My notes and other stuff


Paper: On Nonscalability

I hesitated before picking this paper for this week because it does not necessarily fit entirely well with what I usually post, and also I feel like I have far less background to judge or critically review it. It however offers an interesting lens with which you can look at systems and loosely ties to other readings I've made (I'm thinking of The Safety Anarchist and Seeing like a State). The paper is On Nonscalability: The Living World Is Not Amenable to Precision-Nested Scales by Anna Lowenhaupt Tsing.

The text contrasts the concept of scalability, as used mostly in industrial and economic systems, with something she coins called non-scalability, and then explains why she believes this framing can prove useful.


Scalability is defined here as the ability to expand (and expand, and expand again) a system without rethinking its basic elements, nor without transforming it or its nature. Small projects that can become big without altering their nature can be said to be scalable.

The author cautions that Scalability is only possible if elements of a project do not form "transformative relationships" that may change the elements themselves. These transformative relationships, however, are described as necessary for the emergence of diversity. She points out that scalability projects "banish meaningful diversity", any diversity that impacts and can change things.

She then makes an analogy with pixels, which she defines hastily as uniform, separate, and autonomous—and therefore scalable—parts of displaying an image, and coins the term "nonsoels" to stand for "nonsocial landscape elements" as a general scalable part of a system. Since they're non-social, they're isolated from relationships and can be abstractly manipulated.

To put this analogy in practice, she goes through the example of sugarcane plantations in colonial times, which were started by European colonial powers in America as a way to work around Muslim-controlled production. Basically, all sugarcane planted in the New World was a functional clone, where people would stick a cane in the ground and then it would just sprout. Sugarcane initially came from Southeast Asia, and when it was first planted in the Americas, it had no history of either companion species nor diseases. It had no inter-species ties and was isolated.

This isolation is what made it so amenable to high-scale intensive agriculture. On the island of Madeira, with a warm dry climate and flat land, irrigation allowed ideal control for all growth conditions. All you had to do was prepare the land (by seizing it from native populations they displaced) and growth was predictable and manageable. Enslaved (African) workers made these advantages to the growers even greater:

[S]laves had no local social relations and thus no easy place to run. Like the cane itself, they had been transplanted; and now they were isolated. They were on their way to becoming self-contained. Furthermore, the plantations were organized to foster alienation and thus enhance control. [...] Already considered commodities, they were given jobs made interchangeable by the monotonous regularity and coordinated timing engineered into the cane. Slaves were the next nonsoel, design elements engineered for expansion without change.

Sugarcane plantations could then expand and grow. Their main components were cloned planting stock (low variability), [un]free labor, and open [conquered] land. This led to unprecedented profits. The author compares this approach to what happened later, as factories modelled themselves on plantations, with the segregation of work and nature and their alienation being part of the basic plans.

She points out that for centuries, scalable projects expanded through the world in an ocean of nonscalable diversity, until they took so much room that diversity was forced into "residual puddles." This ever-expanding scalable set of projects (which the author frames as capitalism in general) does not stop for human needs, and knows only further expansion, even as the public widely knows of its horrors.

The author mentions that a more comprehensive view would define scalable projects as the articulation of scalable and nonscalable parts; the nonscalable part are just hidden from the view of the planners and investors.


Tsing starts by stating that nonscalability is by no means supposed to be better or worse than scalability. Scalability never fulfills its own promises, and the world remains diverse and dynamic. There is no normative scale between scalability or non-scalability, and she gives example of both horrible and less bad nonscalability.

The first one is based on the Matsutake mushroom, which is a type of mushroom that can't grow in isolation. It only exists in a state of interdependence with other species, as it grows within the root systems of certain trees in poor soils, and trades nutriments with the host tree. Cultivation of these mushrooms has not really been done successfully. In the US, this mushroom became easier to forage after decades of turning forests into plantations, which failed after drastically reducing the diversity of some northwestern forests, which were abandoned after diminishing yields.

These failed managed forests had poorer soils and different varieties of trees post-abandonment, which in turn made the ecosystem more amenable to the growth of matsutake mushrooms. As the production in Japan went down in the 1980s, the white populations looking for "freedom" and a bunch of Southeast Asian refugees started foraging for their own purposes, with some selling their harvest to independent buyers:

Matsutake foragers in the Pacific Northwest work only for themselves. Most are there because they love mushroom picking—for the freedom of the forest, for the independent searching, and for the money, which they use to support themselves. [...] [T]they do not have standardized work practices that can be accounted for as “abstract labor”; they do not feel alienated from the work process. They are nothing like nonsoels. Since they come for their own reasons, it would be impossible to expand the work unit without transforming it.

By the time the loosely harvested mushrooms are exported by independent buyers, they have become scalable again: sorted by weight, grade, maturity. They have become a capitalist commodity:

Here we have stumbled on another kind of articulation between the nonscalable and the scalable—not the ruins of scalability, but the recuperation of nonscalable forest resources for scalable inventory. Transformation from unscalable process to scalable inventory is what the contemporary capitalism of supply chains does best.

Scalable projects can reap the rewards of nonscalable projects. She calls the people who do this "pirates" who "steal from the work of transformative relations," something she calls "supply-chain capitalism", which at its heart always has this junction of scalable and nonscalable elements. The supply chain's work is one of translation and sorting, of turning the results of nonscalable projects into inventory.

In this model, no means of production requires to be scalable. They can be purely destructive and exploitative: destroy the environment, move on. The impacts of pushing nonscalable work is externalized onto producers, along with the responsibility and its costs.

The author mentions that this pattern is visible in more places than ecological production, and she mentions software as one other example of taking the raw materials for free from the public or commons sources and reaping all of its benefits.


Framing scalability in terms of the articulation of scalable and nonscalable parts is something the author believes to be more useful. She revisits the sugarcane production and points out that most functional clones had a fungus in them that made them grow well, and that once disease started spreading (and sugarcane became socially tied to its ecosystem), it became less scalable.

She also points out the risks around research: if you're gathering data for analysis, the definitions must be shared and applicable to all sources lest the research be nonscalable. Taking a scalable approach therefore risks making the nonscalable (but essential) elements invisible.

More modern approaches, she suggests, should be concerned by looking at the making of diversity by studying and paying attention to the interactions and transformative relations, rather than self-contained nonsocial elements. Emergence could take priority over expansion, and searching "critical descriptions of relational encounters" could help prevent nonscalable effects that keep coming to haunt us.

She concludes that scalability has never been complete. By design, it covers up and blocks transformative diversity of social relations, and tends to leave ruins in its wake. She encourages reconceptualizing the world by considering its nonscalable elements, and possibly helping rebuilding it that way.