Starcraft - Cybernetics and Control Systems

The video game Starcraft and its impact on gaming in general cannot be understated. Originally designed as a "Warcraft but in space" (implying its similarity to the earlier title by Blizzard), Starcraft went on to become one of the first asymmetric Real Time Strategy (RTS) games in which each playable faction had significant variety in playstyle. Previous strategy games like Warcraft and Age of Empires had various factions, but they tended to only have slight variation on a set of basic units. Factions in Starcraft however had entirely different sets of units that they could deploy, making each matchup dramatically different in how it might play out. Strategies that worked against one faction were useless against another thanks to the army that they could field. This alone made each matchup different and gave the game more variety.

However such dynamics are common in the shrinking RTS market today. The existence of three different races was not the reason as to why Starcraft became the first multimillion dollar electronic sport (esport). While Starcraft's presence at the first World Cyber Games event in 2000 may have been due to it being a novelty, its continued presence and its massive growth in viewer shows that there was more to it then just it bringing some novel dynamics to the RTS genre. The secret to Starcraft's success was in having a skill ceiling far beyond that of human capability, while at the same time having highly dynamic gameplay that has yet to be replicated in another RTS title (in my humble opinion).

Easy to Learn, Impossible to Master

Even compared to RTS games that came just a couple years later, Starcraft has an incredibly clunky control scheme. Just moving units around in Starcraft is difficult - the most notorious example being the Protoss Dragoon, a general purpose ranged attacker that suffers from pathing issues that make it difficult to control.

Dark Green fails to properly command his dragoons over the narrow bridge and as such loses their entire army to the tank fortification of Teal without killing a single unit

Yet a skilled player who can properly manage such units can get tremendous value out of them. Dragoons may get stuck on one another when controlled poorly, but when a skilled player controls them they are a cost effective unit that can serve many purposes thanks to having strong attributes.

Being skilled at Starcraft means more then just moving masses of units effectively around the map however. Skill at commanding units in the middle of battle can also lead to dramatically different outcomes. A classical example is that of Terran infantry versus Zerg Lurkers. In a straight up battle Lurkers almost always come ahead, as they are capable of dealing splash damage in the form of line spine attacks that shreds the cheap and fragile infantry.

However Terran infantry are quite fast, whereas Lurkers require brief setup time before they can attack. This allows them to either outrun the lurkers or engage them in such a fashion were their splash damage is negated.

White Terran overruns the poorly positioned Lurkers of Yellow Zerg using effective positioning and then uses the open space of Yellow's base to outmaneuver the remaining Lurkers and secure a massive lead

Such examples of skilled unit movement are not the exception in Starcraft, they are the rule. Pretty much every game will involve fights that enable one player to gain a decisive advantage over the other if they employ superior unit control or tactics.

However it is not just fancy unit control that defines Starcraft. Mechanically Starcraft is a game that encourages not just skillful control over individual battles, but also your entire army, production line and economy. Beyond the micro-engagements you must control an entire system that sustains the forces and armies you field.

Such a system means that the attention of players is split between controlling forces out on the map and staying on top of their production, tech advancement, scouting of the enemy and securing new resources. This division of attention means that players suddenly have options when it comes to conflict. For example you might be able to skillfully maneuver your infantry to cost effectively kill my Lurkers, but if I secure more resources and remember to build units, I can technically come out ahead despite losing the battle.

Footage of a professional playing. Notice how they quickly move the camera around the map, instantly giving orders and making decisions

Finally the attention of players is stretched further by the existence of guerrilla tactics. Even if the enemy has an overwhelming force the weaker player can potentially make a comeback by attacking the enemy where they are weak. Workers are an important part of a player's force as they collect the resources required to produce units and are therefore a natural target for guerrilla tactics. Using airial transports to move high damage units to where players are mining allows for massive returns on investment if you manage to catch the other player off guard.

White Protoss sneaks in High Templar, powerful, slow spellcasters and uses their Psionic Storm ability to destroy large numbers of Red Zerg's workers clustered around a resource patch

All of this pressures the player, forcing them to manage many different variables, while constantly weighing up risks. Their attention is demanded by multiple sources and they must constantly make calls and perform difficult tasks so as to try and get one up over their opponent. Managing such real-time complexity requires many, many years of training - like most competitive games Starcraft requires years of practice and many hours of the game a day. Such mastery does not just mean quick reflexes and the ability to perform tasks on autopilot so they can focus on more advanced strategy, it also means they understand the game as a system and can manipulate it as such.

Closed Systems, Information Processing and Decentralization

Before I draw analogies between a game of Starcraft and systems within the real world I want to make one thing very clear. A game of Starcraft is a closed and bounded system. Outside of the metagame (the assumptions that players bring with them about how the game of Starcraft will go that informs how they play) and hacks that give a player an advantage the game system is effectively closed off from the rest of world. This means that unlike the real world Starcraft can be mapped out by human/artifical intelligence and can be understood fully. Reality on the other hand is infinitely more complex meaning that skill at Starcraft does not instantly translate to military/management prowess (unfortunately for the South Korean state).

That being said, for those of us who haven't spent hours each day training at the game like professional gamers, the space presented to us in a game of Starcraft is far more difficult to navigate, so it may as well be an open system. A professional gamer with countless hours of practice will have instincts that allows them to correctly respond to a given situation far faster and far more accurately then someone without such training. Such knowledge cannot be explained but rather must be learnt through practice. Knowing instinctively what to do in a given situation is, by definition, something that cannot be taught but rather something that must be learnt.

The importance of using implicit knowledge to make snap decisions when dealing with highly complex unpredictable systems was highlighted eloquently by the economist Friedrich Hayek in his famous essay, The Use of Knowledge in Society:

If we can agree that the economic problem of society is mainly one of rapid adaptation to changes in the particular circumstances of time and place, it would seem to follow that the ultimate decisions must be left to the people who are familiar with these circumstances, who know directly of the relevant changes and of the resources immediately available to meet them. We cannot expect that this problem will be solved by first communicating all this knowledge to a central board which, after integrating all knowledge, issues its orders.[1]

While Hayek was critiquing central planning, his critique also describes the limitations any human player will have when it comes to Starcraft. Skilled players will be able to make the right decision within the right timeframe giving them an advantage over unskilled players. However even the most skilled Korean will hit a limit when it comes to decision making due to time constraints. Making important decisions while under considerable time pressure increases the chance of a misplay or ignoring vital information that may have informed a superior decision at that point in time. Even the imposing Google DeepMind AI which has superhuman ability at the game will eventually run into this problem of being unable to calculate you expanded the game space to include more variables in terms of map size, unit availability, resources to manage, enemies to battle and so on. Adopting heuristics to simplify calculation can enable the professional player to make a snap decision in the middle of a high pitched battle that the novice could only make if the game speed was decreased significantly to where Starcraft effectively became turn based instead of real time.

These time constraints on decision making also reveal how various forms of interfacing with systems play off against each other. The Team Melee mode enables multiple people to control a single player within the game. This allows multiple people to control a single army with a single stock of resources as opposed to normal multiplayer game in which each teams of player fight each other, each has their own army.

While this is an interesting mode to play with friends, the real beauty of it shines through when we consider how skilled players have used it to various ends. In his autobiography the gaming professional BoxeR (aka Lim Yohwan) describes how he used the mode in the early day of the games competitive scene to practice for professional matches:

In the case of practicing with the team melee method, because two players are controlling one race, there is almost no lapse in resources and production, as well as control of units. When practicing with this method, one can almost never win during practice, no matter how good and original the strategy may be. It is very hectic, as one will produce units while the other uses guerrilla attacks, and still the other will expand. While barely even scouting properly, when one sees the enemy base operating perfectly without a single break, the critical feeling is quite intense.[sic]

Losing like this again and again during practice, it is possible to accumulate diverse styles of attack. Naturally, there were those that worried that one could lose confidence from losing, but I rather liked it more. Practice matches are just practice matches. Though I lost during practice, I desired victory from the actual battle.[sic][2]

Such an analysis fits within our understanding of how systems make decisions. In order to deal with the complexity that underlines any larger system agents within that system must intentionally prune unnecessary data so as to be able to gain a comprehensible picture of reality in which it can plan its next move. However this pruning process will remove vital information that may be essential for making correct decisions.
In spreading the cognitive load across several minds a more accurate picture of reality can be obtained and more attention may be paid to significant events. Such a method of obtaining more accuracy is in line with the principles of cybernetics (the science of control systems). By spreading the cognitive load across multiple actors the force is more capable of dealing with variety and changes in the environment, as predicted by Ashby's Law of Cybernetics - that only variety can manage variety. The diffusion of responsibility comes with significant benefits in the time constrained environment that is a game of Starcraft. A mind, be it artifical or human has limits to how many calculations it can do in a given instant in order to make a decision. In distributing the load across multiple agents more appropriate decisiosn can be made because the agents are suddenly free from having to focus on particular problems.

We can see how such a control mechanism might work in reality when we look to the turn of the millennium analysis of network-centric warfare. Thomas Hammes describes network-centric warfare as:

An information superiority-enabled concept of operations that generates increased combat power by networking sensors, decision makers, and shooters to achieve shared awareness, increased speed of command, higher tempo of operations, greater lethality, increased survivability, and a degree of self-synchronization.
In essence, NCW [Network Centric Warfare] translates information superiority into combat power by effectively linking knowledgeable entities in the battlespace.[3]

The RTS genre is a vulgar simulcalcrum of such an organizing principle. Outside of having a centralized commander, every unit in faction knows exactly what every single knows, they can, if commanded correctly execute brilliant military maneuvers that would leave an army commanded in a traditional manner in the dust. The network-centric approach to warfare (and any type of organizing) is superior, but it directly challenges the hierarchical command and control model that has dominated both state and private organizing since the industrial revolution. The nature of bureaucracies means that novel ideas about organizing are difficult to implement due to their inherent conservatism. Again Hammes states:

Our advanced information systems are still tied to an outdated, hierarchical organization that slows the dissemination of information. Although specific high-priority commands receive near real-time intelligence, most commanders must submit their intelligence requirements up the chain of command. Each level validates, consolidates, and prioritizes the requests, which are then fed through the centralized staff system to task the assets that will actually collect against the requests. The information is collected, passed to another section for analysis, then put in the form of a usable product, and finally disseminated through the same cumbersome system. Thus, the premier benefit of the Information Age--immediate access to current intelligence--is nullified by the way we route it through our vertical bureaucracy. Not only does our bureaucracy delay the distribution of the intelligence products we develop, it actively discourages subordinate units from tapping into the information themselves, via the Internet. The result is a limiting of the variety and timeliness of the information available to our decision makers, from the strategic to the tactical levels.[4]

Likewise management in Starcraft (and pretty much every other RTS game) is tied to this traditional style of management. However unlike real-world organizations however Starcraft is a closed system with relatively few variables and much lower stakes. Inefficiencies that result from having centralized control can be overlooked. This is why despite the organizational advantage several players have when controlling a single army when up against single professional gamer, the professional has certain advantages that the CEO of a large corporation or the commander of an actual army simply can't have due to the nature of the environment in which they operate. A professional gamer can engage in practice that is 100% identitical to the environment that they operate in professionally for basically free. Furthermore there are no real consequences for failure in either a practice environment or professional outside of whatever stakes they agree to. A commander may be able to simulate a conflict or a CEO may be able to simulate controlling a company, but the simulation will pale in comparsion to the real thing in which the system the controller is dealing with is infinitely more complex thanks to it theoretically including the entire universe. It is this that gives the professional gamer the edge over networked novices, despite the advantages cybernetics says they should have. The closed system of Starcraft means that through countless hours of practice they can acquire muscle memory and heuristics that allows them to quickly respond to any given situation. If you were to create a more complex RTS that was built from the ground up to require multiple players controllign a single force, the professional advantage would be reduced in favor of distributed command.

In this sense Starcraft and similar RTS games are a midpoint between fully networked organization and traditional hierarchical management. A progressive games that reflect the future of organizing would delegate management responsibilities between several players controlling a single force in real time. Unfortunately real time strategies have become increasingly unpopular as of late, largely due to the popularity of MOBA games like DotA 2 or League of Legends in which two teams of players engage in an real time strategy type conflict, except they only have to control a single powerful hero unit and as such do not have to bother with complex army and base management.

While there may be a market for competitive real time strategy games that involve multiple players managing a single army, we've yet to see any achieve commerical success. However I hope that such a game gets made, even if it just happens to be a tech demo or a proof of concept, simply because it would be an evolution in the real time strategy genre that reflects the evolution that we as a society are going through with regards to how we organize.

  1. Friedrich Hayek, Use of Knowledge in Society ↩︎

  2. Lim Yohwan, translator Marencielo, Crazy as Me. Retrieved from ↩︎

  3. Thomas Hammes, The Sling and the Stone: On War in the 21st Century, p. 7 ↩︎

  4. ibid, pg. 192-193 ↩︎