Why does the word “gold” designate what it does? Currently the word gold is taken to designate a specific type of atom, or, more colloquially, any substance that is made up mostly of that atom. But is that what gold always meant? The word “gold” was around long before we knew about the existence of atoms. Gold was once defined by its density, its color, and its malleability. Indeed, even in modern times, most people have a relatively imprecise conception of gold (especially before they learned about atoms). Most people, I think it is safe to say, just think of gold as a rare shiny yellow metal. But let’s go back to historical times when discovering the nature of gold was still a work in progress. The natural scientists of the times were aware of a number of different substances: typical gold, pyrite (fool’s gold), white gold, and brass polished to look like gold. How did they manage to make a decision as to what counted as gold and what didn’t? Whichever of the ordinary features of gold you pick out it is possible that some substance that we now know is not really gold will be picked out as gold by it (or that some substance that really is gold will be left out). There are other malleable metals, other yellow metals, and other alloys of the same density. Picking a combination of these features that will just pick out what we now think of as gold is hard to do. The most direct way is probably through a definition that appeals to none of the ordinary properties but rather the behavior of the metal under certain tests (namely purifying it and testing its density).
But that definition doesn’t seem to have any special connection with the ordinary understanding of gold, why favor it over one of the other possibilities? Why not pick some more colloquial understanding of the term that includes pyrite or excludes white gold, but which just as determinately and objectively defines a class of materials? One answer to this, the kind of answer that David Lewis, among others, would give, is that what “gold” refers to is the natural kind that best fits the ordinary usage of the term. Sure, there are lots of definitions that more or less respect pre-theoretical intuitions about gold, but all of them, except a definition in terms of gold atoms, are unnatural, they would claim. A definition that includes both pyrite and gold, for example, is lumping together two chemically different kinds of things under one heading, and a definition that excludes white gold is drawing a division where none really exists.
Obviously this isn’t an entirely uncontroversial answer because it relies on natural kinds, which not everyone agrees exist or thinks we have access to. But let’s put the metaphysics aside for the moment and just pretend that natural kinds exist. The first problem with this answer to the question, even granting the existence of natural kinds, is that a definition based, for example, on the reflective properties of the substance seems to be just as natural (surely the way light is reflected is a natural division of the world), and, while it doesn’t perfectly respect pre-theoretical intuitions about gold, it doesn’t do that much worse than the atomic definition. Such an objection might be avoided, however, by a more complicated story about natural kinds, which makes facts about composition more fundamental and facts about reflecting light somehow secondary. Again this is a questionable move that drags in even further metaphysical commitments. But again I would say that we should just concede, for the sake of argument, that these are acceptable metaphysical commitments, since arguing metaphysics rarely gets anywhere. Still, there are other questions we can ask. Why do we divide the atoms by proton count instead of neutron count? Most elements have a characteristic number of neutrons that a majority of the atoms of that type have. So, if we instead divided up the atoms into elements by neutron number rather than proton number we would have roughly similar divisions of the world. Among these divisions there would be one that corresponds roughly to the familiar category of elemental gold. Why not say that this unfamiliar division is what gold refers to, since it too would roughly respect intuitions? Certainly there is no metaphysically acceptable way to privilege neutrons over protons for deciding what counts as a natural kind. Thus the appeal to natural kinds breaks down; there are simply too many candidates for the natural kind that might fit our intuitive understanding of “gold”, and this is even more obviously the case when it comes to substances that are defined partly by structural properties (such as crystals), where there are even more candidates for natural kinds in the same region as our pre-theoretical intuitions.
On the other hand, there seems to be something undeniably appealing about the turn to natural kinds. Wouldn’t there by something wrong with a category that included both gold and pyrite but nothing else? The division feels arbitrary and unnatural, and instinctively there seems to be something wrong with any position that would approve of it. Whatever “gold” might refer to surely the candidates shouldn’t include such unnatural things, even if we are willing to concede that there is no immediately obvious natural kind corresponding to our pre-theoretical ideas.
What is it that strikes us as wrong about a gold-pyrite category? Some might say our intuitions about naturalness. But that just raises the question: why do we have the intuitions about naturalness that we do? And on this issue I am not inclined to be tolerant of metaphysical explanations that justify our intuitions. There are a number of well-known problems with intuition that fairly conclusively indicate that they aren’t truth tracking. That doesn’t mean that intuitions can be ignored, and I don’t intend to ignore them, but it means that we can’t explain our intuitions by positing that the world is as they portray it to be. Rather it is often the case that our intuitions serve some other purpose, that we simply dress them up by assuming that they track metaphysical facts in order to justify our use of them. But there is no real need to justify usefulness; usefulness is useful. So what is useful about our intuitions regarding naturalness? Well is the gold-pyrite category a good one? Sure it is an objective division of the world that we can precisely define, but that is not all we look for in a category. We categorize the world because those categories are useful. A first order kind of usefulness is expressiveness; a category is useful because we can refer to it in order to express facts about the world. But, in that respect, every category has roughly the same intrinsic usefulness modulo the other categories that we already employ (redundancy is not very useful). So obviously that isn’t the relevant issue. However, there is also a second order of usefulness that categories can have with respect to how those categories play a role in useful theories. If the K group of properties plays a role in some theory, or better, in a number of theories, than categories that divide along the lines of the K properties are more useful to us. This is why we divide atoms by protons and take neutrons as secondary. The number of protons an atom has is extremely important when it comes to the kinds of chemical reactions it will enter into, while the number of neutrons is important much less often. Thus we divide the atoms by their number of protons because that is the division that tells us the most theoretically important facts about them.
And this explains our intuitive resistance to the gold-pyrite category: it isn’t one of our pre-existing linguistic divisions, it doesn’t give us any significant new expressive power, and, most importantly, it doesn’t play any theoretically significant role. If, however, we did discover something interesting about the gold-pyrite category as a whole, say that it participates in some interesting category of reactions or played an important social role, then our intuitions would change. Obviously we would still think that gold and pyrite were significant categories on their own, because of our existing theoretical commitments. But the new category would also have a role to play, and thus would seem more natural. Or, at least, it should become more intuitively acceptable, if our intuitions are helping rather than hindering us, because now being able to appeal to that category is useful. Sure we could say all the same things by saying “gold or pyrite” over and over, just as we could equally well express all the facts about gold by appeal to the category “atoms with X protons and 1 neutron or atoms with X protons and 2 neutrons or …”. In both cases the mental and linguistic convenience makes the division useful. (The mental convenience is especially important, because it makes the difference between being able to put irrelevant details out of mind so as to focus on what is important, and being unable to do so. Perhaps if we had unlimited mental capacity the only thing that would matter for us is expressivity, and so we would express every idea by appeal to the most primitive divisions, i.e. fundamental particles and their arrangements. But we do not have unlimited mental capacity.)
Now we can finally return to the original question. Why does “gold” designate what it does? Well originally the terms simply designated some gerrymandered, but objective, division of the world. The term existed because it was expressively useful, and since there are pressures in favor of being able to communicate people eventually came to mean the same thing by it, even though some other, altered, understanding of the term would have served them just as well. As time went on it turned out that some of the things designated by gold were rare, and hence socially important as units of value. Thus the terms altered its meaning somewhat to pick out the genuinely rare from the mere imposters (bronze, pyrite). When the discovery of the elements was made gold shifted its meaning again; tracking the scientific divisions was more important than just tracking something rare. Fortunately rarity and the scientific divisions aligned pretty well; the scientific divisions essentially ended up explaining the kinds of tests used to separate the really rare stuff from imposters (purification, density, etc) and why some substances (white gold) met those tests despite not quite looking like normal gold. And this is why gold refers to what it does: it refers to an objective division of the world that is of greatest utility, and which bears some resemblance to the pre-theoretic origins of the word.
And this theory seems to be generally true for all designating terms, not just gold. One advantage of the theory is that it allows that there can be something in the world that the term refers to, which explains how we can make discoveries that inform the way we use the term. And it also explains that reference in a way that avoids an appeal to pre-existing divisions of the world, or any other metaphysical facts, which is an advantage because language is clearly a human invention, and hence how it works can reasonably be expected to be explainable in terms of us, rather than in terms of facts that are, for the most part, independent of language.