I don't usually employ miniatures in my RPG play. We ceased that when we moved from CHAINMAIL Fantasy to D&D.
I have nothing against the use of miniatures, but they are generally impractical for long and free-wheeling campaign play where the scene and opponents can vary wildly in the course of but an hour.
The GW folks use them a lot, but they are fighting set-piece battles as is usual with miniatures gaming.
I don't believe that fantasy miniatures are good or bad for FRPGs in general. If the GM sets up gaming sessions based on their use, the resulting play is great from my standpoint. It is mainly a matter of having the painted figures and a big tabletop to play on.
- Gary Gygax on ENWorld, 2003 (http://www.enworld.org/forum/1263669-post6.html)
I've heard in the past that Gygax didn't use miniatures in D&D, but it's interesting to hear it in his own voice. "We ceased that when we moved from CHAINMAIL Fantasy to D&D." This makes sense in a lot of ways.
One of the things it helps make sense is how the rules for use of miniatures in AD&D don't (to be frank) make single lick of sense. Consider how miniatures don't physically fit on a map at 1" = 10 feet scale, and the truly crazy stuff on DMG p. 10 (make maps at 1" = 3⅓ feet). The reason? Well, Gygax had ceased actually using them as soon as the RPG itself came into existence. The ranges and movements are copy-and-pasted from Chainmail, but he wasn't actually using them directly. In other words, the use of minis became a vestigial, unusable appendage in D&D.
Consider stuff like this. A fireball in Chainmail & OD&D has a fixed range of 24". In AD&D that gets changed to a caster-dependent range of 10" + 1" per level. Repeat that for every spell's range and area in the entire book. Why the enormous increase in complexity (requiring math on the fly just to find any spell's area and range)? Especially when Gygax wasn't using miniatures or a game map in any way himself?
Knowing how Gary would write, I can almost hear how he'd answer this. "It's self-evident that more powerful casters will have greater efficacy, and rules for minatures were included for the kind of person who would enjoy that sort of thing." Something like that. Kind of dodging the fact that AD&D is shot full of complicated rules, everywhere, that he neither used nor playtested; looking good on paper but not playing out so well. (Funny, too, that he's licensing and promoting "OFFICIAL ADVANCED DUNGEONS & DRAGONS miniature figures" [DMG p. 11] simultaneously with abandoning their use in his own play.)
Now, there are other things that this does helps rationalize. One is that it's an excellent point that, in a game with lots of random encounter tables, you really would be hamstrung if you have to go running for different types of miniatures every time a new encounter pops up. Another thing is the need for AD&D rules to now specify random contacts in combat. "Discharge of missiles into an existing melee is easily handled... Assign probabilities to each participant in the melee or target group according to sheer numbers." (DMG p. 63). "As with missile fire, it is generally not possible to select a specific opponent in a mass melee. If this is the case, simply use some random number generation to find out which attacks are upon which opponents..." (DMG p. 70). That sort of stuff.
But the one thing this egregiously overlooks is the interaction of area-effect spells (fireball and all the rest). If melee is an entirely abstracted, Pigpen-like dustup, how do you determine who gets hit by an area-effect spell? Everyone, friend and foe? Whoever you want? Just the bad guys? Random determination? To-hit rolls or Intelligence checks? In all of OD&D and AD&D, I'm pretty sure there's not a single line addressing this question, leaving it entirely ambiguous.
A critical history of D&D would include the following -- Start with Chainmail historical mass rules at 1:20 scale (1 turn = 1 minute, 1" = 10 yards); this includes catapult-fire where players declare the range shot without measurement. Then Gygax develops man-to-man combat, including jousting and the fantasy supplement (using the moves and ranges from 1:20 scale, but never addressing what the new scale is or what should change in that regard); a wizard's fireball simply refers back to the catapult rules.
Now OD&D comes out, and in large part it refers back to Chainmail for combat. "Special Ability functions are generally as indicated in CHAINMAIL where not contradictory to the information stated hereinafter", stuff like that (Vol. 2, p. 5). Of course, the game creator himself is not using miniatures anymore. Questions of scale are given only the most cursory treatment: the combat turn is still 1 minute (fixed from Chainmail's 1:20 mass combat), and not until Vol. 3 are we told, "In the underworld all distances are in feet, so wherever distances are given in inches convert them to tens of feet." (Vol. 3, p. 8). Perhaps that's all the attention you need to the issue if combat has been entirely abstracted at this point.
If I had been more observant, a few years ago when Gary was still with us and generously answering questions in several different Q&A questions online, I really wish I'd asked him this: "What do you do in D&D to adjudicate the effect of area spells like a fireball, et. al?" Are we to assume that OD&D's reference back to Chainmail Fantasy, which in turn references back to Chainmail mass combat's catapult rules, requires declaration of shot range? (In neither of the former cases is it explicitly stated.) Or are we to assume that since miniatures are no longer actually in use on a map, that the determination is done by caster fiat or some random method?
Once I noticed this, it seems funny how much ink is spent in AD&D covering the new hit-the-caster-lose-the-spell rules, and never once addressing who-gets-hit-by-the-spell in its new, purely abstract combat system. Take-away here is two things, I think. (1) Gygax never actually played the game with miniatures at the alleged 1" = 10 feet scale (or 1" = 3⅓ feet, or the stated ranges for moves, missiles, and spells, or anything else), and (2) Area-of-effect spells are left entirely unaddressed in classic D&D, and definitely require some novel, independent adjudication on who gets hit by each individual DM.
Addendum: Another Q&A post by Gygax rules that targets for a sleep spell would be randomly chosen, so perhaps that aims us in a suggestive direction. ( http://www.enworld.org/forum/1972519-post68.html )
Virel: Say a sleep spell is cast at a group of ten characters... Can the caster specifically select the six creatures or six levels he or she wants to be effected?
Gygax: No. Six of the 1st level NPCs would be affected at random.