Direct Link to File. 840 words, 4 minute read, 3 paperback pages

(Chapter 2?) A History of Sorts

One of the central themes (THE central theme??) of this book is to compare and contrast classification (i.e. the construction of categories and the assignment of new “items” to their respective categories) as done by humans vs. that done by machines, and in particular machine-learning (ML)-based classification systems. Most of these machine-based classification systems are intended to replace or automate the work done by humans in creating and assigning categories – and the machines may do this well or may fail in unanticipated ways – so we may ask, how do you humans classify? There are two ways of answering this question. The first is the historical look at how people in various fields have conceptualized classification (on a formal level), whether in philosophy, biology, library science, or other fields. The second is the psychology of how do humans typically classify things, cognitively? Let us begin with the first question about the history of formal thinking about classification, what we might call “classification theory.”1

It’s difficult to pin down how early one should try to go when discussing the history of categories, kinds, sorts-of-things, and so on. In terms of Western philosophy, the name of Plato and Aristotle are the first two names that [typically come to mind].

[todo: now, the next paragraph contradicts with the above statement that we’re going to stick to formal thinking, and western thinking at that.]

However, one could argue that much, much earlier – from the dawn of life itself – categories have existed. An organism must be able to acquire food, and doing so economically and safely involves distinguishing between what is food and “not food.” I would posit that the earliest classification problem ever posed is “Should I eat this?” (And, lest one wants to be pedantic about “shoulds” and volition, I mean “Is this thing before me of the ‘edible’ sort or not?” And am I allowed to say “pedantic” when writing for a general audience? Can I just come back and edit that to read “picky” instead?) Whether an proto-organism (is that word) actually forms a mental category of “things that are edible (or not)”…is an interesting question, and ends up as a “meta-“question about “what sorts of things count as categories and which don’t?” I would say it is up to us to decide how we want to set this up, in other words I will take the view – expressed by many thinkers which we will come to as the chapter unfolds – that categories are conventional, meaning that they are decided by “us” on the basis of how useful they are, and not (necessarily) that they are metaphysically “real” or not.

So, it is a matter of convention to consider whether a plankton’s actions of eating some sorts of things and not eating others amounts its classifying things as food or not food. We should note however that all of this “theorizing” is being done by humans, not plankton, anyway. So perhaps humans and their categories are a good place to start. Again, there are arguably prehistorical uses and conceptions of categories which archeology and anthropology could reveal, but perhaps we will start with what we have writings for.

TODO: note that these biologically-inspired systems are VERY relevant for AI systems, esp simple classifiers. could also rip off story about pigeons from Hanny Fry. ;-)

TODO: Curious: What about non-Western sources? Chinese, Arabic, Egyptian, African?

Charlie: Few philosophers in the Western canon eclipse the influence of the Ancient Greek philosopher Plato. This influence is what led Alfred North Whitehead to claim that all of Western philosophy can be interpreted as a series of footnotes to Plato. Nearly as famous as Plato, and synonymous with his name, is his theory of Forms. Though not the first example of man’s interest in classification theory, Forms are, perhaps, the earliest and most solidified theory.

Aristotle: wrote a text called, in fact, Categories! In this, Aristotle enumerates all the possible kinds of things there are in existence! So, a modest work.

Wittgenstein

Uhhh.

Rosch.

Comte?

When did we start categorizing people? Like for government? Probably since the dawn of time. Who invented bureaucracy? The Romans? The Babylonians? The Egyptians? The Chinese? Ugh. This is relevant for the idea that ML gets used to automate bureaucracy.

How does the history of “formal” thinking about categories fit with the psychology of categories themselves? Should the psychology – even though it’s been going on “forever,” not be a separate topic/chapter? I think so.

Can one argue that, when biological species cease to be able to interbreed, that they in essence form categories?

References

  1. Note I am avoiding using the similar-sounding term “category theory,” which is an abstract branch of mathematics involving graph-like structures, not a “theory” about what you I think of as “categories.” We will not discuss “category theory” at all in this book, although it is an active and exciting field. ↩