A few years ago, I helped out a family friend, Craig Besinque, in designing and playtesting a block-based wargames, which ended up being called Hellenes: Campaigns of the Ancient World.
Craig is pretty well-know in a certain element of the wargaming community - which is like being well-known in the RPG community, except that wargamers tend to be a bit more community-oriented than most RPGer's. He's designed games like WestFront and EastFront for Columbia Games, so it was really interesting to learn about board game/wargame design from him.
In brief, here are the things I learned working with Craig.
1) Designing wargames is unlikely to make you rich. You are working in a niche market, with a fairly small customer base (as customer bases go). They are a reliable base, though - willing to spend money on quality product. This is presumably also true for RPG's - something anyone wanting to design RPG's should heed.
2) You had best be good at math. Craig has a Masters Degree in Mathematics from (I believe) UCLA. He can do fairly complex probability in his head. I suspect that it's a good thing he's not into gambling, or he would be cleaning up on Pokerstars.net.
3) Game balance is all-important. Moreso in this context than some others, since Hellenes was designed to be a fairly short 1v1 wargame. Balance in this context is very similar to what balance means in chess - there should be no sure, or even preferred, route to victory. If there is, the game is "broken" and considered by most people to be unworthy of being played. Which will kill your sales (or even your chances of being published) - see point 1.
4) You can design whatever you want, but if you stick to certain material constraints, there is a much better chance that your game will be published. In this case, Columbia games had certain pre-packaged numbers of blocks and decks of cards. If you used a different number of blocks in your game, it meant the game cost more, which would be a factor in the publisher deciding to release it. Likewise, if you use cards, try to stick with the same number of cards as a standard deck of cards, for the same reasons.
To further explain this point, we should examine a different game - Railroad Tycoon. This game has a ton of different blocks, tiles, markers, plastic trains and whatnot with it. It's a pretty good game, but now out of print. I think that's probably because all the bells and whistles (hehe) cost more that it's worth to produce the thing.
5) You must playtest. Then, playtest more. Then put it out there and get others to playtest it. That is the only way to get good game balance - see point 3.
6) If you want to design a historical boardgame, you must start with the historical part.
It's nice if the game can reflect historical realities, but you should pick what realities you reflect - they should be ones that create interesting possibilities and trade-offs. Guns or butter decisions are what make wargames interesting.
Ideally, players should be making decisions that reflect those that the actual historical sides would have made - in the case of Hellenes, the Athenians need to decide how much of their fleet to take out of the city, whether to focus on attacking Spartan coastal provinces, how much effort to spend putting down revolts, and how many resources to apply to land armies vs the fleet.
7) Limiting resources is a good thing. When you can't do everything, you have to make decisions about where to allocate a very limited pool of resources. This makes each decision a difficult one, requiring much deliberation. Combined with point 6 - it means you can never do everything you want, and you're always making interesting decisions about what you can do.
Many RPG's use one or more of these design principles. Interestingly, older versions of games like D&D tend to use less of them, despite the fact that they are closer, genetically-speaking to those old wargames. Specifically issues like game balance, playtesting and limiting resources don't seem to have been foremost in the minds of the designers of say, 1e Dungeons and Dragons. It's a hybrid game system, and the designers seem to be focused more on the overall experience than applying the lessons of wargame design to the new system.
In more recent years, things have swung back around. 4e, for all it's flaws, pays much more attention to 2, 3 and 7. It would have been nice if they spent more time on 5, though. Curiously, it's frequently panned by fans of the older editions - which is kinda weird, since it owes much more to the initial inspiration for D&D than many of those old editions did.
Ultimately, helping to design a boardgame was a great experience - it's really changed how I look at both RPG's and computer games. I feel like I have a much better view of the decisions that were probably made in the design phase that resulted in the game I'm playing, which makes it easier to mod games, and strangely, easier to like each system for it's own merits/flaws. To understand something is to lose the fear of it, after all.