Same and Different: How Models Contribute to Knowing. A review of Modelwork

ABSTRACT

Modelwork, a collection of articles edited by Martin Brückner, Sandy Isenstadt, and Sarah Wasserman (University of Minnesota Press, 2021), contributes to scholarship about human inquiry by exploring the nature and action of models. Models provide access to hidden realities by bridging the gap between the tangible and the abstract; models also alter perceptions and can have deep and lasting effects on human knowing. The collection argues for the importance of attending to this double action of modelling, and explores important epistemological questions.

KEYWORDS:

• Models
• knowledge
• epistemology
• material culture
• digital humanities
• history of science
• perception
• imagination

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 See also Galison’s excellent 2020 Netflix documentary, Black Holes: The Edge of All We Know, in which the water model is shown, and plays a significant role as an aid to the viewer’s imagination.

2 See for instance Plato’s Sophist, which deals at length with this issue.

3 The story can be traced to the Pali Buddhist canon, Udana 66, in Khuddaka Nikaya 3. With thanks to Willard McCarthy for the reference.

4 UMAP is an improvement over the basic idea of t-SNE, and serves a similar purpose to the older PCA method.

5 For an intuitive understanding of this point, see the excellent tutorial at https://pair-code.github.io/understanding-umap/, especially the 3-D representation of a mammoth skeleton projected from three to two dimensions.

6 The crucial parameters for a UMAP projection are the minimum distance and the n-neighbours, which determine how well local and global structure are balanced in the low-dimensional projection. In other words, they are a measure of just how fuzzy each datapoint should be considered. The diameter of ‘fuzziness’ of each point is also determined by its context: if it is far from other points, it will have a wider circle of relevance.

7 See Owen Barfield’s profound considerations in Poetic Diction.

8 The term ‘tool’ is criticized by Jacques Ellul and David Schindler, among many others.

9 See Eli Pariser’s critique in The Filter Bubble (2012), and Lex Fridman’s hope that algorithms could be developed in much more beneficial ways than those currently in use in large social media companies. It is not necessary for algorithms to optimize for ‘engagement’, which ends up being fear and anger: they could also optimize for ‘happiness’. If they did so, argues Fridman, it would make economic sense as well as moral sense. See his comments on the Huberman Lab Podcast, 19 July 2021, episode 29, ‘Dr. Lex Fridman: Machines, Creativity & Love.’

10 The point is well taken. In fact, the ‘quantitative’ transcription of data that was originally humanistic and text based need not be merely quantitative in the sense of linear scales. It could also take into account relation, and overall ‘shape’ or topology, as other mathematical elements that describe the content.

11 The Centigrade temperature scale is constructed on the basis of an experience (boiling and freezing water) which some peoples had never had before refrigeration and travel made it available to them. This example is susceptible to further critique, since the experience described is also as ‘objective’ an event as can be found. Emphasizing the cultural contextuality of this measurement scale might be pushing it a bit. Water does not freeze or boil at arbitrary or culturally-determined temperatures.

12 Miriam Posner’s work on supply chain software is relevant, as Drucker points out, but should also be read within the specific needs, goals, and incentives that drive all the relations between the creators and users of that software. It cannot be adequately evaluated on the basis of external ideas, no matter how laudable, unless those ideas are also part of the internal dialogue between the actors who produce the device itself. In other words, it would seem like question-begging to lament the absence of attention to climate change within logistics software. Perhaps the first point is to broaden the cultural awareness of the players themselves, and then observe what cultural artefacts and tools they produce.

13 Brann refers between the lines to the problem of induction. She might as well also be referring to the process of backpropagation in a neural network. Can we arrive, through a statistical process of observing regularities, to any ‘sharply precise idea’? It would seem not. Yet we can arrive at highly precise predictions, which seems to imply that there is a fairly precise map of the world embedded into the weightings of the network, even if it is not expressed as recognizable propositions. Perhaps it is also a question of what language and concepts are themselves, which we might understand as heuristics that capture part of the regularity of the network, and which we mistakenly assume to be timeless and ontologically above the mess of irregular reality which occasioned their birth? On this reading, are there then only many different chairs, which do not actually ‘participate’ in the Idea of the chair? And does language, by gathering together similar objects, impose a falsifying light upon them? When Socrates speaks of inborn knowledge, he is perhaps mistaking social and cultural knowledge for a more mysterious process. For when a child is introduced to the world through the mediation of language, many structures and categorizations are introduced which other cultures might order differently. Greater and lesser degrees of attention are paid to different details, depending on their context.

14 Matthew Crawford pointed out how hard it is to believe in the human conquest of matter when one is a motorcycle mechanic, and has to do things like re-boring a piston cylinder first-hand. See Shop Class as Soulcraft: An Inquiry into the Value of Work. Penguin Press, 2009. Metals are not homogeneous, passively awaiting the will of the builder to take form. They resist. Even pure mathematics, which to the outsider seems as ‘logical’ and unsurprising a material as there could be, displays immense surprises in its simplest details, like John Williamson’s UMAP projection of the prime factors of the first million integers at https://johnhw.github.io/umap_primes/index.md.html.

Additional information

Notes on contributors

Jonah Lynch

Jonah Lynch was introduced to the idea of the ‘renaissance man' at an early age, and has found himself constantly attracted to learning about anything and everything. He earned a B.Sc. in Physics from McGill University, then explored Philosophy, earned a Doctorate in Theology from the Gregorian University in Rome and an M.Ed. in Science education from the George Washington University in Washington D.C. He currently conducts history and digital humanities research at the University of Pavia and with the International Institute for Mesopotamian Area Studies.