Communications of the ACM
Does Nature Use Data?
Whereas the milieu of the philosophy of computer science has so far concentrated on formal systems, as described in the Stanford Encyclopedia of Philosophy entry (Turner), many other interesting and enticing subjects invite exploration. Computer science has defined new tools, new objects, in formal settings, but what if we look at those things in themselves, without regard to a formalization? Let us look at, for example, data, but outside the questions of structure, storage, and security--in fact, way outside--in fact, outdoors. Let's ask if nature uses data.
As good philosophers, we ask, immediately, what is meant by "data" (and, at some point, we'd better ask what is meant by "nature"). We will refer to data in the singular, as a mass noun; hence, "data is." We can say that data is a collection of facts, but we are surrounded by facts, in a weak sense, that do not seem to rise the level of data--the amount of a change in a pocket, the cars that pass down the street during the day, the strength of a gust of wind. Data, we observe, is intended to serve some informational need. The definition attributed to Gregory Bateson that a datum is "a difference that makes a difference," adopted by Floridi, is not only intriguingly primitive, but philosophically tantalizing; however, that definition gives the broadest sense, rather than the one we take up here. The one we take up here requires that facts be reified (made into things) and staged, or arranged, for some use; that seems to be what we mean when we talk about data.
We have to stipulate that people use data, via the human brain, and that since humans are creatures of nature, then the answer is "yes, of course nature uses data," on that line of reasoning. To ask the question in a more interesting way, we must pose it outside the constructions of the brain. We simply ask whether a version of nature devoid of brain functions, or external to intelligent agents, would use data. (So that's what we mean by "nature.") We can plow the ground, so to speak, by considering some preliminary questions, by approaching this from outside the general question, starting from more specific questions, creeping up on an issue from below, so to speak, a useful philosophical tactic. Does nature use databases? Does nature use data structures? Considerations of these may assist with the main question: Does nature use data?
Does nature use databases? A database (schema) is a highly structured object, with arrangements suggested by nature but imposed by designers. But nature neither produces nor recognizes organized data about objects that is independent of the objects in space and time in any systematic way. Databases survive even after the objects or entities respresented disappear. For this reason alone, we can conclude on this quick analysis that nature does not use databases.
With a switch in perspective, many intriguing questions about the database itself can be raised. We can move from looking at data through the perspective of nature to looking at nature through the perspective of data. (Taking care with the analytic stance is part of the discipline of philosophy.) Now we are in the world of the data modeler or designer, a professional who faces a daunting task. Is there any object in nature that carries a fixed set of attributes, with fixed sets of values, really? Is there any reasonably-bounded set of relationships in any natural domain? If we were to start from scratch to build an ontology of the universe on paper, for people, and we were to start from scratch to build an ontology of the universe in some data model, for a database, would they turn out to be the same? Suppose, as seems likely, each effort fails at some point; would they have carved out the same portion of the universe to represent? Would they fail for the same reasons? Is a breakdown into entities, attributes, and relationships "natural?" The difficulties of data modeling, as described in a classic account by William Kent, suggests otherwise. What are the alternatives? Is there some greater abstraction, some sort of category theory, of information capture? These pesky questions may shed light on why database design is so difficult, why ontologies are so slippery. Yet databases work so well for us, in so many situations, in spite of their brittle representation of the real world.
We digress. Does nature use data structures? Can we look out the window and find an array, for example, or a linked list, in either its static structured-variable form or its more comprehensive abstract data type form complete with associated methods? Data structures are self-contained first-class objects. They can be passed as parameters. They are rich in representation. Nature deals only with what there is, not with separate representation of it. A footprint forever caught in mud turned to stone may seem to be a rich data structure, but only to us. It is not treated, by nature, as a instantiation of the "footprint" class, providing methods that process data about the foot. Note that DNA just misses the boat, data-wise, on this reasoning, because DNA is of the creature, not independent of the creature.
Does nature use data? Our rather weak definition of data calls for reification--a philosophical notion not without controversy of its own--and arranged somehow for use, which implies some time sequence and some act of reference that takes place after the arrangement. Either we grant that nature takes these actions or we take these actions only ourselves. We can define data so that it only exists in our heads, and is imposed from there onto the world, but that would beg the question.
The answer to "Does nature use data?" would be "no," because we've defined it that way. We can define data so that all phenomena create vast swathes of data, referenced by other phenomena in the attenuated sense that one grain of sand restricts the location of another, but that would also beg the question, yielding the answer "yes" because we defined it that way. Either every little (and big) thing is a datum---pocket change, cars, wind, the placement of every grain of sand, every cell of an organism, the vastness of every stretch of empty space, the heat from solar radiation---because each is a difference that makes a difference, or none of those is a datum, because differences are only made, or noted, or exploited, by us, by humans. This exploration, it would appear, rests quite heavily on the definitions, which await further analysis and clear articulation. Such an outcome will be not discouraging, but rather exciting, to intrepid philosophers of computer science, and even to interested computer scientists too modest to call themselves philosophers.
References
Floridi, Luciano, "The Philosophy of Information," 2011, Oxford University Press.
Kent, William, "Data and Reality: A Timeless Perspective on Perceiving and Managing Information in our Imprecise World," 1978.
Turner, Raymond, The Philosophy of Computer Science, in "The Stanford Encyclopedia of Philosophy," http://plato.stanford.edu/archives/fall2013/entries/computer-science, 2013.
Robin K. Hill is adjunct professor in the Department of Philosophy, and in the Wyoming Institute for Humanities Research, of the University of Wyoming. She has been a member of ACM since 1978.
Comments
Johan Lammens
While doing some data modeling work for a Big Data application recently, I have observed that this is nothing more nor less than a form of knowledge representation. What goes around, comes around ;-)
Robin Hill
As a conscientious data modeler (hello, Joe!), do you have views on how the relation of Big Data to nature might differ from other sorts of data that we have processed for years, such as records and measurements treated as discrete objects? In other words, what can we say about whether statistical modeling (associated with Big Data) is reflected in nature?
Displaying all 2 comments