D&D is about playing a character, rolling some dice, and hoping that they fall in the PC’s favour as the DM and players push miniatures around a grid like a game of chess in which even the queen may not capture the pawn on the first attempt. The numbers are added up and the consequences of the roll determined. The game can be reduced to a robotic level. Indeed, a computer could be programmed to run through an entire campaign in the blink of an eye. But a large part of D&D is role playing beyond pushing a human fighter or eladrin witch around a grid. The players can infuse their characters with personality, which adds an extra dimension even if nothing much else about D&D is even slightly realistic.
In a recent Legends and Lore column, Nod to Realism, Monte Cook discussed the veneer of realism in D&D. He covered such things as weapon damage, rates of movement, and hit points. These little details add complexity to the game, but ultimately, they do not really make D&D more realistic.
As I write up various adventures, I look at realism in a slightly different fashion. I deliberately make battles vague because I know that if a goblin hits a tiefling with a sword, and doesn’t kill the it in the first place, the blow will almost certainly incapacitate the creature. I don’t want my characters to get slaughtered with such ease. (Besides, as I’ve noted before, writing up battles is tedious, which is why Haven of the Bitter Glass has ground to a halt.) In other words, I’m trying to avoid reality in battles.
On the other hand, there are peripheral elements of D&D, which might be described as decorative, where a little realism would not hurt: ecology, architecture, and language and place names.
In Thunderspire Labyrinth, the PCs must place certain items in rune circles to release the Guardian, a green dragon, from a pit. But why would a cold-blooded reptile willingly allow itself to be confined in a pit in a subterranean complex on the off chance that some adventurers might just happen by to release it?
“Because obtaining [sentient mammalian prey] can be difficult, however, greens subsist mostly on woodland creatures, such as deer, bears, and wolves.” (Draconomicon, Chromatic Dragons, 2008:37).
Thus, the ecology is all wrong because the green dragon ought to be out in the forest and not in some dank, dark hole. Some kind of golem would have made more sense.
A more extensive example of ill-considered diversity can be seen in The Lost Caverns of Tsojcanth. The place is populated by monsters which seem to be waiting for some tasty PCs to drop by and be eaten. The monsters all seem to live separate lives with no sense that they would all be in a predatory-adversarial relationships with each other. Some monsters could not leaves the caves at all. For instance, the ettins in the Lesser Caverns would be unable to leave their cave because the gap is too narrow for a large creature. In fact, there wouldn’t actually be enough room for the pair of them.
The architecture used in D&D typically lacks cohesion. Obviously the game has to take place on a 2D surface (although overly keen DMs might build cutaway models which allow a true third dimension), but the sites where the action takes place are still stuck at the level of Doom with a disparate series of rooms strung together with meandering corridors. While dungeons need have no particular vertical unity, towers and other multi-storey buildings ought to have.
Again, Thunderspire Labyrinth contains a dreadful example of bad architecture. The Tower of Mysteries has no vertical cohesion at all (or any toilets and kitchens, or a sufficient number of bedrooms) with the walls at one level failing to match the walls at another. The Chapel of Shades should be a separate structure or, at least, annexed to the main tower, but it doesn’t belong on the third level of the tower. Another bad example is Graystone Fortress in Den of the Destroyer, which looks like a dungeon rather than a fortress.
For practical reasons, everything has to operate on a 2D plane in D&D, but from my perspective, the authors of the various adventures and campaigns ought to go out and buy themselves some Lego and see what issues arise from such bad building design; or just take a look at Morrowind, Oblivion, and Skyrim.
Language is a complex thing and for the most part, there’s no need to create full-blown conlangs for D&D because they can never be truly used unless the DM is going to give the players some sort of language puzzle which can be solved, say, on the basis of frequency. (An example of that can be found in the footer text in some of Eoin Colfer’s Artemis Fowl books; a frequency analysis would soon allow the text to be deciphered.) About as complex as language gets in D&D is an alphabet (always 26 letters just like English) and sometimes a list of words. (I know of Drow and Draconic word lists at least.)
Nonetheless, to do language properly, it’s important to know what sounds it has, its syllable structure (which often determines the nature of the writing system; languages with predictable syllable structures tend to use syllabaries [e.g., Japanese]; languages with more complex syllable structures tend to use alphabets [e.g., European languages]), and the nature of any suprasegmentals and how they are assigned. The morphology of the language may be less important unless part of an adventure involves real examples of the language. The syntax may have some bearing on the formation of place names because of the general relationship between headedness and compounding. Here, too, morphology may be significant.
Some of the best known places in Faerûn are onomastic nonsense. Waterdeep might be a name in the south of France, but in English it’d properly be Deepwater (cf. Broadwater and Freshwater). Feywild doesn’t make sense even when the elements are the right way round (and what was wrong with the Otherworld or Faerie?). Neverwinter isn’t a well-formed name unless someone wants to argue for folk etymology; nor is Shadowfell (if “fell” here means neither “deadly” nor “mountain” ). And as for the excrescent “the” with Shadowfell, Feywild, and Elemental Chaos, I’ve attributed this to a quirk of some varieties of American English which employ the definite article where other varieties don’t.
English place names are frequently composed of two elements where the first is descriptive and the second identifies a particular feature, whether it’s natural or artificial. In other words, these are typical English compounds (i.e., modifier + head). For example, Ox-ford, Cam-bridge, Oak-ley, etc. First elements can be the names of people (e.g. Ayles-bury, Ayles-ford) or groups of people (e.g. Notting-ham, Birming-ham, Canter-bury) or animals (e.g. Shepp-ey, Shep-ton, Shep-ley; Ox-ley) or activities (e.g. Mil-ton) among various things. The second element may be a town or habitation (e.g. -ton, -wich, -ham, -bury, -by, -chester, -thorpe) or a natural feature (e.g. -wood, -ey “island”, -ley, -den “valley” ).
While English retains the old Germanic principles of compounding and noun modification, other languages reverse the order. French is schizophrenic in this regard. In northern France are places like Neuville (Newton) while in the south the same name is found as Villeneuve. Left-headedness in the Romance and Celtic languages is more thoroughgoing than it is in English, hence Waterdeep in Welsh, Dŵrdofn, would probably not even raise a Welsh eyebrow. Compounding in such languages will be noun + adjective or noun (of the) noun.
This is worth noting for wholly invented names, but whether it’s Deepwater or Dŵrdofn, the construction needs to be applied consistently throughout. Similarly, the sound system of the language in question also needs to have been codified for consistency. It’s best for each distinct speech sound to have one rendering, but there is a continuum between English with several ways to write many speech sounds to Finnish, which is relatively consistent, to Chinese in which the correct pronunciation of pinyin transcription requires some phonological knowledge. (The language does not have too many ways of rendering the sounds in it, but too few for pinyin to be transparent to non-native speakers.)