, which in everyday settings we use to sense and detect different materials. We also had to wear face masks and protective goggles to protect our respiratory organs and eyes from the waste. This safety equipment made it possible to complete the work, by keeping the smells and textures of the waste at bay, but it made us feel sweaty and uncomfortable, steamy protective goggles obstructed sight, and gloved fingers were not able to sense the waste directly.

The waste still took a heavy toll on our senses. The general invasive smell of waste in the sorting tent made us not want to breathe through our noses; at the same time, breathing through the mouth was not always easier, as one can also smell through the mouth. There were also certain types of waste, such as feces and blood, that we tried not to look at too closely, to avoid bodily reactions. Dealing with the repulsiveness of waste meant that we had to use our senses but limit them, developing a type of bodily discipline (Muniesa & Trébuchet-Breitwiller, 2010); we had to be brave enough to swing, squeeze, and poke different waste items to recognize them, but careful to not examine certain items too closely, to avoid gagging. Due to sheer disgust, some things may not have been sorted correctly.

Identifying correct categories for the waste was a constant process of probing and interpreting individual fragments and items. Sorters habitually consulted with each other to identify and classify waste fragments into the correct (or agreed-upon) fractions. ‘Which bin have you put X in?’ was frequently asked. Asking for guidance from one another was often a quick way of knowing and involved a calibration of senses between sorters (Harris, 2020). Thus, single acts of translating the waste mass into recognizable objects were often the results of joint negotiation rather than precise efforts either to identify objects using our own individual senses, or to study the rules of the composition study in the instructions we were given by the waste management company (see Woolgar & Neyland, 2013). Despite the written instructions and discussions with colleagues, we came to learn that individual sorters occasionally interpreted the waste fractions differently; sometimes, they corrected their sorting, and at other times, they let these interpretations become part of the margin of error. As previous studies on waste categorization practices have shown, waste classification schemes are often ambiguous, and as a result people create their own schemes and interpretations (Woolgar & Neyland, 2013). Thus, the sensory labor of categorizing waste was in many ways based on shared practices as well as individual interpretations. The sorted waste objects were not fixed, but rather viscous—they were constituted by multiple different agencies and relations between objects, people, and places (Persaud et al., 2019). And in some cases, a correct categorization would have required more knowledge about the origins of the object.

Weighing and quantifying the waste: Translating waste categories into as-accurate-as-possible numbers

Finally, when the 100 kilograms of mixed waste had been sorted through and only tiny ‘smushes’ of waste were left on the table, the weighing process began. Weighing and quantifying the waste was the third phase in the chain of translation that we identified as part of the composition study process and through which the sorted waste was transformed into numerical data.

To weigh the categorized waste, one worker sat in front of a laptop and opened an Excel form. A scale whose screen showed the numbers was positioned next to the person in charge of recording all the data in the Excel form. Other workers started to prep for the weighing; some were weighing the waste in bigger containers with a pallet jack scale, and some were already combining the contents of buckets carrying the same categories of waste. The weighing often started with the contents of the bigger containers, such as plastic packaging, and these numbers were yelled out to the scale ‘manager’. This person started to call out the categories in their numerical order (each fraction had a number), shouting, for instance, ‘and then glass packages and glass’. Others brought the completed buckets and containers and positioned them on the scales. Typically, buckets were queued up in front of the scales, waiting for their turn to be weighed. Once the buckets were weighed, their contents were emptied into bigger containers. After this, the buckets were brought back to their correct places around the sorting table. All the carefully categorized waste fractions were once again treated as mixed waste.

After all the waste fractions had been weighed in their designated containers, the waste smush that had fallen through the grates of the sorting tables needed to be weighed. In this sense, the smush, too, needed to be categorized at the end of the sorting process of each sample, but this categorization was different. The smush was first carefully examined to identify the different proportions of waste that it contains (e.g. coffee grounds and small plastic particles), and overall composition percentages were roughly estimated. This evaluation was conducted through reliance on vision and touch. We, however, often had to rely more on visual evaluation than touch, since the different materials that the smush contained were often thoroughly mixed with each other and it was difficult to separate them.

This process also involved some negotiation, as different sorters had different ideas about the correct percentages. Differences related, for example, to the difficulties of visually evaluating the weight of different proportions of waste. Sometimes, for example, the smush seemed to contain a lot of plastic and only a small proportion of coffee grounds, but wet coffee grounds are much heavier than plastic, and because of this, a larger percentage was given to biowaste. In these cases, the system of sensory perception needed to be oriented to visually evaluate weight, and some of the sorters (especially the employees of the waste management centre) were more trained to do this. This hindered the consistency of the practices of categorization through sensory labor on some occasions.

After evaluating and recording the percentages of different proportions, the mixture was then swept into its own container and weighed. Weights for all of the waste containers and buckets when empty were written on the sides, and this information was also included in the Excel sheet. Excel constantly calculated the overall amount of waste weighed, and in the end, someone checked whether this number adequately matched the sample’s original 100 kilograms (Figure 3).

OPEN IN VIEWER

Figure 3.

One researcher documented the weight of each category in an Excel sheet, while others brought containers to the weighing station and emptied them into the mixed waste.

Although the weighing required lifting and moving the containers containing the sorted waste, it did not require dealing with the waste as closely as the other phases of conducting the study. However, after several weigh-ins, one could get an overall idea of the typical distribution of waste into different categories, and exceptions to this were noticeable. For instance, one day, the textile waste buckets were almost overflowing compared to earlier weigh-ins, and an extra note about this was added to the Excel form. In this case, the knowledge gained during the same phase of translation with earlier waste loads contributed to the accuracy of the weighing and helped us recognize major exceptions to ‘the usual’. Altogether, in a sense, the inaccuracy of the knowledge produced through the sensory labor of categorizing waste (during the earlier phases) was tempered through the process of weighing, which was a more direct and distinct way of producing knowledge and ‘hard’ quantitative data for the composition study.

However, the production of this quantitative data was not straightforward or flawless. The process allowed for errors at the weighing stage, as the scale did not work accurately at times, and exhaustion sometimes caused the numbers at the weighing station to be heard incorrectly. Still, weight was the most essential factor in the composition study, as the end result was reported in kilograms. While plastic and cardboard containers took up most of the space in the sorting containers, the most weight came from biowaste. The final weigh-in never amounted to 100 kg, which means that part of the waste escaped during the sorting process. Some of this was undoubtedly liquids or small particles that escaped from the sorting stations. Moreover, and related to weight being the most essential factor in the study, some of us noticed a change in our thinking after spending a few days categorizing waste:

Now that I have spent four days in the field, I have started to better understand the logic of the composition study. So far, I have mainly focused on identifying different objects, but today I realized that the weight of different materials is more relevant than focusing on whether single objects are categorized correctly. For example, it is not very relevant whether one packed lollipop is categorized correctly, as it is so light that the scale does not even recognize it. (Taru’s field diary, 17 August 2023)

As we gradually came to grasp the ‘big picture’ of the composition study, our ways of knowing waste were shifted. At the beginning of the study, we focused mostly on the recognition of individual objects, such as lollipops, chicken legs or diapers. Towards the end of the study, we became more aware that, in this context, knowing waste required understanding the different material qualities of waste, especially the weight and composition of particular materials. Only the weight of each categorized waste fraction was recorded in Excel—the accurate recognition of individual objects did not matter at this point, as long as the categorizations were done more or less correctly. This also was apparent in our discussions with the employees of the waste company. One worker, for example, said that she does not find waste disgusting, as she mostly just thinks about materials when sorting. After each 100 kg sample was sorted, the experience of gagging, touching worms, and fighting the stench of waste was translated into a neat Excel file that, in principle, provided valid data for comparison.

The inaccuracies of the sensory labor during the first two phases of translation were managed through the practice of weighing at the end of sorting each sample. In this sense, although the whole chain of translation was based on categorizations of individual objects, in the end, masses were more relevant than individual objects from the viewpoint of knowledge production. Further, categorizing the proportions of waste in Excel produced a certain technique of unknowing (see Alexander & O’Hare, 2023) in which all the tensions, flaws and contradictions in the process of categorizing waste through sensory labor became invisible.