The major problems for complex multi-dimensional social science concepts is incoherence, often hidden by the fact that they are also vague. Analytically, precisifying can demonstrate we have incompatible intuitions about the meaning of complex normative terms. Simple vague terms can be precisified with âcoding decisionsâ. Vagueness differs from ambiguity. Ambiguity occurs when a term is used to mean two quite different things and can be handled by the subscript gambit. Power is neither vague nor incoherent. We can identify a simple sense underlying all accounts of âpowerâ. Ambiguous usage concerns the extension to which the simple term is applied.
1. Vagueness and ambiguity
Some terms are ambiguous and some terms are vague. Some are both. The word âchildâ is ambiguous: it can refer to âany offspringâ or to âimmature offspringâ. It is relatively easy to disambiguate this term. Any confusion over the use of the word âchildâ should usually be easy to overcome. One needs merely to point out, for example, that when Martha said âCharlie is my childâ, she was referring to an adult and not a juvenile son. However, it is not always so easy to disambiguate. âPowerâ, as my title claims, is an ambiguous term; and it is a term that has attracted a great deal of conceptual controversy.1 I shall argue that underlying all accounts of power is a simple idea, and that conceptual dispute concerns the relevant extension to which the simple term is applied. Conceptual dispute over context can be solved by disambiguation, elucidating the normative disagreement.
Vagueness is a rather different, and philosophically more interesting problem. A term is vague when it is not clear whether or not it applies to some examples. A simple term that is vague is âtallâ. It can be obvious in some cases that the epithet âtallâ correctly applies, and it can be obvious when it does not apply; but there might be many cases when it is not clear whether it applies. A man who is 6 feet 4 inches in height is surely tall; a man who is 5 feet 4 inches is surely not. But what about someone who is 5 feet 9 inches? Or 5 feet 10 inches? At what heights, precisely, do adult men move from being small, to medium, to tall?
Many vague concepts are simple. When they are simple we can handle vagueness in much the same way that we can clear up confusion over ambiguous terms. For ambiguous terms, we can adopt what David Chalmers (2011) has called the âsubscript gambitâ.2 If a term is ambiguous over two different concepts â âany offspringâ or âimmature offspringâ â we can simply mark each usage with a subscript such as childI and childO (where I = immature and O = any offspring). Chalmers claims that many seemingly important disputes in philosophy are merely verbal. What he means is that there is no real dispute over the nature of the world, merely over the correct extension â the range of applicability â of a term. Once we realize that two people are giving a different extension to a term, we can overcome the verbal dispute with the subscript gambit, and get down to the real issues about the nature of the universe.
The subscript gambit can work for ambiguous terms, and can be used for simple vague terms such as âtallnessâ. The issue is where precisely we draw the line demarcating where someone moves from being short to medium to tall. We can make what, following social science practice, I have dubbed âcoding decisionsâ (Dowding and Bosworth 2018). We decide that small men are those up to 5 feet 6 inches; between 5 feet 6 inches and 6 feet they are medium; and from 6 feet up they are tall. We might justify our usage by the distribution of heights in society, choosing the cut-off points at easily measurable heights (full rather than partial inches) given the actual distribution. Disagreement can be marked by subscripts. In other words, the coding decision is not completely arbitrary, but has no âright decisionâ at the specific interval chosen. What we have done with our coding decision is to specify the extension of the term as we are going to use it.
Some people claim that there must exist a single precisification for vague terms because it is our representation of the world that is vague, not the world itself (Williamson 1994). They hold that there must be some point at which people are tall, or rich, or jolly, and it is just that epistemically we cannot know where that line is. I agree that the world is precise, but I do not think it follows that concepts are. Concepts apply to types of things and, at the margins, what we include as token examples of a type depends upon all sorts of factors. The world is precise, but our language is not precise for all levels of granularity.
Gareth Evans (1978) proved that if there are vague objects there cannot be vague identity statements, and since obviously there seem to be vague identity statements, there cannot be vague objects. This means that vagueness is semantic indeterminacy (Lewis 1988). Semantic indeterminacy can most obviously come about when we categorize a set of items into sets or types. A concept is an attempt to define such a set or a type. The first problem for concepts concerns those items for which we have concrete examples, such as chemicals, or species. The problem is that our definition (the intension of the concept) does not seem to describe every item in the extension. Whilst this might not seem a problem for chemicals (though at the molecular level it is more problematic than might first appear), it certainly applies to species. This is the well-known issue with types (Wetzel 2009, Dowding 2016). A second, perhaps deeper problem is for concepts that are abstract and normative.3 For concepts with concrete examples, we typically begin by describing token examples to give a description of the type, to allow for further identification of token examples. This then might lead to further descriptions, allowing for a general description of the concept. That is, we find an object, describe its type, then look for other examples. For example, we describe water first by its manifest qualities, and then scientifically by its chemical and atomic properties.
However, for theoretical and abstract concepts, including simple ones, we might begin with the description (though not in complete ignorance of how it would be applied) and apply it to aspects of the world. This is how we might think about âtallnessâ. We define tallness and then decide how to apply it to the category of adult men. For more complex theoretical and normative concepts, such as liberty, again we define the concept and then apply it to aspects of the world.4 If we look at the history of such terms, we find, typically, that they start off simply, then become more complex as people consider the normative qualities of the term as it is used.
Our concept of liberty has changed drastically over time. The first meaning of liberty was something like âbeing free from the tyranny of other peoplesâ. It was applied collectively to a people (Greeks), not to individuals. Then in Greece it became associated with being a citizen of the (partial) democracy that then existed. Freedom was a legal category associated with controlling government. Then Zeno and the Stoics associated the term with some inner characteristics of people. These same ideas were repeated first in Rome, and then much later in medieval Europe, with the idea of self-rule, freedom from the interferences of others, from the government or as non-domination (see De Dijn 2020). Of course, how we define such a term in part depends on what we see in the world, but for normative terms the boot of analysis, so to speak, is on the foot of intension, whereas for scientific terms the boot is on the extension. Rather than finding examples and then describing them, with normative terms concerned with social life we describe what the term entails and then look for examples. Or we look to create examples with our moral conventions and laws. To be sure, as we study the world, we re-examine our concept, but still the intension is where the action is.
In science, where we have vague concepts, either we make coding decisions or we eliminate the vague term and replace it with another. In the social sciences we often turn qualitative terms into quantitative ones for statistical analysis. For example, we might code a set of events as examples of insurgency and non-insurgency. We will have to make decisions on precisely what is to be included in each of these categories. In most instances the coding will be clear, but there will be marginal cases for which decisions have to be made. One might be able to argue for including a particular example in either category. However, as long as we are consistent in our decision-making, and as long as there are not too many marginal cases, any individual coding decision should not have a big effect on our overall analysis.
At other times we do not use the vague concept at all, but eliminate it from the analysis, only bringing it back to summarize our findings. Tallness is an example of doing this in the social sciences. Height (in males) has been correlated with various social and economic characteristics. For example, Melamed and Bonzionelos (2012) show that taller men tend to be promoted more often than shorter ones, and Loh (1993) shows they tend to have higher starting salaries, whilst taller candidates tend to be more successful than shorter ones in US presidential elections (Stulp et al. 2013). However, these studies donât divide men into âtallâ, âmediumâ or âshortâ. Rather, they conduct regressions using precise height, along with other variables, and are thus able to conclude that, on average, taller people tend to be more successful. The vague term is useful for summarizing the findings; it is, indeed, more accurate as a prĂ©cis than reporting the results themselves. Saying âtaller people tend to be more successfulâ is a more accurate portrayal of the findings than giving the precise estimates of the regression as a claim about the world in general, since these estimates are based on samples taken in specific places and times. The vague term can be entirely accurate for the world, whereas the estimates are only approximately true. So, for simple vague concepts, we can either handle them much like ambiguous ones, with coding decisions, or we eliminate them in our scientific analysis, then bring the terms back for efficient and accurate reporting of our estimations.
I have argued, however, that many of our moral and political concepts are not simple (Dowding and Bosworth 2018). They are complex and multidimensional. They are also vague. Like simple vague terms, at times it seems obvious to what they apply and to what they do not. At times it seems obvious that the citizen is free and the slave not; or that the people of this democratic nation are broadly free, whilst the people residing in that dictatorship are not. However, the precise boundaries of how free this person is in relation to that person, or one society in relation to another, is often not clear. There are times when the term âfreeâ seems obviously applicable or obviously inapplicable, but there are some cases when it is not clear. Dowding and Bosworth (2018) argue that when we try to precisify complex, multi-dimensional, normative terms such as âlibertyâ (that is, we make coding decisions), we discover that our precision simply uncovers conceptual inconsistences.
The problem is that we have intuitions about different aspects or dimensions of the concept, and our attempts to bring precision reveal that these intuitions conflict. We might be able to coherently precisify each dimension of the concept, but we find it impossible to put all the dimensions together into a coherent whole. For example, we find that when we want to maximize the value of one dimension, that process conflicts with maximizing the value of another dimension. Any way of trying to optimize over the dimensions bumps up against other intuitions that we have. This process ought to be familiar to most moral and political philosophers. We define a concept, and then someone comes up with an imaginary case that suggests the definition is defective, so it is redefined, only to face a further imaginary counter-example. The problem comes into clearest focus when we think about how we might measure the extent of some concept under different conditions.
So, at least some of our cherished moral and political concepts are complex and multi-dimensional and also, we discover under analysis, incoherent. The idea of âincoherenceâ here is ...