I always look for intensity. Starcraft kind of leveled off after a while, just I’m not sure how you would make rock-paper-scissors intense. Depth is related to intensity. I think many would agree that Starcraft didn’t get that deep, though deeper than Warcraft. You can’t just keep adding unit types because you run into the limits of humans (breadth), but the limits of programs are very different (depth).
Earlier you said roll-your-own AI. I don’t mean to domineer, it has potential too, and games would go a lot faster-- entire tournaments could be over in a second. We would all examine the replays after, lick our wounds, and debug.
On the downside, if you allow teammates to share routines, the distinction between races vanishes. You know, not necessarily, I take that back. Didn’t Starcraft have a shared-control mode at one point?
Well, what tree am I barking up? Does your game resemble Starcraft?
I have no idea, I’ve never played Starcraft. But no sci-fi theme if that is part of what you mean.
The Wikipedia article covers a lot my observations actually.
en.wikipedia.org/wiki/Real-time_strategy
Even with a dev editor, the “click fest”, a la the article, potential is still there, though there is the aspect of preparation. Do you want to waste time, kill it, pass it, or spend it?
bigfoot,
Thank you for letting me share this idea. I think it influences the design of a game because computing power can trump some possible strategies. You don’t want a guy renting an hour on the Amazon Cloud when everyone else is running pentiums. Instead, I’m picturing writing a program to “do the clicking” for you. Perhaps we would use a command-line interface to issue orders, and not do much in-game editing. Perhaps some users would use voice rec. software and issue verbal orders.
I can’t wait to see such a game in the future, and if you’re the first to make one, it’s that much sooner I’ll get to play it!
You know what it reminds me of, is a turn-based programming game.
I thank you, and everyone else as well, for sharing. I hope to rid the game of click-fest and it will actually play rather slowly because I hate being rushed.
I’m not gonna be the first to make it, that really calls for someone much more advanced than I. Sorry. 
I am trying to eliminate as much needless and aggravating micromanagement as possible though.
Turn-based is a whole other argument, perhaps my favorite. Since it is the only computer genre that I’ve regularly bought games of. 
Sorry, if I don’t seem to respond to everything mentioned. I’ve been a bit preoccupied with bug stomping (debugging is such a boring term) lately.
UPDATE 3-27-10:
Been debugging, or rather, filling in gaps in my knowledge that were causing crashes through coding mistakes.
Found various problems that were already drastically lowering frame rate, got it back up to about 50, even with path-finding. Quite happy about that.
Experimenting with different user-interfaces in the quest of what is simple and smooth to use, but fits the game perfectly.
If you’re willing to entertain my above idea a little bit more…
You don’t want the player’s side to be a solved problem. In starcraft, it is, just click faster. In chess, it is as well, just use deep blue.
If you had a computer / automated assistant to “do the clicking for you”, as above, would starcraft / command&conquer etc. just be boring? Or would you still get fun from examining weaknesses, your base, your opponents’ bases, etc.?
Would you always use the same routines, that is would one always be best? Or could you constantly be improving on them?
I never stopped entertaining ideas. I read them, I think about them… I think some more, and then I might do something. People tell me I think slowly, I just say I try to examine all points of the issue.
A computerized assistant to micromanage small details, while the player can do what he wants. Same routines? I suppose you could constantly improve upon them, though I have little inclination to make it in game editing. AI players should be able to do all that, perhaps the player’s micromanage routine should start out very similar to the AI’s, and then the player can gradually improve upon it.
NOTE: The only rts that I have played, or the first two AOE. BFME 1 and 2, War of the Ring, and very briefly Halo Wars. (all played because of friends, or I found them for a great price) So, please explain any other rts game references. Thanks. 
As I recall one of the AOE versions had a Trebuchet unit. That’s similar, roughly, to say, siege tanks and lurkers, the units in Starcraft I was using for examples, if you want to scroll back.
For example, you could give a trebuchet group “standing orders”. When the opposition gets too heavy, they should pack up and roll out. And the ground troops should provide cover. And workers should work on repairing them.
Those kinds of things are especially difficult to get a computer to do. You almost always need a human “eye” to intervene at some point, but only for one judgment call. So, let the human just make that, and leave the rest to for-loops.
This is basically a Tile-based-RTS. Although I’m quite proud of myself in that the unit movement doesn’t quite show that. 
So AI isn’t gonna be terribly difficult in that matter, but it is AI, so it is gonna be hard just because it can.
I don’t have to scroll back, I quite remember and have been thinking it over since I heard the names. Siege units in the back, troops up front, repairmen close at hand. There is no comparison to the human eye, it wins every time.
I keep wanting hex grids.
The thing that AIs can’t do is adapt-- diagnose what went wrong with its strategy and compensate-- either during the game or between games. I’ll argue it’s a full process of diagnosis per se, no less.
You get to notice a lot more stuff if you free up the clicking.
You could compare it to Agile Programming-- the analogy being Agile Starcraft. In Agile Programming, teams of two write code; one person is looking over the other’s shoulder, and mentally keeping track of things like which identifiers he’s in the middle of changing etc. The advantage is the team saves time on design sessions because the team doesn’t need everything in complete detail before they get to work.
What if you had a friend looking over your shoulder in AOE?
I think hex grids are coming out in Civilization 5, if you happen to like turn-based. 
EDIT: For curiosity’s sake, I’m taking the evening off and messing around with the possibility of hex-tiles. Wouldn’t take me that much to adapt the current setup to it (I always try to make everything as adaptable as possible). I’m at least looking at it, very much doubt I actually will though.
I have never seen an AI that adapts well. Never.
I really never click much, I’ve gotten pretty good at finding bottlenecks and stationing all my troops there and the AI will send their forces piecemeal, easy as pie. I’ve waited three hours on AOE2 just watching the AI. I notice a fair amount, at least in my humble opinion.
The only thing wrong with that analogy is that I’m doing this mostly by myself. All programming and a good part of the art, though I do have a fair amount of help on the art.
I would play multiplayer AOE2 with friends. We set the map to no fog-of-war to make it interesting, we watched each other.
I have Civilization 4. Promise not to laugh when you see it “running” on my machine though, if you can even call it that.
I think it would be easier to do hexes in the general case first. Consider each tile as a circle such that each point “buckets” into the circle it’s closest to the center of. Then your game model could be on the real plane, instead of a discrete grid. It’s not obvious that each square falls into integer-number categories; some tiles might have more neighbors than others where they all didn’t have the same number and weren’t in the set 3, 4, 6.
I think Pong and Tron would be good starting points to examine principles of A.I., as they’re simple and it’s easy to make them too hard, although the latter does interfere if we want to study adaptation.
Well, for my part, I always look at A.I.s as an obstruction and a chore, that is, playing against them.
Another good starting point could be to consider putting an A.I. on a human’s team. This occurs in Civilization 4 in the ultimately simple capacity of asking a war ally to attack one city, and the A.I. grants or denies the request.
To AOE’s defense, the AI task is taking second burner to the tasks of modelling and rendering. I wonder what a full-time AOE would look like, that is, the Deep Blue of AOE.
I very much doubt I’d laugh. For a few weeks I played Civ4 with an OLD graphics card. All the ground was black, the only way I could know a tile was by mousing over it and reading the description. That took me forever and hurt my brain. Not to mention that eight minute loading time for a small map. But now I’m running it very well after getting my current computer.
I’m still considering hexes, the only real changes I would have to make to my code would change tile placement slightly for every second row. Change one function which returns all the surrounding tiles. And a few other small things.
An obstruction and chore pretty wells sums up my feelings about AI in most games.
I’d love to play Deep Blue on AOE. Even though I would lose, I would lose to a beautiful attack.
EDIT: No hex tiles. I made a backup first, and then fooled around with the game and I just don’t think it will fit in well. But fooling around with it did bring to my attention some things I could improve on, so I’m going back and changing the tile system around a bit.
UPDATE 03-30-2010:
Unit movement/pathfinding/targeting is under a complete redo. Before I’m even done I with it, it is going considerably faster than before. Not to mention its looking a whole lot better. Halfway a mix now between navmesh/tiles and using a*.
In the process of finishing up small details on my “design documents” which is close on twenty pages of random(to other people) notes, drawings, ideas, questions, etc.
Updates are sporadic, because it is a bit difficult to find enough time to work on it every day.
NOTE: A release will not be available until semi-completed status, or help is needed for some problem.
UPDATE 04-04-2010:
Nothing important, but a bit more object-oriented structure.
A bit of optimization, a steady 60fps on a core2duo 2.4ghz. Even with a bit of pathfinding.
Afraid that’s really about it, had a bit of extra work I’ve had to do first before I could work on this.
I think “clickathon” has a more satirical ring to it than “clickfest”. </off topic>
UPDATE 04-08-2010:
Buildings are starting to take shape.
A small menu is up (I mean small, it took me just three minutes to do).
I’ve got fed up with restricting tile-based, so I’ve entirely redone that to “3D” non-tile-based movement.