Certain genre terms become so widely engrained in our language that we are easily fooled into believing we know what 'genre' means - which is to say, we learn to interpret genre in a particular fashion. In fact, genres are particularly ill defined - genre terms are a class of words which define by inclusion certain instances from certain media. However, the same is broadly true for all nouns - a noun is like a genre term applied to reality. When we use the word 'tree' we refer to those things which we consider a tree - which may include, for instance, plastic trees, virtual trees, some bushes, and even logical trees.
It is no secret that my philosophy is strongly influenced by Wittgenstein. To be honest, this isn't saying much, as Wittgenstein is widely considered to be the greatest philosopher of the 20th century, so many living philosophers claim his influence. Prior to Wittgenstein, much philosophical discussion was dragged down by logical assumptions that simply weren't as universal as they felt to the individuals touting them.
I often summarise Wittgenstein's views by stating "the meaning of a word is how it is used", but this is merely a convenient aphorism. Elijah T. Beaver provides a nice summary of Wittgenstein's views in this regard:
The contention that meaning is use in Philosophical Investigations [one of Wittgenstein's books] is related to his idea of ‘family resemblance’. A family resemblance is a vague grouping of things that are together because of their similarities, but have no one thing in common. The example Wittgenstein uses is that of games; all games do not have one underlying trait in common, but are grouped together under the heading ‘games’ because... he "can think of no better expression to characterize these similarities than ‘family resemblances’." He compares this to a rope; in that there is nothing within a rope that runs throughout the entire length, yet all the tiny fibers woven together are still parts of the rope. The uses of a particular word are much like a family resemblance in this way. Not all the uses of any particular word are necessarily going to have something in common with all the other uses (excepting, of course, that they are referred to with the same word). They will, however, be related to one another in a vague sort of way. This resemblance is not by any means perfect, and so the specific meaning of the word is still dependent on its particular use and not on anything that tightly binds all the uses together under the word.
Wittgenstein’s belief that meaning is encapsulated in the use of a word is by no means an exact or all-encompassing idea. It is merely assumed, as are many of his beliefs, to be a generalization that holds true much of the time.
What Wittgenstein says of words is particularly true for genres - which exemplify the notion of a family resemblance category, and are inherently resistent to logical definition. Nowhere is this more clear than with genre terms such as "RPG". The diversity of views as to what constitutes an RPG is vast, and attempts at logical definitions fail because the term is overloaded - it refers to both tabletop RPGs and CRPGs, for a start. But, in point of fact, what tabletop RPGs have in common (collective storytelling) is very different to what CRPGs have in common (formalised and stylised progress mechanics, as an example). Attempts to make the term cover both types of games face an uphill struggle. This is why I and others employ the term CRPG for Computer Role-playing games (or DRPG, as Corvus does, which is presumably for Digital Role-playing games).
But of course, even making this split doesn't solve the problem, as we still have within CRPGs the Western style of CRPG, the Japanese style of CRPG, not to mention Hack styles and strat-RPGs...
Many people, and I have been one of them in the past, are deceived into thinking that they can derive logical categories from what genre terms mean to them - and then apply those logical categories in discussions with other people. Now it is sometimes possible to apply logical categories to words - there are, for instance, logical definitions for measurement terms such as 'metre' and 'litre'. But some caution is advised.
If you investigate measurement terms, you will find that the definitions were originally quite arbitrary - and have only become standardised over time. For instance, the metre was at one point defined as the distance between two scratches on a bar made of a platinum-iridium alloy kept Sevres, France (near Paris). Nowerdays, we define metre in terms of the speed of light in a vaccuum, because we currently believe the speed of light in a vaccuum to be a constant (it may not be, though... we can't be certain with only a century of data).
But genre terms are not measurement terms - they resist logical categories. There is no international standard RPG held in a locked room which we can turn to in order to establish what is or is not an RPG.
Attempts to take family resemblence categories and make them into logical categories can produce quite ridiculous results. Warren Spector once told me that his team insisted that the character in Deus Ex had to have a name that the player could enter because members of his team would not accept it as an RPG otherwise. They believed that if the name of the character was fixed, it would become an action adventure, and cease to be an RPG. I have always found this somewhat ridiculous... Japanese RPG games rarely allow the player a choice of names, but are widely accepted as RPGs, and even in tabletop games the player can have a role assigned, as in the Indiana Jones RPG in which the player plays Indiana Jones.
Successful genre terms (i.e. terms whose usage has caught on), like 'survival horror' which began as a marketing phrase for 'Resident Evil', catch on because they ring true easily to a wide number of people... In some respects, one could say that 'Resident Evil' is the platinum-iridium bar of the genre term 'survival horror'. But of course, which of the game's features one chooses to denote the genre is still open to interpretation - is it a factor of mechanics, structure, or of atmosphere? The answer varies from person to person.
There is a great temptation with words that rely on family resemblence to assume that "majority rules" - that when sufficient people believe something, it becomes true. This is misguided. It would be fairer to say that once a majority definition has crystalised, it is harder to make oneself understood without playing the same language game.
In this regard, it is worth noting that language 'drifts' over time. 'Factoid' used to mean an unverified or inaccurate piece of information. However, more recently it has taken upon a different meaning, that of a brief and somewhat interesting fact. I can already deduce from what I hear on the radio that the meaning is drifting so far towards the second definition and away from the first that it is only a matter of time before the second definition - currently seen as a 'usage problem' - becomes the de facto meaning of 'factoid'. Eventually, the original definition may disappear entirely (or at least, become categorised by lexicographers as an 'archaic' meaning).
In social and behavioural sciences, the terms 'emic' and 'etic' are used to distinguish between the persepective of the individual (emic) and the perspective of the observer (etic). Robert Anton Wilson was the first person I read who was writing about 'emic reality' versus 'etic reality' - that is, one's personal reality, versus the (strictly hypothetical) reality that exists outside of our perceptions. (For shorthand, I think eMic = my reality, eTic = "True" reality). We assume etic reality is out there, but we can never know it - except through our perceptions, from which we form our own emic reality.
So many arguments occur because people assume that words (and genres) have a meaning that applies in etic reality - but words only have meaning in our personal, emic realities. It just so happens that, by chance and the channeling effect of media and culture, our emic realities are sufficiently close that we can communicate meaningfully - most of the time.
We are all free to make and use whatever definitions of words (and by extension, of genres) we wish - although the extent to which these words and genres overlap with other people's definitions will vary. No two emic realities are the same. We should feel free to invent genre terms on whatever grounds we wish - with the understanding that other people may not accept these terms, and that if they do not, we are unlikely to be able to convince them. It's all part of the rich pageant of language which is the substructure of each of our private realities.