Foundations of a Sucessful RTS

By Tom Cadwell aka Zileas

RTS is one of the fastest growing and most popular genres of computer games today. RTS has several sub-genres, which can basically be seperated into a multiplayer and single player as well as economically driven, and non-economically driven and finally between operational and tactical. For the most part, and there are exceptions, economically driven, multiplayer, tactical RTSes have been the most popular to date, and likely will be in the future as well. Most of these RTSes are members of the Dune 2 Descendant chain, and include RTSes such as Command and Conquer, Total Annihilation, Dark Reign, and the Crafts.

RTSes of this type seem to have a common set of pitfalls. The biggest pitfalls deal with the treatment of game phases. Few RTSes have managed to go past the basic build-and-rush phase (thus never leaving the early game for the most part, or just having a very linear escalation in technology and armies). Those few that have gotten past this often have difficulties dealing with the late game. The most common attempts at solutions seen are superweapons, an even bigger rush, and attrition. I define a superweapon as a weapon of immense destructive power (And cost) that can be operated out of a heavily fortified area (i.e. your base), with no additional risk for attacking. The TA bertha is an example of this. A bigger rush and attrition should be a little more intuitive concepts. Each of these solutions has their own problems, some worse than others. Another pitfall lies in the task of diversifying multiple races, while at the same time maintaining balance. The biggest dilema that I believe game developers have to deal with is the fact that they are hard-core gamers, and can't help but want to make a game that is for a hard-core gamer, which often means more features. Generally, additional visible features (i.e. buttons on the screen, additional stuff you have to do, etc) make the game more daunting to a new user, and end up reducing the marketability of the game to casual gamers, which are the majority of the gamers who might potentially buy the game. On the other hand, hard-core gamers recommend games to their casual gamer friends... Its a delicate trade off. This, and the other recurring problems in RTS, like any problems, have a number of solutions.

I feel that good balance theory, and a firm recognition that "realism" (in terms of obeying physical laws, etc) often detracts from gameplay, is important to solving these problems. I've summarized my viewpoint on these issues with a list of design principles and balance principles.

Principles of RTS Balance:

1. If race A builds up a unit mix, and attacks race B with it, there should exist a cost-effective counter for B that is available around the same time, and requires slightly less time to build than race A's unit mix. This is best accomplished by strong counter units, as well as a slight defensive advantage (which is innate to any RTS because of the firepower maximization of a micromanaged tactically sound defense formation). Along similar lines, every build order should have a corresponding weakness of magnitude related to the potential gain of it against certain other build orders. Fast expanding should be rushable, marginal teching should not be, etc.

In the case that certain attacks are only non-cost-effectively countered, there must exist an equally available, and equally powerful attack (in relation to the counters) for the second race, otherwise an imbalance will exist. Also, it is usually best for there to be several cost-effective counters to mixes, varrying with strength. The strongest counters should have corresponding weaknesses. Versatality should always be a little bit weaker in general, though a versatile unit mix needs to be a viable strategy -- Players like the idea of having to counter, but they don't like the idea of up-to-the-second reconnaisance being necessary for victory. The viability of versatile unit mixes makes this possible. It should be possible for a player to regularly, but not constantly, do recon, and learn what they need to do to optimize their current versatile unit mix. On a smaller scale, versatile units should be less powerful compared to specialist units, not only to not make versatile units superior to specialist units (which creates obsolete units -- a bad and confusing thing) but also to avoid too many unit mixes with minimal weaknesses. Sometimes, though there will be a cost effective counter to a build, the situation is still an imbalance as the "production forks" leading out of the situation are highly favorable to the race that forced the other race to build a counter -- Timing is so so key, and though 2 unit mixes may appear balanced, even in terms of fastest build time, the route you take to get them doesnt necessarily have to be the fastest build time and COULD be imbalanced.

Some game designers have tried to balance the game and diversify the races by giving certain races higher strength at various phases of the game (rush/early, midgame, endgame). This is dangerous and should be avoided if possible. By dominating an early phase, you alter the balance of a later phase by going in stronger.

2. Balance should take the form of a balance between unit mixes, not a balance between individual units. This ensures more diverse races, and generally will result in more skill differentiation as countering an attack (or making a good attack in the first place) involves more than just building a large number of one unit. I do not feel this detracts from instant addiction in the least -- in fact it might add to it when it comes to playing online. Generally speaking, each unit, or even unit type, should have a very effective counter even so so as to make unit mixing all that much more powerful. Compare tank rushing in RA, to what was seen in later RTSes and it is clear that this is important.

3. You should always consider maximum firepower concentration. Many super-long range units is very bad. The siege tank in starcraft is an example of a unit that is unrealistically short ranged (though this didnt bug any players as far as I can tell), but which is balanced due to its low mobility and not really all that long range (compared to a bertha in TA). The siege tank, on a side note, is also an example of why sacrificing realism for gameplay is often good, and has worked in the past.

4. Range and attack power should be closely intertwined. Long range siege units should always hit for very little, so as to just force a response, or slowly wear down a target. It really is necessary for gameplay to sacrifice the realism of LR units hitting hard -- Yes thats how "real' war is but I feel that this creates too many potential imbalances and generally puts too much power in one unit. It is possible to somewhat balance out a long-range unit by making it innacurate, but against larger groups this effectively becomes a long range area effect unit of normal accuracy. Though realistic, I think that this is only a partial solution, and just making the long range weapons weak in general is a better solution.

Some games have dealt with range dynamics very well to date. A strong combat dynamic of ranged, melee and various types of artillery as seen in myth 1 is very important, and has led to the success of several games, including myth. One mistake dozens of game designers have made is to make long range and medium range attacks only marginally lower than short range attacks. Because of range attrition (i.e. the damage a unit sustains while rushing up to bear short range weapons to bear), particularly with impassible terrain features, it is dangerous to make any long range attack very powerful. Powerful short range area effects are necessary to encourage unit mixing however. Weak long-range area effects may also be acceptable. Super-low rate of fire, high damage long-range area effects are potentially imbalanced however, even if long-range units logically hit very hard. This is because of the fact they eliminate skirmishing, a great component to have in an RTS, and because they tend to be devastating with map lag. Also, long-range units should not be able to skirmish well at all for obvious reasons; In other words they should never have much mobility compared to other units. There are many other issues with range and attack power, but to cover them all now would take at least 5 pages.

5. Map lag is the solution to many early game problems. By not allowing units, especially early units, to cross the map too quickly, you solve many problems with rush fests.

6. Combat formulae and the various values units have need to be very flexible, and have a wide array of effects. This makes patching and fixing balance if problems are discovered after release extremely simple.

7. Every attack should have a risk. Though cool, I feel that "superweapons" are bad as they do not incur risk. Also, a per-attack cost is not to be confused with a risk. There should always be some opportunity for your opponent to counter your attack and destroy your resources cost effectively. Even with this premise, you can still have cool, mega-destuctive weapons that carry risk. The nuke in SC is far better than the nuke in TA for this reason.

8. Endgame driving forces must be in place. I personally feel that attrition is the way to go, otherwise things tend to degenerate into a swirling powering fest devoid of strategy. Bigger isnt necessarily better -- after a certain point players find the administration of controlling 100s of units with any degree of adequecy completely overwhelming. Keep it simple -- it doesn't have to be huge in # of units to seem epic, and if you tend to have less units in general, you can also pump up the FX of those units.

Also, static defenses need a very effective means of being cracked so that a player who suffers from attrition will be unable to continue. Basically, I feel turtling should be a bad long term strategy. If it is powerful, not only will you get stalemates, but you will be reducing the excitement of the game by making very very very standoffish situations. I could discuss this for hours at end why this is the case, but this is supposed to be a very general, and very short paper.

9. Mobility is one of the most important traits a unit can have. An extremely mobile unit should never be very powerful, but these can be in the game. Fast reaction time, and pinpoint, on-demand firepower delivery are powerful tools indeed. It is dangerous to have low mobility high firepower/hps units, particularly as cornerstones of a race's units, as they will create map-specific imbalances in that their response time will be horrible on larger maps but acceptable on smaller maps -- And you have to balance their power relative to one, but no matter what you do you muck up other size maps. If you were to go with a fixed size map, you'd still have this problem with any units that have movement constrained by impassable terrain, as some maps would take longer to transverse than others.

Design Principles:

1. If it isn't fun, it shouldn't be in a game. Giving players lots of adminstrative tasks is silly. Players like action, and players like making split-second decisions. Battles are far more interesting than spending 20 minutes planning out your base and hitting your units over the head to build what you want where you want. Tedious building tasks should only be implimented for play balance reasons, or because there is no other way to keep the player busy without creating an imbalance. If they wanted administration, they'd go buy Sid's latest game. Peon pumping, or the tediousness of expanding as Protoss in SC are good examples of where a boring task was put in to encourage balance. Supply depots are a good example of a boring task put in for absolutely no good reason. A boring task is better than no task at all, but you can always replace a really boring task with something more fun I think. I can think of a number of ways the early games of various RTSes could be improved on -- too many designers are locked in the mode of, "we have to start at an early point in order to make the strategy deep" and just overlook the fact that the very initial buildup is silly. Why didnt you start with more power in your commander in TA? Why didn't you start with 7 peons in SC? Beats me.

Along similar lines, units should generally not do stupid things. Nothing frustrates a player more than their units acting idiotically and requiring constant babysitting. The AI should help the player, not annoy them.

2. Gameplay over realism. You need the game to "make sense", but adding additional realistic features, though cool on the back of a box, eventually detracts from gameplay in many cases. If you have to sacrifice one or the other, sacrifice realism.

3. The aspects of a game that appeal most to the hardcore gamer should tend to be either in the form of hidden features (hotkeys listed in the manual but not visable on the screen, combat bonuses for cover, etc), or in deep strategic depth that only comes out with lots of play. Even a hardcore gamer may be daunted by 101 buttons on the screen. A player will only discover and explore just how awsome your game is if they get hooked in the first few minutes. Basically, don't clutter the screen with buttons -- keep it to a very sane low limit.

4. Sound effects and visual affects are extremely useful for making a game instantly enthralling, while in no way affecting play balance. For this reason, they should be used extensively. Visual effects also look good on the back of the box -- in many cases way better than those 40 realistic features that might diminish gameplay do.

5. Strategic wealth should always be sought after. This is what creates the true fans. True fans will recommend your game to their friends, which means more sales for you. Strategic wealth isnt an obvious thing however. More units isnt the way to accomplish this -- complex unit interactions or a simple but elegant resource system is. Just as chess is simple yet elegant, so should an RTS be. You should keep the game simple, but make the simple parts (which should be, for the most part, the ONLY parts) of the game interact in extremely complex ways that are not readily apparent. I truly feel that this is the singular reason SC was probabaly the best RTS to date. Note that complexity only is relevant to the "game system", as in the unit costs, the ways resources work, the ways unit work -- it has nothing to do with how advanced the engine is (well for the most part -- some engine things will change this), and it certainly doesnt have much to do with how good the GFX or SFX or story are.

6. Give units a "purity of purpose". Rather than think, "this race aught to have a unit, and it logically would have these abililies", think, "what does this race need to flesh out its combat abilities in a unique way", and then come up with a unit that fits the racial "feel". Also, avoid giving units more than one major use, and if you do, make them very expensive. A spellcaster should not be a very good warrior, nor a warrior a very good spellcaster, etc. Also, don't have more than one unit per race that does essentially the same thing. This encourages unit mixing and reduces confusion for new players. You shouldn't add units that add very little to the game besides "another unit" as these just add confusion, and little else.

7. Never let a unit become obsolete. Many RTSes have made higher tech levels superior in every way. This only confuses players, and can also lead to more imbalances. It certainly cuts down on strategic wealth.

So What Does All This Mean?

Unfortunately, a lot of game designers just stick things in randomly, and never apply them to a central philosophy. The most successful games of our time have all taken that approach -- though this philosophy may or may not be widely embraced by RTS developers, I feel that -A- philosophy that is adhered to is crucial to a successful game. Those features that bug players a lot, invariably are the ones that stick out from the rest of the game, because they werent scrutinized with regards to the central design philosophy.

Everything in a game should have a purpose -- It shouldn't be there "just because it sounded cool" or "just because it would be realistic". There has to be some gameplay benefit from every single aspect of the game that the user deals with. Game design is as much an artistic thing, as an engineering problem. Just as a game should be artistically cohesive, its parts should have functions and usefulness beyond aesthetics. Everything in a game must be aesthetic AND augmenting to the gameplay.

by Tom "Zileas" Cadwell
April 2nd, 1999
Updated April 20th, 1999