Well then, I felt the inspiration to type another entry into this series on game mechanics. Buckle up, grab a coffee, hit that subscribe button and please donate to my patreon… oh wait never mind.
APM in an RTS
This game being an (MMO)RTS over really long time scales needs to take into account that APM, an integral part of RTS design and balancing is no longer a factor. APM stands for Actions Per Minute, how many effective commands you can give in a limited time frame. This has implications that are straightforward and some that are more subtle.
Now normally when you play a standard RTS like starcraft, total annihilation, command & conquer, age of empires, or their families of descendants (warcraft, 0 AD, openRA, empire earth, supreme commander, planetary annihilation, etc, etc.), you are playing a game that’s meant to last for no more than a few hours or so at most. Most matches are better measured in minutes. This means quite simply, that the game has to go from the humble beginnings of a single structure and a handful of units to a war raging across the whole map with dozens if not hundreds of units in that time.
Lets take the original starcraft as an example, being the most thoroughly analyzed of RTSs. When you start you have a “command” structure which can produce workers, and a handful of workers. So you queue up a worker to be built with your starting resources, and then send your workers to mine. Then you fiddle your thumbs and wait for the workers to have gathered enough resources to build another worker. But not if you’re an experienced player, then you start to micromanage your workers in this stage of the game.
There isn’t anything else to do, so the most effective allocation of your actions is to make sure each worker takes the nearest mineral patch and does not lose time by going for one that’s already occupied. This micromanagement of the workers collecting resources leads to slightly more minerals being available slightly faster. Its almost negligible, maybe another one of the cheapest combat units is gained all in all, but hey, that’s not negligible at all when it’s the first and therefor only one on the field.
As the game picks up, and you start building out your base, engage in some scouting and military production and fights, worker micro takes a back seat to managing those things. That is to say, the pay off you get for performing worker micro is far lower than the pay off you can get for performing other actions. The priority of micromanaging workers sits at the bottom of the priority list, well below the point you can reach on your APM budget. This pattern constantly repeats as the game continues; your first combat units you want to micro to perfection, but when they’re a part of a big army with many such units, you simply do not have the actions to spare to micro the most out of each individual basic unit.
So what happens if we play with infinite APM?
So then, lets imagine playing starcraft at a pace that it’d take days or months to complete, thereby eliminating the APM economy. This leads to a number of problems. For one, you now have time to bother with optimizing each worker’s resource gathering. It wasn’t particularly interesting with 4 workers, but by the time you’re on 80 workers, oh hell no. Thus we identify the first problem with eliminating APM: a lot of grinding, repetitive action is now not dropped from the priority list.
To an AI which is unhampered by our human slowness, such is more or less the situation, and this leads to very interesting strategies and playstyles in starcraft AI tournaments. I’d suggest anyone to have a look at those to see how the lack of an APM economy leads to very different strategies. Imagine hundreds of perfectly microed kiting units (mutalisks), or a super early rush with the terran SCV workers which have the ability to repair one another, and thus beat any other worker force as they can be pulled back and healed just in time, provided they hit before military units come into the game. Of course, “AI”s are still rather dumb, and humans with thousands of actions per minute at their disposal could probably figure them out and beat them easily.
This leads up to the second problem, without APM there is eventually a single optimal strategy. While in the above examples it might look like a wide variety of strategic options has opened up, this is always true of the starting phase of what is essentially a new game under the APM free conditions. Eventually, a single strategy will be mathematically optimal, and can be executed with mathematical precision. At best, a rock-paper-scissors gamble will arise due to a lack of information. But gambling isn’t strategy.
In the situation with an APM economy such a thing can only exist theoretically, as the execution of the optimal strategy will always involve micromanaging every asset to perfection, and no one has the time and ability to do so. Depending on ones ability to handle ones assets with the amount of actions one can muster, suboptimal strategies might be dominant to theoretically more optimal strategies, if they can actually be executed well within APM limits. (i.e. the infamous attack-move blob) Thus the second problem: stripping APM strips the aspect of execution from a strategy, narrowing the scope of viable strategies.
Lastly, in RTSs one can best queue up one thing a time. Idle resources waiting in queues to actually be spend later give you neither more economy nor more military, thus you can better spend it right away to either purpose. In RTSs this is again a differentiator of skill, where a more attentive, trained player can cycle around his queues to keep them filled one at a time, always spending resources as fast as he can to maximize his presence, while the less skilled might need to settle for longer queues (I do).
Again, in an RTS without relevant APM, the strategic principle still applies, but the associated multitasking skill has not translated. It has become just another nag every so many hours to be paying attention and add that next thing to the queue. Or settle for less efficient bulk orders. It’s not a differentiator of skill, but of constant attention to little nagging details.
Of course, I realize this game isn’t starcraft. The specifics of the problems are not the same, worker micro is not even a thing in Outscape, and so on. Yet the fundamental problems illustrated with Starcraft remain in Outscape. The grind is very real, and the optimal strategy a thing of math.
Automating the grind
I’m going to do a bit of conjecture here, and say that the problems with the snowball in A2 might not have been so much a problem of snowballing itself, but rather the amount of repetitive work that it entails, akin to having to micro the 80 or so workers of late game starcraft. That is to say: the problem isn’t so much that there is a rapid exponential curve in economic strength and empire size, but rather that this translates to an equally rapid exponential curve in soul sucking click work.
This topic has been visited before in multiple ways by multiple people, but I’d like to add from the angle of an RTS without APM. In the RTS with APM, the automation takes over for you, while being slightly less efficient than you. And this is fine for those games, yet in the context of Outscape, remember how one should still micro automated workers in the very early game of Starcraft? It’s not a lack of automation that makes players do that, but rather that the automation isn’t as good as they are.
So let that be the first point here: automation has to do as good a job as the players themselves would, or players are still nagged into doing things manually to get an edge. (Why did he win? Well, he spend 3 hours a day doing the same thing over and over again. Better do the same next time…) And without delving further into the solution of automation, let me add that this also means you have to give players a lot of freedom in how they wish to fine tune the automation, so they can adjust it to the circumstances.
However, automation in itself is not the greatest of solutions. As I frequently hear when I suggest such features here and elsewhere “why not just let the game play itself?”. You basically give the player options to interact with the game, and then you have the game handle those interactions, automation always is a little self-defeating in a game.
Redesigning the grind
With these things in mind, lets look at alternative ways to design against the grind. For that purpose, lets first look at what we actually want from a strategy game. One might argue over the definition that is to follow, but fundamentally, the fun of a strategy game is a process where one observes a situation, analyses that situation, creates plans to deal with the situation, then executes that plan, and thus creates the new situation on which to repeat the process. Then, by carefully analyzing how the various assets of the game contribute to this cycle one might be able to cut away a lot of the more grinding aspects and retain or even expand upon the strategic choices they embody.
Take for example your humble farm or other population cap raising building. What do these buildings provide to the game’s balances and the player’s strategic choices?
- Provides a resource cost for continued population growth.
- Halts population growth if not built.
- Locks up a certain percentage of population to sustain population, with upgrades costing resources that can improve that ratio.
- Can potentially be targeted to reduce local population over time, should such a precision strike ability be implemented.
I feel the above list pretty much covers everything farms have on offer from the perspective of strategic gameplay. And yet, on a planet with some 500,000 population I’ll be building some 20, manually checking when to build them nearly every time the planet is checked, not to mention that I have to upgrade them one by one due to them going offline during the process.
Of these things, what do you need the level of simulation for that is provided by having individual farm buildings? If something pops to mind, it worth the price in grind that it entails? Normally, I’m all in favor of little details like individual buildings or people and so on, but only when they also mean something on that individual level. Here farms only have meaning as the aggregate of their numbers. So maybe we have an avenue here to get rid of a lot of grinding repetition, while maintaining the strategic choices and influences provided.
One obvious way to do such is to make combine individual farms into an agricultural sector. One unique entity per planet. For example, let us instead of a farm building imagine an agricultural sector.
The agricultural sector has a pool of invested resources (credits, beron, power), which determines the % of population that needs to work in the agricultural sector to sustain the whole population. Depending on the size of the current population in relation to the size of “resources invested” pool, this percentage rises or falls, and stops further growth when that leads to labor shortage. Equally, it stop improves when it hits a certain minimal value. We can either manually commit resources to the agricultural sector, take them away at a loss, or potentially set an investment rate equivalent to the population growth rate. Damage can be translated into a reduction of the pool size. Techs can improve the minimal value, and the costs for upgrading can be mimicked by exponentially rising the costs the lower your ratio gets.
The result is that you have just one object in the game to provide the relevant balances and strategic choices, and only needs to be interacted with when those choices are made. Unlike a gazillion farms, these sectors at least have the scope to be different to one another, particularly when you start connecting them to local circumstances or events. This furthermore has the benefit that you can now make your balances continuous as they no longer need to jump per-building, avoiding threshold issues. Regarding production in whole sectors also gives scope in to very simply implemented diminishing return schemes or other balance tweaks.
To wrap this bit up: paradoxically, it is precisely because you have all the time in the world to do every little thing that can give you an edge, that every little thing needs to scrapped from the player’s interaction with the game in these circumstances. Given that every little thing that might give you an edge can no longer be ignored when you have time to do every little thing that might give you an edge, it’s suggested to redesign sections of the game for every little thing which gets stuck on repeat.
What happened to execution?
Moving on to the second problem, that of an optimal strategy. Of course, this problem is not caused by a lack of APM economy, but it is aggravated by it. APM is not the only factor here that is influenced by having this much time, I’d add focus of attention and simply time to analyze to the list. All these things greatly interfere with theoretically optimal strategy.
In an RTS, there typically are things happening everywhere, and yet your camera view can only be at once place a time. I call this the focus of attention. Make a distraction here, and so you miss the airdrop there. Winning by directing where your opponent is holding his camera at the crucial moment. Being afk can be much the same, with the crucial distinction htat being afk also means one does other things that makes life worthwhile. One should not be punished for having a life outside of Outscape.
Likewise, time to think is necessary to respond adequately. So one can deny his opponent the time to do this. Keep up the pressure, keep him reacting, he will do things which are not very clever sooner rather than later, and then you pounce.
APM has been well discussed, and taken together these three resources complicate and add skill to the loop of strategic gameplay I mentioned earlier: observe the situation, analyze the situation, create and execute a plan, repeat. Their interference makes the technically suboptimal strategy viable, and occasionally the downright impossible. The air unit might swoop in on the unattended anti-air unit while it is on move command. It’d be a fatal error in judgment, unless it works.
One might argue this is a downside of RTSs, where it devolves into a clickfest, where the strategies and counters are a-priori known meta, all that makes a difference is clicking faster, and I wouldn’t necessarily disagree. I too would like an RTS where the clickfest factor is eliminated. But then, we must contend with the question of how to make up for the lost strategic possibilities. Because the strategies in an RTS are usually a-priori known meta, the decision tree more or less mapped out in advance, and all that’s left is the execution. These problems are magnified, not overcome, without the clickfest.
Maintaining strategy in the face of perfection.
Truth be told, I have no grand all-encompassing ideas here either on how to avoid a straight line to a strategically solved meta. I do however have some ideas, and things that might serve as inspiration.
For example, take a look at correspondence chess, where people play chess by mail or email, taking days or weeks until they receive a move. It’s super slow chess. It’s turn based and therefor not the same, it’s also a perfect information game (i.e. both players know everything), making it harder to maintain strategic depth. Unlike tic-tac-toe however, chess or go isn’t a solved game.
I could write a far longer article on why I think chess is a good game, but fundamentally, it’s a game where every piece can interact with every other piece via a limited resource of tiles. Thus even though there are only 32 pieces, the fact that they can all interact means that the web of possibilities is quite large. The limit of 64 tiles on the chessboard then forces these interactions to occur. One can use pattern recognition and planning to make this advantageous, yet the sheer volume of possible futures introduces an element of uncertainty.
If we take this line of thinking back to Outscape, it must be noted that most pieces, be they fleets or planets, have a very limited scope of interaction. Fleets are specialized; there are fleets to combat other fleets, fleets to mine, sweep mines, infiltrate, detect, bombard and invade, but all these specializations don’t really come together outside of the necessity to have the (counter-)ability when you need it somewhere. While I can understand how this might appear more strategical, as you need to create fleets with a purpose, it also constrains all fleets in their scope of interaction. You’re not only snipping most threads in the web of possible interactions, but thereby make optimizing each ship for its purpose a foregone conclusion.
I’d suggest making the differences between ships more one of degrees than differences of kind. Think of the enterprise in star trek: the next generation, it can ship cargo, do scientific exploration, fight in combat, etc. One could go for such all-round ships. Or one might still specialize, but having that choice makes things more connected. Consider each purpose that a ship can have a “tile” in chess which pieces might traverse with more or less ease, but every piece could.
Lets talk about uncertainty, which I think is a basic considering, chess provides this by obscuring the final result in deep web of possibilities. In my present play through of A3 I had to adjust my expansion/colonial strategy several times due to new information becoming available (corruption, the cost of shipyards, etc). This is fun, as it makes me re-evaluate my strategy and adjust my plans. But now these mechanics are known, they will not provide uncertainty again. Not knowing what was in my neighboring systems provided uncertainty too, but now I’ve scouted everything a 100LY radius, that’s gone too. Only my fellow players can provide uncertainty now. But they too are reasonably predictable as the only real change, war, is not really profitable for them as they cannot hold on to my planets anyway, and bombarding them away is just a waste of resources.
An element of randomness in what exactly your empire gets to work with might work very well here as well. A degree of randomness in the rapports or information you get might work. Techwise, I quite like the sort of game where you cannot click a tech to research, but get them more or less randomly as in alpha Centauri. Stellaris’s variant on this system gives you a choice between 3 to 5 options.
But why not go for a full random system of tech upgrades, maybe influenced by the circumstances around your tech centers? I.E. a tech center on a planet filled with farsu mines simply has a per hour chance to upgrade some aspect your farsu mines or mines in general (build cost, power drain, population cost, mining rate, deep mining rate, whatever else), which you can then implement empire wide. Add in other factors such as the availability of credits to the likelihood of what gets upgraded. Likewise science ships increase the odds of discovering something beneficial to upgrade your ships.
I realize this specific proposal is much too far reaching to be implemented at this stage of the game, but I bring it up to highlight how the upgrades would compound into player and play through unique game objects. Add the ability to trade those unique objects with other players for a constant re-evaluation.
Lastly, a “problem” of both EVE and outscape is the vastness of the galaxy. One is not forced into unwanted interactions mostly, as one can avoid the enemy, and anything that can catch up with out is necessarily weaker for the ability to overtake you. Supplies might provide for a limited boxing ring; olzine being a prime candidate. Supply depots being points of interests for contest.
Aside from supplies one can imagine other “connector” mechanics or objects, which connect a majority of the game objects in a single “space”. For example, characters or other boons that only be used once or one place at a time make for good connecting mechanic, provided multiple options for their deployment are reasonable.
So to wrap this bit up, removing apm and other time based constraint from strategical decision making, you’re creating a situation in which a given strategy can be optimally executed. To avoid a straight descent into an optimal meta, I’d suggest to re-evaluate for more interaction between game objects, more spaces in which game objects connect, and more inherent variability between game objects.
Playing RTS in very slow motion leads to several problems:
- Repetitive actions becomes the norm; as small scale optimizations are easily the most numerous.
- Using automation is only viable if it performs as well as the player, and can take into account changing circumstances as well as the player, or at least be set as such.
- Redesigning to eliminate the need for such small scale actions might be more prudent.
- Strategically, a guarantee of perfect execution of a strategy limits the scope of strategies to a predetermined meta.
- Thus there can be no easily deduced optimal strategy. This can be done both by creating a more connected web of possible interactions, as well ensuring such an optimal strategy varies wildly with changing circumstances.
Alright, that was way too long.
I wont do it again.
For a while.