Saturday, May 10, 2008


For this article, the majority of the specific examples will come from Warcraft III: The Frozen Throne.

In my mind, decision-making encompasses two fundamental points to gameplay: Macromanagement and general army movements. As a brief definition, macromanagement (henceforth, "macro") is the player's ability to spend his resources. The difficulties here are "on what" and "when." Do we make Grunts, Headhunters, Kodo Beasts, an expansion, tech to Fortress, stack gold at 50 food? None of these are the right answer all the time. As economies get larger this question expands itself. We can make a lot of things all at once, and we're probably going to want more than just one type of unit. The second point of difficulty, choice, skill differentiation, whatever you want to call it, is the usage of a player's army. Now, don't confuse this with micromanagement, there's no decision made in whether to micro well or not. However, where and when to attack, what to creep, where to position oneself on the map, these are all decision to be made. Much like the first point with macro, we are making these decisions with limited knowledge.

Unless the game is created in such a way (or through the usage of cheats), we don't always know what our opponent is up to. We usually don't know his exact army composition, we aren't always 100% sure if he's trying to expand, tech, upgrade, rush, etc. What's the difference between a Human player tower rushing you and a Human player expanding? His army's headed to you and not to a gold mine. That's it. No matter how much you scout, your opponent has those two options open with the same opening. Every time. Now, defeating each of those strategies requires different tactics. Clearly you cannot truly build to counter just one of those choices if your opponent can simply change his strategy on a whim, while for the Human player, attempting to fast expand and attempting to tech quickly are mutually exclusive builds in many situations.

So we've established that we make decisions with limited knowledge, but let's try to break down what this means for us as players or designers. First, however, I am going to break down the four main categories that macro decisions fall into:
  1. Unit Production: Pretty straight-forward. Obviously what to produce is still in question, but producing units gives the player the ability to attack, defend, or creep (in some games). These can be to support other units, counter a specific set of units an opponent has, or just be a generally useful unit. A point to players: Think about why you're making this specific unit. What is it countering? What is it adding to your army? Will your opponent easily counter it? Has he already countered it as part of his army? A point to game designers: Make sure players have more than one option for defeating a certain unit. Otherwise undefeatable two-unit combinations may exist: "Unit A" and the counter to whatever counters "Unit A."
  2. Tech: Teching has one purpose and one purpose only: It unlocks other units. Want to stay at a Great Hall all game? Well then you're stuck with Grunts and Headhunters. Want Gryphons? Get a Castle. Sometimes this may overlap with other categories (i.e. Blacksmith allows Riflemen but also allows upgrades).
  3. Expanding: Another fairly straightforward point, expanding allows a player more resources to spend on any of the other three categories. This is one of the riskier options as generally expansions cost a great deal of resources. This is probably the hardest point for designers to perfect: a) Expansions may be the best option always, in which case there's no decision-making behind expanding, it's simply a fact of the game. b) Expansions are almost impossible to put up, though in this case, if the rewards are great, players may build strategies around securing expansions, which furthers the longevity of a game.
  4. Upgrades: Researching an upgrade improves current and future units. There are many ways in which this takes shape: Siege Mode to allow (surprise!) Siege Mode for StarCraft's Siege Tank, Orcish Unit Armor to improve the longevity of nearly every unit in the Orc army, Animal War Training to make Knights, Dragonhawks, and Gryphons more durable. Some may be "one-shot" upgrades, such as Siege Mode or War Training, some may have various levels as with the Armor upgrade. Ultimately, upgrades are an investment in your troops. Barring a few cases, your army will generally perform better with one more Bear than it will with one more level of Strength of the Wild. But since upgrades will affect every unit built, they can be a strong choice in the long run. Players must weigh the improved efficiency for a certain group of units against their need for more immediate improvements in army strength. A player must also be wary of those units becoming obsolete, so as not to waste resources.
Mechanics (such as Warcraft's upkeep system) may affect how strong these options are: Few players transition flawlessly from 50 food (no upkeep) to the 60-70 food range (low upkeep: 70% income) without first waiting for a sizable bank of gold. Undead in particular is known for staying sub-50 until their gold mines near depletion, suddenly surging towards three-digit food counts. Because of "upkeep abuse," amassing upgrades and possibly expanding while waiting at 50 food proves to be a valid strategy. However, unit production is more effective now, so players who stack gold for only a short time can overrun overzealous bankers. In matchups like Night Elf mirror, one of the key components is to see which player breaks 50 food and pushes first and whether his opponent can efficiently repel the attack.

As players, we have to make decisions on what to make and when to make it all the time. 90% of the time, these choices are sort of pre-determined. With the advent of replays and the distribution of common strategies, most players know "make a bunch of Gargoyles against Elf" and "make mass Druids of the Talon against Orc." The answer to "when" is usually "all the time." If you have 135 gold, make a Talon. Sometimes this is interrupted as players purchase Scrolls of Healing, but the macro plan for the game is pretty straightforward. However, even a matchup as "simple" as mass Talons has its complexities. How many Archers does one make at tier 1? How many Wisps? How many Wisps do we replace after building our Ancients of Wind? What heroes to we make? When do we start tier 3? When do we get Adept training? When do we pause for Heal Scrolls? If you're a Night Elf player and you answer is "I dunno" you've got some work to do.

Even without set strategies, or in games where there are set strategies but many possible openings, we have to scout - often. Certainly, effective play requires a "fluency" with the game. You have to know what can come out of Beastiary before you can hope to counter it. Truly top-level players can see small differences: less Peasants than should be expected, another Farm, maybe the lack of an Arcane Tower, and smell "Fast expansion" and make a beeline across the map for a creepjack. All of this comes with experience. Also, as you may have noticed from this paragraph, macro and army movements are intertwined. The counter to seeing less Peasants is "let's go creepjack." Your opponent made the macro decision to expand, you made the army movement decision to find him. Your macro will likely be different as well, maybe fast air to deal with a land-locked opponent, maybe fast tier 3 to secure expansions with (teching for the Tiny Great Hall).

So what's the point of this article then? Two things, I suppose: A word to developers and a word to players; we'll begin with the latter.

Players: If you're trying to learn a game, or a matchup, or a strategy, or a counter-strategy, there's more to look for than you may think. Look through the player's eyes. What is he making? How does this change among other games of this specific matchup? What caused these changes: What did or didn't he see? What about army movements, when does he creep, attack, heal, creepjack, etc? Every change is effected (verb form, look it up (thanks xkcd)) by either a movement or macro choice by his opponent. Every pro player is simply a dictionary of stimulus-counter combined with good micromanagement abilities.

Devs: Obviously we don't need creeps to keep the games interesting (StarCraft, CnC3, etc all work well), but in order to keep games interesting, it's nice to have a sprawling game of rock-paper-scissors to play. It doesn't have to be just with unit counters, but expansions, attack timing (killing tier 2 buildings before they complete is a big part of Warcraft 3. It's non-existant in CnC3)... I guess one thing I noticed, if we are to compare WC3 to SC, StarCraft's army movements are all about the economy. Attacks and defenses are about stopping, delaying, and damaging expansions/economies. Warcraft's are about not only that (albeit much less frequently), but hero levelling, item gains, and the prevention of those acquisitions. I suppose that makes the army movements in Warcraft more complex with more factors that weigh in. Ultimately, when you make a game, it absolutely must have a sort of give-and-take, both with units and their counters, but with overall decision-making, how and where use the army you've built.

As with the previous article, I've recorded a podcast for your further enrichment. All podcasts can be found here.


Anonymous said...

Sun Tzu makes the distinction between tactics and strategy. Strategy is exactly like the "strats" that are available for wc3 builds - the cookie cutter stuff that non-pros follow to win games.

Tactics on the other hand would be things like positioning and unit counters. They are more individual concepts than a set recipe. Such as, take the higher ground, trap the opponent, harass when you're weaker than him.

++respect for xkcd reference ;)

White-on-black is friggen difficult to read.

Interesting read, nice ideas. Read Sun Tzu's Art of War if you haven't already. Your conclusion though was somewhat thin... "when you make a game, it absolutely must have a sort of give-and-take" It would be very interesting if you expanded on this point - how to achieve that in the specific way that makes it fun. Some games do have give-and-take but they fail to provide entertainment - the point: what sort of give-and-take interactions do players enjoy?

I think you have an opportunity to discover really original concepts with your professional background in game playing, so keep on thinking on it - I look forward to your next articles.

Karthick Gopal said...

excellent writeup, I would request more writeups on the player aspect and learning and improving on it. That would be really awesome to read. But just a request please tag your posts as warcraft/starcraft/ etc so it makes it easier to search topic wise.

Else excellent work Sir!