I've been trying for the last few weeks to write a post grounding my thinking about narratives in my understanding of complexity, insisting that stories are complex systems. The implication, of course, is that understanding complex systems is a fine way to understand stories, which seems correct to me, or at least, promising enough to explore. However, when I asked AI Gemini to critique my post several times, it repeatedly objected to my comparison of narratives to frogs in an attempt to show how both were complex systems. I tried several ways to address Gemini's complaints, but it persisted, telling me:
You have doubled down on the frog metaphor, which remains your weakest logical link. … You are conflating the method of analysis with the nature of the object. A complex system doesn't become a complicated one just because a human looks at it through a reductionist lens.So yes, I'm arguing with an AI, and perhaps I am struggling to take its criticism seriously, so in this post I want to clarify my ontological and epistemological treatments of complexity. Quite fortuitously, I just read an essay Mapping degrees of complexity, complicatedness, and emergent complexity (2018) by Timothy F. H. Allen, Preston Austin, Mario Giampietro, Zora Kovacic, Edmond Ramly, and Joseph Tainter who take a strictly epistemological approach to complexity, explicitly distancing themselves from debates about whether complexity materially exists in the world. Their abstract says:
Throughout the paper, we carefully eschew ontological issues, and sort out the epistemology of complexity. We try to explain why the ontology of complexity makes no sense to us, much like significance is neither material nor ontological.
Epistemology-Only Complexity: By my reading, Allen and his colleagues argue that complexity is not a material or ontological property of the external world, but rather a relational property between a system and the observer attempting to model it. In other words, complexity is a property of how we know reality, not reality itself – epistemology and not ontology. Drawing on the work of Robert Rosen, they define a complex system as one that requires a nonsimulable model, or a living-logic, and would need an infinite number of distinct formalizations to capture all its qualities. For these authors, neither organisms nor ecosystems are real independent of human abstraction. Therefore, complexity arises entirely from the way a system is addressed and the abstractions human observers use to define the purpose and constraints of a system. They conclude that they "are not sure what the ontology of complexity could be".
Onto-Epistemological Complexity: Most of the complexity theorists that I follow disagree, claiming that this rigorous separation between ontology and epistemology is fundamentally impossible. Paul Cilliers argues that because we always deal with the world and our descriptions of it simultaneously, complexity is a combination of a system's ontological attributes and the epistemological function of our understanding. In his essay Why We Cannot Know Complex Things Completely (2002), he says:
Even if we acknowledge that our descriptions of the world are not perfect, we would like to maintain that they are not merely instruments, but that they enhance our knowledge of the world as it is. There is a complex dialectical relationship between the world and our descriptions. When we try to understand the world we are always dealing with ontological and epistemological issues simultaneously. (84)In her 2012 dissertation The Problem of Complexity, Preiser notes that scholars such as Lesley Kuhn and Gregory Bateson maintain that in the natural history of human beings, ontology and epistemology are "mutually constitutive". How we perceive and act in the world determines our beliefs about its nature, meaning we are bound within a net where the two cannot be isolated. Preiser summarizes her position on the issue this way, and by the way, references my two key complexity thinkers Edgar Morin and Paul Cilliers:
For the purpose of this study, the corresponding arguments offered by Cilliers (2002, 2007b) and Morin (1992c, 2008) will be adopted, namely that in the process of studying complex systems, it becomes impossible to conceive of epistemology, ontology, methodology and causality as if these were four separate distinguishable domains. (94,95)
Ontological Complexity: Many other scholars that I read take this position a step further by explicitly grounding complexity in ontology. In his essay What Is Complexity Science? (2001), Göktuğ Morçöl asserts that complexity science broadly assumes a "realist ontology"—the belief that systems and their nonlinear relations exist objectively out there, independent of the mind. In his essay The Wicked Nature of Social Systems: A complexity approach to sociology (2015), Anton Tornberg references the work of David Byrne and Gill Callaghan to argue that complexity science should be understood as an ontologically founded framework which asserts that the social world actually consists of complex systems, and must be analyzed as such. In explaining his approach to social systems, Tornberg says (28):
While traditional theories provide concepts and causal connections, attempting to explain social phenomena, it makes more sense to think of complexity theory as an ontologically founded framework that asserts the specific ontological position that the social world consists of complex systems, and if we wish to understand it, we need to analyze it in those terms (Byrne and Callaghan, 2014: 8)
Similarly, Minka Woermann (Bridging Complexity and Post-Structuralism, 2016) emphasizes that philosophical complexity is primarily an ontological position, insisting that in the ontological view of complexity "the world is inherently complex, because there is no central organising principle and the system is open" (18). While these ontology-first thinkers do not deny the complexity of epistemology, they begin by grounding complexity in their ontologies: because reality is complex, we should use complex approaches to studying it and knowing it.
Points of Agreement: Despite this profound disagreement about whether complexity is grounded in epistemology, ontology, or both, all these thinkers agree on a few points that will be relevant to my exploration of rhizonarratology, my belief that narratives are nonlinear, root-like systems that grow in unexpected directions. They are all aligned on the epistemological consequences of complexity: on the incompressibility of complex systems and on the necessity of reductionism for understanding complex systems.
Timothy Allen's reliance on Rosen's idea that complex systems are nonsimulable aligns, at least superficially, with Paul Cilliers' and Edgar Morin's assertions that complex systems are incompressible. All agree that a complex system cannot be fully mapped or reduced into a simple mathematical model or theory without losing the very complexity one is trying to study. In their introduction to the special issue of Emergence (3:1, 2001) Kurt Richardson and Paul Cilliers say, "Possibly the most important “law” associated with such systems, in light of the project to develop the Theory of Complexity, is that these systems are incompressible, i.e., any description claiming completeness must be as complex as the system itself" (8,9). This nonsimulable incompressibility presents real problems for all our scientific and aesthetic models of reality and how they work. All our models – our calculations, formulae, paintings, plays, and poems – are all incomplete and often leave out the very qualities that make the system notable in the first place.
Complete knowledge, then, of any complex system is not available to humans, and because complete knowledge is impossible, all camps agree that researchers are forced to make choices to reduce complexity in order to understand it, to model it. This creates boundary problems for everyone. Allen points out that "thingedness" comes from human decisions to place boundaries on a fluid experience, just as Cilliers and Woermann note that drawing boundaries around a system is a normative act of mental construction and convenience. In his book Complexity Theory and the Philosophy of Education (2008), Mark Mason explains the predicament this way:
The idea that complex models are not isomorphic with the complex systems they purport to represent has been defended in detail by Cilliers (1998, 2001). In a nutshell, the argument is that since we cannot understand something complex in all its complexity (as humans we have limited means and limited time), models, by definition, have to reduce complexity. … Complex systems, however, are by definition ‘incompressible’: they cannot be ‘reduced’ without losing something (Cilliers, 1998, pp. 7–10). (208)
Summary of Epistemology Vs Ontology: To summarize the views of these two camps, while Allen et al. believe complexity is purely a product of human discourse and abstraction and other theorists believe complexity is an undeniable feature of reality, both agree that our knowledge of complex systems will always be limited, provisional, and fundamentally shaped by the observer's models and boundaries.
However, even these two seemingly similar concepts illustrate the fundamental difference between the epistemology-only camp and the onto-epistemological camp of Morin and Cilliers. For Morin and Cilliers, the shortcomings lie on the mismatch between our knowledge and the nature of the complex systems under observation. For Allen, the shortcomings of models lie totally on the side of the knower creating the model. This seems to me to point to an inevitable contradiction in the epistemology-only argument.
To see this contradiction clearly, we have to look at what Rosen calls the internal predictive model. If I understand Rosen's (and thus, Allen's) view of nonsimulable, then all complex systems create and use an internal predictive model of itself and its environment. The model is a kind of living-logic, a set of functional relationships within the complex system that allows it to anticipate and adjust itself to reality. It allows an entity to work outside the box or beyond the blueprint or physical and chemical reactions. For example, our body’s circadian rhythm is an internal model of the Earth's rotation. However, our body doesn't just react to the sun coming up as a light meter or a thermometer will; rather, it anticipates sunrise by shifting hormone levels and performing other physical and mental tasks before we ever wake up. This internal modelling ability allows our body to change its present state based on a predicted future state, which follows from its ontological model of how the world works. And though changing our circadian rhythm is arduous, it can be done if conditions change enough. In other words, a complex system is self-organizing, and this inherent, nonsimulable modelling capability is part and parcel of that process.
This internal modelling system is an ontological characteristic of all complex systems: you and me, dogs, slime molds, rivers, galaxies, and novels. We all dynamically form models of ourselves and of all those other systems that perturb us, and we do it to maximize our fit into reality, and we do it largely nonconsciously. We complex systems perceive our internal state, our external states, inputs of energy, matter, information, and organization; we process that input, adjust ourselves to make best sense of and use of that input, and then output our own energy, matter, information, and organization. We humans are not conscious of most of this activity. Most of us are unaware of our circadian rhythms until our sleep is disrupted in some way and we are left groggy and irritable the next day. However, this modelling dynamic is part of what a complex system is. It is ontological and nonsimulable. Allen et al. seem to limit the entire idea of complexity to our conscious bit of modelling: to our computer simulations, calculations, plans, and stories. Moreover, they make this bit discrete, separating it from other ways in which complex systems model themselves and their worlds to best use the resources they have and to fit into and cope with their complex ecosystems. I don't see any advantage in limiting my view of complexity this way.
I think modelling processes vary by entity, with different parts and procedures, affordances and limitations. Humans seem to have a more sophisticated modelling process than do slime molds; however, both have an innate system that allows each to model itself and its world and to try its best to fit in and even thrive. Narratives have an analogous system, I think. A narrative's internal modelling apparatus includes the expectations of the genre, the rhythms and syntax of its language, the internal logic of its particular mix of characters, the expectations and capabilities of audiences, the gifts and aims of the author, its social, historical, and cultural contexts, the vehicles for production and distribution. A relevant example is this post that I am now writing and you are now reading.
- Expectations of genre: This is a blog post, a hybrid genre sitting somewhere between a personal journal, a newspaper editorial, and at least in my case, a scholarly essay. It is more conversational, with a strong sense of interaction between an author (I) and audience (you).
- Quick lead: Posts don't build to their point, they start with it, often in dramatic fashion, with a clickable, sometimes flagrant title: 10 Reasons You Won't Die Today. I'm not very good at this.
- Formatting: Readers often scan a post before they commit to reading it, so posts often provide headings for an outline or callouts to emphasize main points to keep the reader from clicking on to something more interesting. I'm not very good at this either. I added headings to this post just now, mainly because when I started writing, I had only vague ideas about what I might say.
- Language: Of course, I write in English, but that's a very large space. Because of my professional history, I tend to words, phrases, and sentences that use philosophical, technical, and literary terms. I try to avoid the impersonal scholarly style, but … This ain't for everybody. Probably very few people have discussed nonsimulable internal predictive models this past week. Why would they?
- Characters: Given the topic, I think I have a fine cast of characters: Timothy Allen and David Rosen on one hand and Edgar Morin and Paul Cilliers on the other. They each bring some real intellectual value to the plot.
- Plot: I think this post has a genuine conflict, rising action, and a resolution – at least, a resolution for me.
- Audience Resources: I assume an educated audience, if any, with some interest in complexity and narrative and some hope that I can bring value to that conversation.
- Author Resources: I can write long posts and I have read about complexity and narrative. I can type, not an insignificant skill for blog posts. I'm getting older, but I think I can still make sense, and now I have an AI assistant who will check my writing for errors and inconsistencies. It finds them and helps me fix them before my human audience reads this. This is a new element in the author's internal toolbox, but it works just as well for audiences. Don't quite grasp Rosen's nonsimulable internal predictive models? Ask Gemini or Claude, and they will tell you more than you want to know in under sixty seconds.
So what do I believe about the nature of complexity? I follow Morin and Cilliers in thinking that complexity is both ontological and epistemological. Reality is itself complex, and we need complex ways to map and model complex reality to create the most useful knowledge. Both my lived experience and my reading convinces me that complex ways of knowing best map complex realities. And like Byrne and others, I believe that complexity is ontological first and epistemological second. Because the world is complex – both frogs and novels – we need complex modelling strategies – including stories – to live well within it and to understand it as well as we can.
No comments:
Post a Comment