User:Tom Morris/The Definition Delusion

A commonly recurring anti-pattern in discussions on Wikipedia is the desire for a definition to solve all arguments. You see this frequently in regards to deletion debates over notability: I saw an RfC a while back that argued that we should come up with a definition of what makes a notable civil aviation accident, based on the number of people who die in such an accident. You also see this kind of argument come up in discussions over what should or should not be included in articles, in the lead of articles or in the scope of articles (as in, whether that content should be in one article or moved into another article and then pointed to with a "see also" link from a section of the main article).

As someone trained in analytic philosophy, the desire for clarity, preferably with easy to apply necessary and sufficient conditions, is understandable. If it is possible, simple, clear distinctions and divisions are preferable to vague hand-waving (or many far, far worse things). But this demand for clarity can lead us down utterly irrational quests for clarity where none can be found. Instead what ends up happening is a particular point-of-view gets enshrined as the One True Definition even though that definition does not reflect the complexity of the subject matter at hand. The pushers of the One True Definition then come up with ad hoc explanations for the bits that don't fit their analysis, and use the consensus of the definition to beat others over the head with.

On defining

In his book on Informal Logic, Irving Copi lists five purposes of definition. It is worth recounting them here.

1. Increasing vocabulary.
2. Eliminating ambiguity.
3. Reducing vagueness.
4. Theoretical explanation.
5. Influencing attitudes.

The first two are pretty clear. Reducing vagueness is pretty important in practical circumstances: Copi gives the example of a decision by the 1966 Supreme Court of North Carolina deciding that yachts are not "motor vehicles" and that the state's 3% sales tax does not apply to yachts but a 1% tax specifically on yachts applies instead.

Theoretical explanation is a strange one, and is mostly restricted to science. Think of water: defining it as H₂O tells you what it is, but it also provides a theoretical explanation. Water is, at core, two hydrogen atoms and one oxygen atom. This is useful for theoretical understanding of chemistry, say, even if the more informal definitions of "liquid substance that comes out of a tap, freezes at 0°C and boils at 100°C" etc. may in practice have exactly the same extension in the real world as the more theoretical definition.

The final reason, obviously, is what people do when they aren't abiding by WP:NPOV. Think of the religious evangelist who says something along the lines of "Christianity boils down to being a good person". Sure, being a good person is pretty important in religion, but the reason he is making that statement isn't so much to reduce vagueness but to influence attitudes and get people to think religion equals good (and non-religion equals bad).

Why definitions don't matter

The simple act of defining something isn't actually very important. In contemporary academic philosophy, arguments about definitions are considered a bit of a waste of time. This is because people can define things however they want. To reuse a classic example from the philosophical literature:

Grue: An object X satisfies the proposition "X is grue" if X is green and was examined before time t, or blue and was not examined before t.

Nelson Goodman defined the term "grue" in the 1955 book Fact, Fiction, and Forecast in order to give an example of what he termed the "new riddle of induction", a problem with inductive reasoning beyond the problems David Hume discovered (see problem of induction). I won't discuss Goodman's argument here: feel free to read Grue and bleen (and Goodman's writings) to find out more.

Grant the meaning to the word "grue" that Goodman gives it. It may be the case that some object is grue in a particular context.

But in ordinary day to day living, the value of saying that something is grue is irrelevant. For the purposes of writing an encyclopedia article about some object which may satisfy the truth conditions of the sentence "X is grue", the question of grueness is completely irrelevant. If you are writing an article about emeralds or sapphires or water or blueberries or ultramarine pigments or the symbolic colours of political parties (the Green Party, "reds under the bed", "Blue Dog Democrats" etc.) the precision or otherwise of the word "grue" is irrelevant. What matters is that all these objects have an actual colour like blue or green.

The value of defining a term is not whether or not things match the definition, but whether the analysis provided by the definition is useful in aiding others in understanding what you are talking about. Consider the variety of terms used to describe political affiliations. Some are very vague and generalist like "liberal", "conservative", "left-leaning" or "traditionalist", while others are very specific, like "minarchist", "social democrat", "economic populist" and so on. Precise definitions here are useful for some purposes: if you are trying to clearly delineate minute differences amongst socialists, more precise terminology is needed than if you are trying to determine the broad ideological status of a broadly left-of-centre political party like the Labour Party (UK).

At a certain point, definitions cease to matter. If you wish to understand, say, the British Labour Party, an argument about whether they satisfy the strict conditions for being "socialist" or "left-wing" (however you choose to define those) will tell you a lot less than just reading the history of how the party came into being, how it evolved, its leadership, policies and actions in government and in opposition. Definitions are tools to think with, but there's a lot more to thinking than just defining stuff.

Context specific definitions

Definitions are used within contexts. Consider the term "Christian" or "Christianity". It should be roughly clear what being a Christian is, but it really isn't. At best, if you are trying to come up with a global, context-free definition of Christianity, about the best you can say is "some kind of religious inclination that involves the figure of Jesus Christ". I hate to appeal to authority here, but I've read a fair amount about different branches of Christianity, different sects, heretical movements and so on, and this is about the most accurate definition I can come up with that encompasses Christianity as a whole. (But I'm not a professional theologian, so what the hell do I know?)

And to any Christian or to anyone who wants to have a meaningful discussion about religion, it is a completely useless definition. A huge number of major religions "involve the figure of Jesus Christ". Islam has a view on who Jesus is, and most Christians will want to deny that Islam is somehow a branch of Christianity. Nontrinitarianism probably passes muster, and many Christians will strongly dislike any definition of Christianity that includes things like Unitarian Universalism or other denominations.

Christianity is something more like an identity for many and when pushed, an individual person's proclaimed Christianity may boil down to little more than "well, I'm not a Muslim or a Hindu or a Buddhist or a Jew, and I was born in a Christian country, ergo I'm a Christian". Kierkegaard described this kind of attitude under the label of "state Christianity", and it still exists today. In the UK census, plenty of people have been known to tick the box "Christianity" because they are unthinkingly "C. of E." (Church of England), because, like eye colour or dialect or football team preference, this is just something one inherits from one's family. Do they—as the evangelicals demand—have a personal relationship with Jesus Christ? Only at Christmas—and they have just as personal a relationship with Rudolf the Red-Nosed Reindeer.

But in some sense they are Christian, if by Christian you mean "descended from a line of people who broadly identify themselves as Christians rather than another religion in a country that is historically a majority Christian country". The question is whether that definition is useful. If you are a fundamentalist preaching a sermon on who is going to heaven and who is going to hell, you are not talking about the same thing. Arguing about whose definition is "right" is beside the point: in one context, the word and definition is serving one purpose, and in another it is serving another.

A lot of the time, it is far easier to drop arguments about definitions and refer to the things that form the basis of those definitions.

Take human beings. The discoveries of natural science lead most reasonable people to believe that human beings and apes share an ape-like common ancestor. In that sense, human beings are apes. In another sense, you don't often see human beings locked up in cages in your local zoo with a little plaque next to the cage describing them as an ape. The definition of the word "ape" is context-specific. In the broad brushes of evolutionary history, human beings are apes–quite smart apes, capable of writing encyclopedias and inventing robots that can wander around on the surface of Mars–but with a large quantity of similarity in DNA and phenotypes to apes.

That kind of definition is quite irrelevant in an article on American football. American football is not being played by apes, it's being played by human beings–admittedly human beings who consider it wise to smash their heads against each other at high speed while chasing a ball. Within the context of discussing American football, whether football players are apes in the biological sense is as irrelevant as whether their jerseys are grue in the sense Nelson Goodman meant.

Beyond definitions

So what do we do instead? Firstly, we spend less time worrying about definitions. If a talk page discussion reaches the point of saying "we need a simple, clear definition of what it means for something to be X before we can say so in the article", strongly consider whether or not that is actually a useful way of spending one's time. It probably isn't. Consider whether games of definition tennis are actually routes for point-of-view pushing or, indeed, original research. Go back to Copi's list: eliminating ambiguity and reducing vagueness is fine, but when there is inherent ambiguity or vagueness in the topic or in the sources, trying to make it go away is beyond the scope of Wikipedia's role. We are here to reflect the world's knowledge, not point out the conceptual flaws in our attempts at knowledge.

Instead, work out what properties or characteristics of the subject one shares with the concept for which a definition is sought, and the properties or characteristics it does not share. Then seek reliable sources which discuss in detail those characteristics. Definitions are simply a handy device for bundling together commonly shared properties. When the usefulness of a definition breaks down, you simply unbundle the properties and discuss them individually. (But beware: original research concerns loom here too!)

Finally, be prepared to accept vagueness. Vagueness is inherent in our language–think of the Sorites paradox–and maybe also inherent to the underlying world. It is the job of Wikipedia to reflect rather than fix that. Reducing the amount of vagueness in the world is simply out of the scope of the project.

See also

• Fallacies of definition

If you enjoyed this essay, try The Reliability Delusion, wherein I rip into our holy reliable sources policy.