Game design is an interesting topic...does a designer have a specific methodology in mind when he designs a game or set of rules? Does he have a design philosophy?
There are many game designers that I admire, for different reasons: Frank Chadwick, Sam Mustafa, Rich Hasenauer, Buck Surdu, Phil Barker. It is sort of like movie reviewers - you may not agree with everything they write, but you value their opinion - either because it agrees with yours or you know what to avoid because they recommend it!
Chadwick I admire because of his thorough, detailed period knowledge. While I don't necessarily agree with how things are done in his rules, I completely admire his knowledge and design decisions based on his philosophy. If you ever get a chance to hear him present at an HMGS East convention, I urge you to attend.
Mustafa - I admire his rules because they are usually unique; a different way of approaching a problem. He's also a very good writer, and his meanings are pretty clear.
Hasenauer - all you have to do is say "Fire and Fury"; that says it all. One of the cleanest sets of rules out there.
Buck Surdu - the "Look Sarge No Charts" series of rules has a number of unique concepts; always interesting to me.
Phil Barker - the legend! What an evolution from the WRG ancients rules up through DBA/DBM. To me, DBA was such a radical departure from his previous work that this alone earns him the respect due a unique designer.
If you look at each of these - they had a specific vision in mind when they wrote their sets. I don't think they spent much time consciously worrying about defining those visions, they just knew what they wanted to have in their games. Is there a mechanic that is historical? Unhistorical? Valid? Invalid? Unique? Most likely, they knew where they wanted to get and just played around until they got there...I think there are not too many "eureka!" moments in game design...most games are 5% inspiration, and 95% persperation. It takes a LOT of development time to clean up mechanics, get the text clean, make sure there aren't inconsistencies.
My personal design philosophy can most easily be summed up as "Comparative Recognizable Patterny".
Recognizable Patterny means that you HAVE to include things that "feel right" for the period. For example, you HAVE to have lines, columns, and squares if you're doing a tactical Napoleonic game. You HAVE to have unit weight/usage classifications if you're doing an ancient or medieval game, you HAVE to have different gun weights and capabilities if you're doing WW2. If you're going to do a grand tactical game, you HAVE to emphasize fog of war and command decisions.
Comparative means that I select an average baseline for the "normal" troop capability, and then you rate/adjust everybody else up or down from there. Figure out the %'s that you want to have for success at any task for an "average" unit/individual/weapon, and then you're set. As an example - a Sherman's 75mm gun is better than a short 50mm gun, but worse than an 88 (duh). Crude example, but you get the idea.
I prefer results that force players into decisions - whether that is movement or combat. Random sequence decks are my preferred method of limiting the choices a player has in controlling his force at any point in time, but there are certainly other methods to do that. I prefer combat methods that incorporate morale results - I find separate morale test procedures to be tiresome and outdated. I want the players to be placed in the role of commanders and leaders - they should be involved in rallying, directing, leading...most definitely not involved in the process of testing a morale level of a unit. There is no single rule process in a set of rules that I dislike more than a separate morale test process. Barf.
Movement - I enjoy randomization of movement, but with that movement being influenced by command and situational environments. Fixed movement I find to be boring and predictable. That has nothing to do with being "historical" or "unhistorical", but its just a personal choice. I've experimented with Horizon Movement in Les Grognards 2nd edition, and think it has some merit depending on the game period, scenario, and game group. It certainly moves games to conclusions quickly, but it doesn't give some players the "bit by bit" development of a game that they enjoy. However, it is a great game design exercise.
I also firmly believe in simple, simple, simple. It is WAY too easy to just add a modifier, or a process, to handle another bit of information. I avoid "double jeopardy" - never modify the same thing twice...for example: if you think rough terrain should slow troops, you could modify the move rate, or you could modify the leadership modifier for movement rates. BUT - DON'T do them both! That just adds time to the process, proves that you can think up modifiers and clumsy processes...avoid that!
I base my designs on a LOT of reading and research to identify "recognizable patterny" to make sure the game feels "right". While there is no absolute ruler of "historicity", you know what "feels right" when you play the game. You don't need a PHD to find that out, you don't need somebody to tell you how right or wrong this or that game process or concept is - if you like it, you'll play it. If you find a game's processes irritating or to produce results you don't agree with, you won't play it.
I don't worry about placing rule design into the category of "art", "science", or "history". Its a little bit of all. One of the most overlooked, vital aspects of rule writing is just that - the WRITING. Rule writing is not like writing a novel - its much more a technical writing exercise, once the rules concepts and processes are established. If the writing isn't clean, clear, and precise, it doesn't matter what concepts are being introduced - the result is too garbled to be useful to anyone.
It ain't rocket science!