Jekyll2018-12-28T01:16:07+00:00https://evizaer.github.io/feed.xmlNo Hidden InfoA strategy game design and programming blog.Carrot or Stick? Fixing an XCOM Design Problem2018-12-27T00:00:00+00:002018-12-27T00:00:00+00:00https://evizaer.github.io/Carrot-or-Stick<p>The XCOM series reboot started with XCOM: Enemy Unknown, released in 2012. The game had a tightly-designed combat system with a big problem: all designed incentives led the player towards maximally conservative play. Instead of engaging deeply with the combat system, players found it more prudent to slowly and methodically creep forward through missions to expose themselves to as little risk as possible. Playing in this way is far less interesting than what the system could possibly support. Conservative play leads to few dangerous and complex situations–conservative play avoids the most interesting situations the game has to offer!</p>
<p>The developers recognized this issue and tried two different approaches to fixing the problem in follow-up releases.</p>
<p>The first attempt was in an expansion for XCOM 2012, Enemy WIthin. Enemy Within added “meld canisters” to maps which would expire after a certain number of turns. This design tries to encourage the player to move more aggressively and dive into riskier, more interesting combat encounters by placing timed rewards in the scary unknown. Once the player had collected a few of cannisters, they could use the meld to buy powerful augmentations for their soldiers.</p>
<p>In XCOM2, the developers ditched meld and instead added failure timers to most mission types. After the timer expired, the player would suffer a major punishment–usually any soldiers still on the map when the timer expired would be instantly captured or killed by the enemy. By putting a big punishment on a timer, this design forces the player to move at an uncomfortably high speed and accept risks in order to forestall a large punishment.</p>
<h2 id="carrots-are-not-enough">Carrots are Not Enough</h2>
<p>On its face, Enemy Within’s meld canisters should encourage players to engage more deeply with combat. It’s a reward for playing faster and more aggressively–the first such incentive the developers added to the design. In isolation adding such a reward seems good, but the results of this design intervention were mixed because of XCOM’s character progression system and the relative scale of rewards and punishments.</p>
<p><img src="/images/Meld.png" alt="Screenshot of a meld cannister in XCOM2012" />
Losing a powerful character is a terrible blow to the player. Making a character incrementally more powerful is a boon, but on a comparatively much smaller scale. The player will see their many characters achieve increments of progress 20+ times during a campaign. Aside from characters gaining promotions and unlocking new skills, the player advances through a tech tree replete with helpful items they can equip to their characters to make them more effective in combat. A meld canister doesn’t even represent one such increment, it only represents abstract progress towards an improvement for one character. The player may suffer the injury or death of her soldiers for overextending when trying to reach a meld canister, but the quality of the reward for success is often minimal or non-existent. The reward is disproportionate to the potential costs.</p>
<p>The reward also fits uneasily with campaign progression. Skilled players will reach meld canisters more often, which improves their characters faster, which makes it easier to reach meld canisters in future missions. This positive feedback loop further destabilizes the difficulty curve of a game that already suffers from frustrating negative snowball effects and boredom-inducing positive snowball effects. If the designers change the difficulty curve of the game to suit a player who can get most of the meld canisters, less-skilled players will be underpowered and frustrated by challenges as they get deeper into the game. If the designers leave the difficulty curve unchanged, players who get meld reliably will stomp through most of the game with the power meld buys. Perhaps a dynamic difficulty curve could solve this problem, but the scope of this design change is likely beyond the money and time limitations of the dev team, aside from being complex and difficult to design well under ideal conditions.</p>
<p>A positive incentive to take risks in combat doesn’t work well with the rest of the XCOM series’ design fixtures. The game penalizes poor play with sharp and large losses to the player’s effectiveness. Incremental rewards can’t balance such potential penalties. Any such rewards put more stress on an already unstable campaign difficulty design while doing nothing substantial to preventing players from playing in a boring but safe way.</p>
<p>As a player gets better, they can move more quickly across the map because they’re more confident in their ability to deal with difficult encounters. The game threatens a less-skilled player with terrible penalties for overextending their squad, and overextension is easier to do by mistake for a less-skilled player because they are less capable of dealing with complex combat encounters and limiting risk while under fire. The better player can play faster because he can better deal with the consequences of playing faster. The worse player stands to lose a lot from playing faster. For a positive incentive to draw a player out of super-conservative play, the reward has to be at least proportional to the penalties of failing due to the risks of pursuing the reward. In XCOM, the penalties are steeper for less-skilled players, but the meld reward is the same as it would be for an expert player. Meld as an incentive away from super-conservative play doesn’t work.</p>
<h2 id="xcom-2s-attempted-fix">XCOM 2’s Attempted Fix</h2>
<p>XCOM 2 tries to fix the crawling-forward incentive structure through two additional mechanisms:</p>
<ul>
<li>Before enemies spot the player’s squad, enemies see less far than player characters. The game calls this special condition “concealment.”</li>
<li>Most missions are on timers with harsh penalties for exceeding the timer.</li>
</ul>
<p>The concealment rules allow the player to lay a free ambush at the beginning of the first combat encounter in each mission. Since the player can see enemies while pods are inactive, she can maneuver her units into good positions from which to fight and activate the pod when it is best for her. This initial advantage makes this first combat encounter easier and thus makes it easier to survive (or get through unscathed) if additional pods are activated in the course of the first combat encounter.</p>
<p>The player can gain much useful information from exploiting the fact that her characters can see significantly further than enemies while in concealment. With this information, not only can the player plan the opening firefight in some detail, she can also scope out the composition and likely location of other pods in relation to the mission objectives. She knows a lot more about the likely course of the mission in XCOM2 because of this additional information than she could ever get in an XCOM2012 mission. From this, the player gains significant agency in how they pursue mission objectives and face combat challenges–such agency is sorely lacking in XCOM 2012.</p>
<p><img src="/images/xcom2-evac-point.jpg" alt="Screenshot of a contested Evac Point in XCOM2" /></p>
<p>Balancing this boon, the designers added timers to most mission types. These timers alone are enough to force players out of their defensive approach for some portion of the mission. Failing to complete the mission before the timer expires results in the player losing characters who haven’t reached the extraction zone (a small area of the map, usually on the other side of the map from the starting place). This punishment is similar in scale to the how the game can punish the player for putting their squad in danger. Because the player faces the loss of their characters if they’re too passive as well as too aggressive, they’re pushed towards middle ground–risking enough to get the mission done within the time limit, but not risking so much to suffer avoidable losses during that push.</p>
<p>Since the player starts in concealment and can see pods before she activates them, creeping forward slowly is less appealing. Concealment allows the player to scout enemy positions at little risk when the mission is at its most threatening, so the core reason for creeping forward withers into a minor concern instead of an overriding fear. XCOM2012’s disincentive for seeing enemies no longer applies. After the player has completed the initial ambush, she is in a similar situation to XCOM2012 but she can only creep forward for a few turns or she risks losing characters due to the timer expiring. The player can use information gleaned while in concealment about the likely location and composition of enemy pods to make aggressive moves that would be unjustifiably hazardous in XCOM2012.</p>
<p>The timer is complemented by mission objective design. The last mission objective in most XCOM2 missions is for the player to move her squad to a specific small area on the map. The player doesn’t have to kill all the enemies on the map. The enemies can behave aggressively to prevent the player from getting to her positional objective, and the player has the choose to engage or bypass whatever resistance stands in her way. This dynamic is far more complex than the typical systematically-hunt-and-destroy-at-minimal-risk dynamic of an XCOM2012 mission.</p>
<p>XCOM2’s timers fix the creeping forward problem by disincentivizing super-conservative play as much or more than the game disincentivizes overextension. The timer would create too harsh and arbitrary a punishment if added with no further modifications to the formula of XCOM2012, so the designers also added concealment. Concealment allows the player a temporarily safer environment in which to scout and an advantage in the first combat encounter of the mission. With more information at their disposal, the player can make plans to cope with the potential sharp penalty the timer may impose in the not-so-distant future.</p>
<p>XCOM2 does a much better job at putting the player into tough but survivable combat situations. Since this is the core of its tactical combat, I view it as a design success and an improvement over XCOM2012.</p>The XCOM series reboot started with XCOM: Enemy Unknown, released in 2012. The game had a tightly-designed combat system with a big problem: all designed incentives led the player towards maximally conservative play. Instead of engaging deeply with the combat system, players found it more prudent to slowly and methodically creep forward through missions to expose themselves to as little risk as possible. Playing in this way is far less interesting than what the system could possibly support. Conservative play leads to few dangerous and complex situations–conservative play avoids the most interesting situations the game has to offer!Incentives and Intent: XCOM’s Creeping Forward Problem2018-12-01T00:00:00+00:002018-12-01T00:00:00+00:00https://evizaer.github.io/Incentives-and-Intent<p>You can better design games when you can keep two clear and separate pictures in your head at once: what your game incentivizes the player to do, and what you want the player to do. Often designers only have a grasp on what they want the player to do so their playtesting consists of doing those things and making sure they work. Regardless of the clarity of the designer’s vision, the player does not have access to it and will follow the game’s incentive structure. Even relatively small misalignments between intent and design can cause the end result to look markedly different if not broken. For instance, an RTS with a slight imbalances in unit design will see players neglect most of the unit types available to them in favor of building the one imbalanced unit or its counter.</p>
<p>Designers benefit from achieving a critical distance from their work and trying to re-learn their games so they might play in a way independent of their own preconceptions of how they intended the game to be played. This is no easy task, but practice analyzing the incentive structure of games where you don’t already know designer intent can equip you with the tools to achieve critical distance and analyze how incentives influence player behavior in your own game.</p>
<p>The basic process is this</p>
<ol>
<li>Pick out a class of decision that feels off to you.</li>
<li>Write down the design elements that factor into that decision.</li>
<li>Check your list by putting yourself back into the same situation and re-making the decision based only on the factors you wrote down. See if there could be other viable options you could’ve chosen based on those factors.</li>
<li>Add factors until your choice is the only one you can imagine taking given careful consideration of the factors you listed. (Also toy with removing factors if no instantiation of a factor can make you change your decision.)</li>
<li>Trace factors back to design elements and conduct thought experiments like “what would I do if X worked like Y instead of Z?” Start from the simplest changes that might fix the decision which feels off.</li>
</ol>
<p>In this article I’m going to analyze a part of the incentive structure of XCOM 2012’s tactical combat. The goal is to learn why the optimal strategy when not engaged in firefights with aliens is to slowly creep your squad around the map, revealing a minimal number of tiles each turn.</p>
<p>XCOM 2012’s combat system is otherwise dynamic and tightly-designed. It should encourage its players to fully embrace its complexity. The design should encourage the player to show aggression so that the player can get into interesting, dangerous situations more often and thus experience the full intellectual and emotional gamut the combat system can deliver.</p>
<p>In this article, I will analyze the incentive structure of tactical missions in XCOM 2012 to find the causes of boring and degenerate map exploration tactics, then talk about how XCOM 2 tried to improve on its predecessor’s design in this area.</p>
<h2 id="the-rules">The Rules</h2>
<p>In XCOM, the player and enemies alternate turns in which all of their characters can act. the player can issue orders to her characters in any order, but each command is permanent. The player’s turn ends when all her characters have taken actions or double-moved. Each character can move and act, or double-move. Actions include firing at enemies, reloading a weapon, using special abilities, and going on overwatch.</p>
<p><img src="/images/xcom2012-1.PNG" alt="Screenshot of XCOM2012" />
(Image taken from <a href="https://www.youtube.com/watch?v=T-cBJc38O_k&list=PLXctaw5JGF4LcidFVdkQMV1tc2DfC8x3D">Beaglerush’s fantastic let’s play</a>.)</p>
<p>The player can only see the tiles within line of sight (LoS) of her squad members at the moment. The majority of the map is shrouded in darkness. Hidden throughout the map are several groups of enemies (called “pods”) that move each turn. Hidden enemies do not move in predictable ways. Some missions also have a small number (or just one) pod that sits in a fixed place on the map, or patrols a very small area like the main room of a UFO.</p>
<p>If the player doesn’t encounter enemies for a few turns the game gives the player a clear visual cue which tells her the general direction of a nearby pod. Aside from this, the player can only get hints to the position of hidden pods when flaws in the fog of war allow the player to see faint traces of aliens opening doors or breaking windows that are out of the squad’s line of sight.</p>
<p>The challenge in any mission comes from defeating the enemies on the map. At any given time enemies are either in an active state, where they’re fighting the player’s squad, or in an unactivated state, where they “wander” the parts of the map that the player cannot see.
A player activates an entire pod when any member of that pod gains LoS on any member of the player’s squad. Since the player cannot activate less than one pod at a time, the pod represents a basic unit of danger in the mission.</p>
<p>When pods activate, the members of the pod get one free move each, which they typically use to get into cover. If the pod activated on the player’s turn, the player’s turn continues as normal after the enemies “scamper.” If aliens walk into the player’s LoS during the aliens’ turn, the aliens scamper and the aliens’ turn ends.</p>
<p>The objective of a mission usually involves killing all enemies or reaching some location. Either way, the player must move across the map into unknown territory to fulfill the objective. Each tile the player reveals may contain an enemy and thus activate a pod, even when the player is currently engaged with another pod.</p>
<p>Soldier and alien LoS is reciprocal. Once the player sees enemies, those enemies will activate. A notable exception is the “battle scanner” item, which a soldier can throw into the distance and gain temporary vision of enemies such that they won’t activate. Most of the time the player will activate enemies by moving one of her characters such that it gains vision of an enemy.</p>
<p>Characters can be put on overwatch instead of shooting. On overwatch, a character will fire at the first enemy that moves within their LoS. When a character is on overwatch and a pod activates, that character will get their overwatch shot on any enemy that scampers through their LoS. In the next turn–which comes before any aliens can shoot–that character can take another shot. Overwatch effectively grants a soldier an extra shot on aliens that just activated.</p>
<p>Dealing with one pod is manageable but can be tricky, dealing with two can be lethal, dealing with three is likely to cost the player multiple characters or wipe out her squad entirely.</p>
<h2 id="the-disincentives">The (Dis)incentives</h2>
<p>The costs of attempting the unnecessarily dangerous in XCOM are high. Characters that die in missions are gone forever, and injured characters may be unavailable for several missions.</p>
<p>The design of the strategic layer of XCOM steadily increases the challenge level of tactical missions with minimal regard for the player’s success or failure. The game offers a hard-pressed player no breaks and often drives her into a campaign-ending death spiral through the cumulative loss of resources over several missions. Death-spirals take this shape: the player’s most powerful characters get injured, meaning inferior characters go on missions and have a higher likelihood of also suffering injury or death. Since the top character is injured, she is not gaining experience and advancing. Weaker characters go on missions that are above their means, so it’s more difficult for them to survive so they can grow in power. On the whole, the player with less skill falls progressively further and further behind the mounting difficulty, defeat feeding on defeat.</p>
<p>The game makes it abundantly clear to the player thematically and mechanically that she shouldn’t want to take injuries or have characters die for any reason if it can be avoided. The simplest way to avoid such pitfalls is for the player to engage as few enemies as possible and engage them one at a time.</p>
<p>The mechanics shape combat into a trial of risk minimization techniques. The player will rely on deterministic ways of dealing damage, like grenades and rockets, and seek the safest possible engagements by crawling forward and fighting as few aliens at a time as the game will let her.</p>
<p>Incentives combine to constrain the player into playing in a way that reveals as few tiles as possible on their turn and ensures that multiple squad members’ turns are available to deal with pods that may activate. The resulting play is a deliberate crawl across the map. The player’s first move is to advance her forward-most character, then follow with the rest of the squad such that no move but the first reveals tiles. Most of the player’s commands are revealing nothing, and most of the player’s time is spent waiting between turns or watching character-running animations.</p>
<p>What would the player gain by completing missions quickly? The game’s systems provide no incentive to do this, but a player may find it boring to play as conservatively as the game allows. A knowledgeable player trying to play her best is forced into the awkward situation of having to choose between spending a lot of time playing in a boring but correct way, or spending less time playing in a more fun way but one that lowers their chances of success for no in-game benefit.</p>
<p>This fun-for-effectiveness trade-off hurts the design. The design punishes the player for engaging the most stimulating and interesting kinds of dangerous combat situations. The design gives the player no reason to seek challenge, so the smart player studiously avoids it.</p>
<p>In the next article I will look at the way XCOM2 and the Enemy Within expansion for XCOM2012 try to address this problem.</p>You can better design games when you can keep two clear and separate pictures in your head at once: what your game incentivizes the player to do, and what you want the player to do. Often designers only have a grasp on what they want the player to do so their playtesting consists of doing those things and making sure they work. Regardless of the clarity of the designer’s vision, the player does not have access to it and will follow the game’s incentive structure. Even relatively small misalignments between intent and design can cause the end result to look markedly different if not broken. For instance, an RTS with a slight imbalances in unit design will see players neglect most of the unit types available to them in favor of building the one imbalanced unit or its counter.Tension and Interesting Decisions2017-05-21T00:00:00+00:002017-05-21T00:00:00+00:00https://evizaer.github.io/Tension<p>Sid Meier’s famous quote that “games are a series of interesting decisions” rests at the core of how many designers understand their craft. But we seldom examine what makes decisions interesting. Intuition can take us pretty far, since we all have plenty of experience with tough choices we’ve had to make throughout our lives, in games and out. But the structure of these choices, and how to design such choices in a controlled environment like a game, is seldom discussed in much detail. Three freely available pieces of media that have tried to tackle this problem in games design are <a href="http://www.gamasutra.com/view/feature/174832/the_more_you_know_making_.php?print=1">Jon Schafer on what makes decisions interesting.</a>,
<a href="https://learn.canvas.net/courses/3/pages/level-6-dot-1-tips-on-decision-making">Ian Schreiber’s tips on decision-making</a>, and
<a href="http://www.gdcvault.com/play/1015756/Interesting">Sid Meier’s GDC Talk on Interesting Decisions</a> (<a href="http://www.gamasutra.com/view/news/164869/GDC_2012_Sid_Meier_on_how_to_see_games_as_sets_of_interesting_decisions.php">Leigh Alexander’s summary</a>). I find all of these to be vague and unsatisfying. This article is my attempt to build up a theoretical groundwork for interesting decisions in strategy game design.</p>
<h2 id="narrowing-the-scope">Narrowing the Scope</h2>
<p>For the purposes of this article, I will constrain “interesting decisions” to be goal-directed decisions made from a systemic perspective. I’ll pass over some other putative types of interesting decisions, like cosmetic character customization, in the interest of having a clearly circumscribed area for discussion.</p>
<p>A critical part of interesting decisions I will not go into in this article is that of relevance to the player in a broader sense than systemic concerns allow. Few players are interested in making decisions about things if the only reason they have to make decisions about them is that a game is telling them so. We’re interested in that which is personally relevant to us, and for many players that relevance is predicated on thematic choices that may have nothing to do with the system. Winnie the Pooh strategy games will turn off many potential players based on nothing more than that the player sees a child-oriented theme and assumes the game isn’t relevant to them. This article will not discuss how to put decisions into the right cultural context to make them accessible and meaningful to players more broadly.</p>
<h2 id="trade-offs-the-starting-point">Trade-offs: the Starting Point</h2>
<p>Meier, Schreiber, Schafer agree that trade-offs make interesting decisions possible. I’m left with a lot of questions, though. What makes a trade-off? How can we look at a system without playing it and see where trade-offs occur? Are some trade-offs better than others for creating interesting decisions?</p>
<p>Trade-offs are an emergent property of decisions which arise from conflicts in incentives. When the player can’t do everything he wants to do in a given span of time, a trade-off pops into existence: progress towards one desired outcome comes at the expense to some amount of progress towards another.</p>
<p>To better articulate the causes of trade-offs, I say that some game rules are <em>in tension</em> with one another. They vie for the player’s attention as he tries to decide what to do next and in so doing provide the raw materials for strategy: if he hopes to win, the player has to piece together some coherent plan to take advantage of the natural synergies between rules and diminish or eliminate forces that get between him and the game’s overarching goal.</p>
<p>To understand this in more detail, we must explore goals and sub-goals.</p>
<h2 id="first-goal-structure">First: Goal Structure</h2>
<p>All incentives flow from the structure of sub-goals formed by the interaction between the game’s goal, play environment, and available player actions. Any discussion of the tensions between game rules must start with a sub-goal analysis. To do this analysis, start with the game’s goal and work backward to the game’s starting condition, one step at a time, making reference to player actions and game concepts where possible.</p>
<p>Here’s a goal analysis for Company of Heroes’ adversarial multiplayer:</p>
<ol>
<li>The game’s goal is to reach 500 Victory Points before your opponent does.</li>
<li>To reach 500 VPs, you must control more Victory Locations at a tick than your opponent.</li>
<li>To control VLs, you must capture it and prevent your opponent from capturing it.</li>
<li>To capture a VL, you must have at least one unit close enough to it for a number of seconds.</li>
<li>To have a unit close to a VL, you have to move units to it and fight enemies who might stand in your way.</li>
<li>To move units you must have units to move.</li>
<li>You acquire units by building them at certain buildings.</li>
<li>You have to build buildings which permit you to build units.</li>
<li>You build buildings using your starting resources and engineers.</li>
</ol>
<p>Resource acquisition and the game’s economy is actually a parallel goal structure built into those last three steps.</p>
<ol>
<li>To build units, you must have sufficient resources, buildings, and technologies.</li>
<li>To build buildings and research technologies, you must spend resources.</li>
<li>To acquire resources you must control capture points of different types.</li>
<li>You control capture points the same way as you control VLs.</li>
</ol>
<p>I could go further and do a similar analysis for what it means to control space. This analysis involves talking about the counterplay between units. I will leave it as an exercise to the reader in the interest of brevity.</p>
<p>This analysis provides a high-level overview of what the player is incentivized to do in their pursuit of victory. Tensions arise when there are multiple ways to reach these sub-goals, where multiple concrete sub-goals vie for attention (e.g. should I try to capture the north or south VL?), or where progress towards one sub-goal can be traded for progress towards another. A goal analysis doesn’t show you where interesting decisions happen or why, but it reveals the superstructure which gives interesting decisions room to happen.</p>
<p>Now that we’ve got an analytical tool to help us find the sub-goals, we need to be able to combine that with an analysis of player actions to see how those actions create tensions in pursuit of the sub-goals.</p>
<h2 id="player-actions">Player actions</h2>
<p>Tension occurs as a product of multiple appealing actions the player can take. These are the verbs players use to interact with the game. They are always parameterized further. E.g. to move a unit, a unit and a destination must be specified; to attack, at least an attacker and defender must be specified.</p>
<p>Player actions have a few properties which exhaustively define them:</p>
<ul>
<li><strong>Payouts</strong> – some combination of resource or condition changes that happen once the player takes the action. (Ex. The payout for moving your piece into an enemy piece in chess is that the enemy piece is removed from the board. The payout for building a Barracks in Star Craft is that you can now build Marines.)</li>
<li><strong>Costs</strong> – some combination of resources or constraining environmental changes that must happen in order for the action to happen. If you can’t pay the costs, you can’t take the action. (Ex. The cost of a Marine in Star Craft is 100 minerals, 1 supply, and a certain build time at a specific barracks.)</li>
<li><strong>Conditions</strong> – what the details of the environment must be for the player to be able to choose the action. Conditions are like costs except they aren’t “paid”. The objects of conditions are not changed by taking the action. (Ex. You can’t build a marine unless you already have a Barracks built.)</li>
</ul>
<p>Payouts, costs, and conditions do not need to all be expressed as conjunctions of constants, like “pay 2 stone to build a wall segment”. These properties can also be expressed as functions of one another, as in “If x enemies died last turn, deal x damage to an enemy where x is the number of red mana you pay.”</p>
<h2 id="components-of-tension">Components of Tension</h2>
<p>Rules (player actions and environmental rules alike) come into tension with one another as a result of three properties interacting to create clashing incentives: exclusivity, situationality, and indirectness.</p>
<ul>
<li><strong>Indirectness</strong> – it must involve choosing between actions which indirectly move the player towards the game goal, thus obscuring their value. Without indirectness, the player can easily value each choice and choose that which has the highest value.</li>
<li><strong>Exclusivity</strong> – each choice needs to consume resources that, if lacking, adversely effects the player. Without exclusivity, there is only one option, or the player can simply always choose all options.</li>
<li><strong>Situationality</strong> – circumstances should weigh heavily on which choice is preferable. Without situationality, you repeatedly choose the same move once you’ve decided it’s best.</li>
</ul>
<p>Actions generate tension through having asymmetrical costs and payouts to others. Asymmetry of conditions leads to strong situationality.</p>
<h2 id="tension-in-action">Tension in Action</h2>
<p>Now let’s bring it all together by analyzing the structure of RTSes like Company of Heroes.</p>
<p>The actions of building units in RTSes typically exhibit all three necessary tension-generating properties.</p>
<ul>
<li>Units cost varying amounts of shared resources, creating exclusivity.</li>
<li>Which unit to build (and if you should build any right now) depends on what you’ve seen of your opponent, creating situationality.</li>
<li>No victory condition is satisfied by producing a certain unit–you typically have to build and move many units to achieve the victory condition–thus there are usually at least two levels of indirection. The above goal analysis of Company of Heroes shows that there are five levels of indirection in that case.</li>
</ul>
<p>Another core problem for the player in Company of Heroes, as with all RTSes, is force allocation. This problem arises because of exclusive positioning and localized effects of units. Guns can only fire so far, and maps are far bigger than any gun can fire. Even off-map artillery in Company of Heroes can only effect a small area of the map at once, though it can be deployed anywhere. On-map units can only effect the area around them, and can be rendered irrelevant to an important engagement through mistakes of positioning. Exclusive, situational positioning is the bread and butter of Company of Heroes combat.</p>
<p>Not only are the units you can produce and allocate to parts of the map situational in a broad sense, but they can be used to better or worse effect tactically. This second layer of situationality arises from rules like cover, armor thickness variation on tanks’ sides and rear, and machine gun teams’ cones of fire. All of these modify payouts conditional on positioning relative to other units and/or map features.</p>
<p>The beauty of Company of Heroes, and indeed most RTSes, is in the interaction between the force allocation and unit construction problems. There are three layers of situationality at play: strategic production, strategic deployment, and tactical combat. The payout of building a unit is thus situational in a dynamic and complex way. This layering of situationality is a powerful design technique, since it tightly interlinks several interesting kinds of decisions that have partially overlapping sources of tension.</p>
<p>Although force allocation seems very different from force composition/unit construction problems, I hope you can see how the common conceptual vocabulary I’ve introduced in this article makes sense of them with respect to the designer’s intentions of making a system of interesting decisions. There’s a lot more to say about this topic, and many higher-level features and interactions to explore using this groundwork. Stay tuned!</p>Sid Meier’s famous quote that “games are a series of interesting decisions” rests at the core of how many designers understand their craft. But we seldom examine what makes decisions interesting. Intuition can take us pretty far, since we all have plenty of experience with tough choices we’ve had to make throughout our lives, in games and out. But the structure of these choices, and how to design such choices in a controlled environment like a game, is seldom discussed in much detail. Three freely available pieces of media that have tried to tackle this problem in games design are Jon Schafer on what makes decisions interesting., Ian Schreiber’s tips on decision-making, and Sid Meier’s GDC Talk on Interesting Decisions (Leigh Alexander’s summary). I find all of these to be vague and unsatisfying. This article is my attempt to build up a theoretical groundwork for interesting decisions in strategy game design.Representing State2017-03-20T00:00:00+00:002017-03-20T00:00:00+00:00https://evizaer.github.io/Representing-State<p>When you play a game, you manipulate game objects using various actions, all of which are defined in the game’s rules. In this post, we’ll look at game objects and how to represent their state.</p>
<p>To make this descriptive effort easier, I’m going use, as a reference point, a generic SRPG or character-based turn-based strategy game.</p>
<p>A character in our notional SRPG is a game object: Just some thing we hang state off of that the game’s rules bring to life. I call the smallest piece of state on a game object a dimension. Dimensions can be a position in space, a maximum amount of HP, current amounts of HP or some other resource, a list of status effects currently afflicting (or aiding) this object, etc. Beyond those more obvious cases, a character also has a series of dimensions relating to what they can do–like their abilities, magic they know, and reagents they are carrying to be used in that magic.</p>
<p>Some dimensions are straightforward in how you represent them. HP is clearly best represented as a number. But you can represent other kinds of dimensions in different ways. You could look at status effects as each having their own dimension on the character they effect which can either have a value of “yes” or “no”. Perhaps a poison effect has some additional dimensions of its own, like how many turns it will last and how much damage it does per turn.</p>
<p>As the poison example illustrates, dimensions can also have dimensions of their own. A character’s position in space is articulated in two or three dimensions, each of which can vary independently.</p>
<p>Let’s dwell a while longer on space. Space is very interesting because it’s a pre-established (outside of the game design) relationship between two or more dimensions. Usually you don’t have this kind of thing going on with game objects’ dimensions. HP doesn’t have any clear connection with another dimension on a character. It’s just a singular number that is typically separately modified by game rules. Space is way more interesting than that, because you have two or three dimensions which have a built-in concept of adjacency. You can define adjacency in a number of ways, but you know from the very concept of space that you’re going to have consistent adjacency across the whole space. You can even choose to define adjacency as totally fluid; you can see this in Paradox Interactive’s grand strategy titles, which feature a map of provinces which have odd shapes and all different numbers of adjacencies to other provinces.</p>
<p>Adjacency is a simple way of saying “given one position, what other positions are one step away?” But you also have this built into singular numerical dimensions, like HP. 6 HP is one step away from 5 HP. Arithmetic operations provide us with a kind of adjacency over numbers. We use this all time in damage and healing amounts for different character abilities, just as we have ranges on attack and movement.</p>
<p>But what about status effects? They seem different. Since they’re just a bunch of separate labels, the only concept of adjacency they may possess is through turning them on and off. So you get a lot of variety out of space, a lot less out of HP, and simpler still are status effect which seem to have almost nothing to offer us.</p>
<p>Though the game’s rules tie all dimensions of game objects together in a variety of ways that end up being what is so spellbinding about games, some kinds of dimensions need more from the rules than others. The meaning of a status effect lives entirely in the realm of the rest of the game’s rules. Space and numerical dimensions have some room for different magnitudes of changes by virtue of existing mathematical relations. status effects have nothing of the sort beyond the most basic “on” and “off” transitions. The game rules, then, define all the meaningful relationships between the status effects. Being frozen in place ties into the character’s ability to act but also makes fire attacks potentially melt the frozen character and free them. Perhaps against a neutral-status character a fire attack begins burning them, instead. This kind of dynamic response isn’t built into status effects as a concept, it’s articulated entirely through game rules that are defined outside of state.</p>
<p>This property means status effects are the most basic kind of dimension a game can have. The simplest kind of state a game object can have merely has those same two values a typical status effect has: on or off. You can’t get any simpler, because with only one possible value, the dimension would be incapable of varying and would cease to be state, it would be some inherent unchanging property codified in the game rules.</p>
<p>Numerical dimensions can be defined in terms of simple dimensions by using several simple dimensions together to represent a binary number. So “on” + “off” + “on” would represent 5. This is not practical, but it shows that all state is an elaboration on the simple dimension combined in some way. Space is just two or three numerical dimensions taken together as one dimension. There’s a natural complexity progression from simple dimension to numerical dimension to spatial dimension.</p>
<p>We already knew that we could represent games in computer code, so what’s the point of this exercise? It’s an introduction to a simple way to represent state in game: game objects, which are logical groupings of some number of dimensions. This terminology and way of breaking down state helps us to represent games concretely and precisely, and gets us one small step towards having an objective and concise representation of game systems that we can reproducibly analyze. With such an analysis we can prove that a game has certain properties, and then we can talk about the dynamics of games systems in a precise way through talking about how these properties interact. From there, we have a stable foundation for having substantial knowledge about what designed game systems do.</p>
<p>In summary:</p>
<ul>
<li>Game state consists of game objects which consist of dimensions.</li>
<li>The simplest dimension is binary: on or off. Like a status effect in an RPG.</li>
<li>Numerical dimensions can be derived from binary dimensions. We do this for the convenience of using mathematical relationships to quickly articulate a wider variety of state changes. The simplest is adjacency–better known as adding or subtracting 1. Through applying adjacency multiple times in various patterns, familiar mathematical operations like multiplication and division happen.</li>
<li>We can group dimensions together, creating space. There are then more complex adjacency rules relating the dimensions that comprise space.</li>
</ul>When you play a game, you manipulate game objects using various actions, all of which are defined in the game’s rules. In this post, we’ll look at game objects and how to represent their state.Near Randomness And Discontinuities2017-03-04T00:00:00+00:002017-03-04T00:00:00+00:00https://evizaer.github.io/near-randomness-and-discontinuities<p>In my article introducing <a href="http://nohidden.info/agency-and-randomness/">near and far randomness</a>, I talked in vague terms of the distance of randomness from the player’s actions, and how randomness forces variation in the course of the game’s events, sometimes at the cost of <a href="http://nohidden.info/agency/">agency</a>. Now, to further elaborate on the thought process of near and far randomness, I will elaborate on what makes certain cases of near randomness onerous, and through that develop ways to maintain the closeness of randomness to player action, but at the same time tame the damage done to agency.</p>
<h1 id="when-randomness-matters-discontinuities">When Randomness Matters: Discontinuities</h1>
<p>Players care much more about random results that cause certain numbers to cross certain boundaries. Even direct and seemingly close randomness can be effectively blunted through an understanding of when changes caused by randomness immediately matter and when they don’t. Randomness denies agency most when its effects immediately generate a big response from the system.</p>
<p>Consider hit points in a typical Dungeons and Dragons-like combat system: The specific number of HP a character has doesn’t matter much to the game’s rules unless it crosses a certain critical threshold: 0. At that point the rules intervene in a profound way by severely curtailing the player’s choices with that character. Yes, this is a weird way of putting “the character dies”, but I want you to be thinking about the peculiar way that a number, usually representing some resource, doesn’t mean much until it crosses a threshold, then it means <em>everything</em>.</p>
<p>I call this a discontinuity. It’s a point at which a small, incremental change in the amount of a resource causes an outsized effect after otherwise not meaning much.</p>
<p>Some discontinuities, like death, are forced by the rules, some merely open up new options to the player. In the same D&D-like game, the amount of gold coins a player has allows them to buy different items. We can see each item they could buy as a discontinuity in the value of gold pieces. If you want to buy a 5gp sword, the difference between 3 and 4 gp is meaningless, though the difference between 4 and 5gp is huge. But it only matters if a few other conditions are fulfilled: player has to be at a place where he can buy a sword, and the player has to volunteer to spend that money. There are, thus, a series of decisions and eventualities between the random event and the player actually buying the sword. This is in contrast to the automatic, rule-enforced action of dying at 0HP, which is universal (it has no conditions attached, except for some very rare cases) and involuntary.</p>
<p>The player’s agency is damaged more when a difference decided by randomness crosses involuntary and unconditional boundaries. When an enemy dies, it matters a lot to the player, since that’s the end of a threat to the player character’s own life. If that threat hangs around one or more additional turns because of dice rolls, and those dice rolls happen at the last possible moment between the monster being alive and being dead, the player has had the causal chain of an event that really matters to them torn out of their hands. This is an acute loss of agency. When the player has to do 5 damage to kill an enemy and their weapon does 4 to 6 damage, nothing stands between the player and his valuable short-term objective except for the roll of the dice.</p>
<h2 id="close-dice-rolls-far-randomness">Close dice rolls, far randomness.</h2>
<p>To demonstrate how we can improve the use of randomness by purposefully avoiding discontinuities, let’s take a look at randomized damage in that generic Dungeons and Dragons-inspired RPG mentioned above. The critical discontinuity in the act of dealing damage occurs when the opponent’s life total is lower than the max damage of the weapon, but higher than its minimum damage.</p>
<p>Since damage is determined by rolling dice, at what time the player manages to reduce the monster’s HP to 0 is highly variable. Rolling a few 2s and 1s on six-sided dice over multiple turns can be the difference between the opponent being long dead and being very alive. That’s a ton of near randomness.</p>
<p>But if we make some alterations to this design, we can make that randomness less near without moving it to an earlier time and thus giving the player time to react to it. The crucial step is to stop thinking of damage numbers in terms of their absolute values, and instead think about them in terms of the variance they cause in the time it takes to kill an enemy with a typical HP total.</p>
<p>Immediately this change of orientation produces some counter-intuitive results. Allowing weapons to do 0 damage seems pretty innocuous if you’re only looking at absolute numbers. But if you look at it from the perspective of time-to-kill, it pushes the maximum value out to infinity! The player can sit there and has a non-zero chance of swinging at an opponent forever without killing it. That’s way worse than if you had all attacks do a minimum of 1 damage–it’s so much better to have combat be potentially of variable length but definitely end at some point.</p>
<p>Aside from making it obvious we have to get rid of 0 damage, we can use this new perspective to see two techniques to push randomness out:</p>
<ul>
<li>
<p>Make sure we only vary time-to-kill by relatively small amounts if it’s a one-on-one fight. (It gets more complicated with multiple friends attacking an enemy and vice versa, but there are ways to handle that, too.) This means to keep damage numbers from varying by large percentages.</p>
</li>
<li>
<p>Never let randomness account for the killing blow. If the player does 4 to 8 damage and the enemy has 6 HP left, have a rule that either says that the enemy will definitely die or that it will definitely live with a predictable amount of HP.</p>
</li>
</ul>
<p>The net impact of these changes is that players will be able to predict when something is likely to die more accurately as it gets closer to dying. The design still allows for variation in when exactly something will die, but it doesn’t submarine the player’s agency by potentially randomly granting an enemy one, or many, more turns of life right before it <em>should</em> have died.</p>
<p>Through using this technique, we’ve moved what otherwise seems like close randomness into a place where it looks more like far randomness, even though we haven’t changed when the dice are rolled much at all! It’s all because we have an understanding of the discontinuities in some critical numbers in the game state.</p>In my article introducing near and far randomness, I talked in vague terms of the distance of randomness from the player’s actions, and how randomness forces variation in the course of the game’s events, sometimes at the cost of agency. Now, to further elaborate on the thought process of near and far randomness, I will elaborate on what makes certain cases of near randomness onerous, and through that develop ways to maintain the closeness of randomness to player action, but at the same time tame the damage done to agency.Agency And Randomness2017-02-27T00:00:00+00:002017-02-27T00:00:00+00:00https://evizaer.github.io/agency-and-randomness<p>Players and designers spend an incredible amount of time and effort grappling with randomness’ role in strategy games. In this article I’ll develop a model for understanding the impact of randomness on agency, which will hopefully clarify the use of randomness as a design tool.</p>
<h2 id="the-context">The Context</h2>
<p>Strategy games are about the player expressing skill in decision-making. Strategy games give the player <a href="http://nohidden.info/agency">agency</a> in trying to achieve some pre-defined goal under varied conditions. What the player chooses to do must have an important role in determining if they achieve their goal.</p>
<p>The player expresses skill in how they react to whatever happens in the game, in terms of how much they can use events to get to their goal. It’s critically important for designers to always be thinking about how the player could’ve done something differently with the knowledge she had when making a certain decision. Trace the causal links between the player’s success or failure, the information they had, and their actions. Is there a clear causal link, or did the result come out of nowhere? Did a monster spawn so close to the player’s avatar that the player could do nothing but put up meager resistance before being slaughtered? Could the XCOM player have known that they will miss three attacks in a row and that would give their enemies time to kill one of the player’s squad members? Was the player locked into a losing path and all the good decisions they made after a certain point were futile, even though they had no real way to know?</p>
<h2 id="randomness-as-a-design-tool">Randomness as a Design Tool</h2>
<p>We can think of a strategy game as a big list of possible things that can happen which is determined by the rules of the game and the participation of the player. The player is interested in finding out what’s going to happen, and their uncertainty about the result combined with their power to influence the result pushes them on to play more.</p>
<p>In a single-player game with no randomness, the player is the sole decider of what happens and the game acts as an impartial judge that figures out the immediate results of the player’s input then waits for more input. In the player’s pursuit of the goal, their job is to filter through all the stuff that can happen so they can find a path that will lead them to achieve their goal. This can get boring, because the player will filter out most of the variety of the game in their effort to achieve the goal, since the game allows way more to happen than the player <em>wants</em> to happen. This hurts variety a lot, making the game become monotonous and boring way too quickly. There may be ways to get around this by designing systems with many complex emergent properties, but it doesn’t seem like this approach has seen any practical use. The standard technique is to use randomness. The designer can deploy randomness as a way to lessen the player’s filtering power. Randomness can force new events into the player’s consideration by virtue of the player simply not being able to control them. This is how designers use randomness to achieve more variety.</p>
<h2 id="near-and-far">Near and Far</h2>
<p>Since we care about just how much and in what ways the player can react to random events, we need to pay close attention to How much game-metered time elapses between when the player becomes aware of the result of a randomly-determined event and when the player is directly affected by it. I use the concepts of “closeness”, or “near” and “far” randomness to capture this idea.</p>
<p>An enemy randomly spawned within range to attack the player before the player can act is an example of very close randomness. An enemy spawned within sight of the player but several turns of movement away from attacking the player is an example of randomness that’s more far than near. The real consideration here isn’t actual physical distance on the game’s map, but a measure of time the event spends in the player’s awareness. Near randomness is often <em>too close</em> in terms of map distance–but that also makes it <em>too soon</em> in terms of time.</p>
<p>Near randomness hurts agency but aids variety by forcing potentially large changes to the game state that the player can’t do anything to mitigate or filter out. But if we push randomness further away from the player’s inputs, we provide variety without compromising agency in any significant way. The randomness then makes the player play differently through providing new stimuli for the player to react to. This expands the number of possible combinations of events the player can encounter while maintaining the player’s agency by avoiding salient, powerful near rendomness.</p>
<p>You can think of randomness like a powerful light. If you put it right next to your eye, it blinds you. You have lit up your surroundings but you’ve rendered it pointless since you now can’t see. Only when you put the light an appropriate distance away does it illuminate your surroundings in a way you can actually use.</p>
<p>If you are familiar with the terms “input randomness” and “output randomness”, you have just seen a derivation of their meaning. Input randomness is far, output randomness is near. But note that this turn’s output randomness can be seen as next turns input randomness–yet in the near vs. far model this is not the case. This turn’s near randomness did its damage to agency, and the only purpose it serves is to force variation in events. Near randomness never “turns into” far randomness as you play a game, or vice versa. Nearness is a property of when the randomness becomes a consideration for the player. In contrast, input randomenss is about if or when the random event becomes a consideration in the player’s strategy.</p>
<h2 id="a-catch">A Catch</h2>
<p>Far randomness is not a cure-all. Since designers do not have solid and implementable abstract models of decision difficulty and player skill, there is no way to verify that the difficulty variation caused by the far randomness is even and fair. To fairly assess the player’s skill, the player must play through multiple sceanrios that test them in different ways and through this multiple-sampling approach of measuring skill the peaks and valleys of difficulty are averaged out. Far randomness works best when there is a relatively short standard unit of play, and the game can assess the player’s skill by aggregating the results of all plays.</p>
<h2 id="conclusion">Conclusion</h2>
<p>We can think of randomness in terms of how close it is to the player. Since players express skill by reacting to events in ways that draw them closer to their goal, giving the player plenty of room to maneuver around random events allows us to add a lot of variety to designs without it costing much agency.</p>Players and designers spend an incredible amount of time and effort grappling with randomness’ role in strategy games. In this article I’ll develop a model for understanding the impact of randomness on agency, which will hopefully clarify the use of randomness as a design tool.Agency2017-02-21T00:00:00+00:002017-02-21T00:00:00+00:00https://evizaer.github.io/agency<p>Agency is the player’s sensation that they participate meaningfully in the game.</p>
<p>Of the three concepts in <a href="http://nohidden.info/Introduction-to-the-AVA-Paradigm/">the AVA paradigm</a> (<a href="http://nohidden.info/analogy/">analogy</a> and variety are the other two), agency is the most psychologically complex and slippery. Agency in games is an extension and subset of agency in real life. Let’s start there.
/</p>
<h2 id="from-world-to-game-world">From World to Game World</h2>
<p>We believe, by default, that we have agency in our own lives. We inherently possess a feeling that we have power over our own bodies and minds in the real world and that the world will respond when we kick or hit or spit or talk. At the very least we’ll feel our own body doing those things–we hope to see others react in some way, and to change what’s going on around us with our actions in perceptible ways.</p>
<p>This immediate sense of agency leads us to believe that we have agency in broader events, as well. Superstitions around luck or divine favor are a part of all cultures. People who hold these beliefs are believing they have agency over distant events which have no causal connection to their action except perhaps in a minor detail. The human mind seems to seek and assumes agency wherever it can.</p>
<p>In video and board game design, the game designer defines a world in which the players participate. Part of this definition includes places and times where players can intervene. Players have agency when they can use their will to make perceptible progress towards their goal. This is the simplest way of conceiving of agency.</p>
<h2 id="a-simple-model">A Simple Model</h2>
<p>From this we can derive a simple measure of agency: What’s the difference in goal achievement rate between a competent human player (or merely someone who knows the rules and can play in a goal-directed fashion) and some stand-in automated player who always chooses his actions at random? A stand-in definitively exhibits no will whatsoever, thus is cannot possibly have agency. So if we compare the result of random play with the result of goal-directed play, we should see the goal-directed play win signfiicantly more often if the player has agency in the game.</p>
<p>In a game where the player doesn’t have many actions to choose from at a time, this model for agency may have some use. Checkers, for instance, allows so few possible moves in most board configurations that the stand-in could at least play a portion of the game in a way that could be mistaken for sensible. Later in the game, when there are more choices to be made, the stand-in will most definitely be dominated by a goal-directed player, though. Unsurprisingly, you do have some agency when you play checkers.</p>
<p>But think, instead, about this kind of analysis applied to XCOM 2. There are so many possible actions the stand-in could take on any given turn, and so many of them don’t make noticeable progress towards the mission’s goal, that the chance of the stand-in succeeding at a mission is nearly 0%. Merely being cognizant of the goal of the game is enough to make your actions significantly better than chance at moving you towards a positive result.</p>
<p>What does this actually tell us? Not much, I think. Players express some non-negligible absolute quantity of agency merely by being conscious of the rules of the game and the game’s goal when they play. The player in XCOM 2 is having a real impact on the result of a mission–even if they’re playing erratically and naively they have a significantly better chance than if the system were played randomly.</p>
<h2 id="salient-moments">Salient Moments</h2>
<p>But we don’t think, when playing a game, of the absolute amount of agency we have. We may make a hundred small decisions over the course of an XCOM 2 mission, but we don’t comment on the fact that we had a lot of agency in the 90 move actions we made–we comment on the 3 75% shots that our characters missed.</p>
<p>Players don’t assess their agency continuously over the course of the game, taking into account all of their actions. they pick out salient points where their agency was denied or subverted. Agency, in the player’s mind, is indellibly tested not in the accumulation of willful action, but in the important moments where the difference between success and failure wasn’t a missed insight, but instead was a missed dice roll. The more clear and direct this dice roll’s influence appears, the more it drains agency.</p>
<h2 id="causal-chains">Causal Chains</h2>
<p>It’s not just important that the player can do things related to the goal, they have to <em>feel</em> like their actions have a direct relationship with achieving the goal. Players feel agency when they identify causal chains in gameplay that go through <em>them</em>; sequences of events where the player can see clearly that their actions have direct effects that relate to the game’s goal. There are many kinds of game rules and combinations of game rules that can get in the way of these causal chains, most often by obscuring them with many additional important links between player action and results.</p>
<h3 id="randomness">Randomness</h3>
<p>Games where the player feels a strong sense of agency eschew salient random events. Designers must be very careful of where they put randomness in a game, because it is, in effect, another actor in the game world that pulls on the causal chains the players cling to, sometimes ripping the chains out of the players’ hands. To minimize the pull, randomness needs to have effects that the player can see coming and react to. When the player can react to randomness, randomness can act to give gameplay some variety without pulling much on the causal chains the player traces their agency through. Miss chances and randomized damage on attacks are, thus, dangerous modes of introducing variety because of how they can damage agency.</p>
<p>In other words, the player feels they have agency when they have the ability to react. Throwing uncertainty at the player before they can react is directly denying their agency by failing to allow the player to interpose his will between important events.</p>
<h2 id="the-uncertainty-paradox">The Uncertainty Paradox</h2>
<p>Players can also feel their agency is constrained by games where long run prediction is possible. Players don’t want the game’s result to feel pre-determined, even if it really was only determined by their actions. When the player feels like the game’s set up against him and the pieces are just mechanistically moving him towards a specific conclusion, agency dissolves. It doesn’t matter that the player had many choices with clear outcomes along the way. A great game keeps the player feeling like they have agency in the outcome for as long as possible, which means not only does the game system need to resist solution, it needs to signal clearly to the player that it is not solved and there’s room for maneuver and eventual success.</p>
<p>Often designers use randomness to give the player the impression that the game is far from mechanistic and that fortunes can change–but in order to have the most positive impact, the player needs to be able to react to the product of randomness before it directly impacts their chance of achieving their goal. Miss chances and swingy damage rolls can add some interest to a game by maintaining a feeling of unsolvedness to what would otherwise be boring lame-duck situations, but this benefit necessarily comes at a steep agency cost.</p>
<h2 id="too-much-action">Too Much Action</h2>
<p>I have primarily been talking about single-player games so far in this article, but multiplayer games have a unique potentially agency-damaging feature: other players. The designer has to keep in mind that the agency of all players is important in a game–agency for all can be hard to maintain when players can set one another back by various means of attack, thus taking a player out of contention for victory early. A complex game with high-contrast, impactful actions for all players to use seems great, but the result of these impactful actions must not change the face of the game too radically, or by the time a player’s next turn comes around they may be looking at a world that bears no resemblance to the one they made decisions about last turn. Though a player’s actions can get lost in the complexity of a single-player game just as well as a multiplayer game, when you’re trying to design an exciting game for multiple players you face a unique challenge in keeping each player’s activity impactful while not, through cumulative actions of all other players, leaving one player’s actions totally buried.</p>
<h2 id="and-the-kitchen-sink">And the Kitchen Sink</h2>
<p>Agency is such a complex topic that even just this brief overview of important concerns feels woefully incomplete. I will come back to agency in numerous future posts to fill in the gaps left here and elaborate on many of the concepts I’ve presented so far.</p>
<p>I’ll leave you with a few more quick points to consider:</p>
<ul>
<li>Enacting the rules in a board game makes people feel like they have more agency in the result than they actually have, because they physically move pieces around a board which makes an impact in a way that stands in for abstractly making an impact through goal-directed action.</li>
<li>A good interface that feels satisfying to use, similar to enacting rules when playing a board game, gives the player a feeling that their actions are having meaty effects, even if the game rules don’t pull through well. Contrast XCOM with and without animations that play when a shot misses.</li>
<li>When you’re low on knowledge about how a game works, your sense of agency is diminished. The player keeps playing because of the promise the game’s analogy shows, or the player is intrigued by what they’ve already learned of the rules–or perhaps it’s just a matter of hype from other players.</li>
<li>If a game is too difficult, players will give up and claim they never had a chance. A frustrated player can easily spiral downward into a feeling of powerlessness, as if they had no agency in what was happening and the game was just taking advantage of them. Usually this leads to tossing aside the game, regardless of how much absolute agency you can show by comparing the player’s actions to those taken at random.</li>
</ul>Agency is the player’s sensation that they participate meaningfully in the game.Analogy And The Process Of Design2017-02-16T00:00:00+00:002017-02-16T00:00:00+00:00https://evizaer.github.io/analogy-and-the-process-of-design<p>In my <a href="evizaer.github.io/analogy">article on analogy</a>, I hinted that designers design abstract mechanics and then apply analogy to make the system most accessible to players. That is not a representation of how most people design video games today, it was merely a set-up for talking about analogy without diving too deeply into the properties of the abstract systems that underlie games.</p>
<p>In reality, strategy game design often starts with analogy and relies on it as a way of making design decisons throughout the development process. Designers pick what kind of a game they want to make through reference to genre convention and personal experience with other games, history, and fantasy. Most design decisions are informed by analogy, not by a rigorous understanding of abstract systems of rules and the dynamics they create. Decisions are made based on what mirrors or captures a part of some interesting interaction in the real world that the designer wants to model and harness. Designers also take–often very direct–inspiration from interesting interactions in other games; existing, working systems provide a stable starting point that designer aims to elaborate or improve on.</p>
<p>We just don’t have significant institutional knowledge in the game design field of how to work directly in the abstract–the design practices I’ve seen and read about always make extensive reference to source material from the real world or other games. These guiding analogies provide a skeleton which is then given flesh and adjusted in its particulars by the designer. The designer takes educated, though often wild, guesses at the flesh of the game based on primarily her intuition, then uses iterated playtesting to figure just what muscles need to be stronger or weaker to make the skeleton stand up. Just as players use a game’s analogies to relate to and mentally synthesize the game’s abstractions, the designer relies on analogy as progenitor and proxy for abstract systems to have any hope at grasping the logical relationships that should exist between game elements. The analogy is driving the process more often than not.</p>
<p>Analogy is not reliable when used as the primary tool for designing games, though. There are numerous analogical justifications for just about any game rule, even quite bizarre ones, and the rules suggested by analogy are often manifestly bad rules. For example, we have to build these analogically-awkward systems of multiple avatar lives in many gamse in order to allow the player to engage in a natural process of learning through repetition and experimentation. Failure means (analogical) death, so the player’s avatar must go through multiple lives. But let’s take analogy as our guide here: if the player is represented as a human avatar and that avatar dies, the game ends just like life does–eternally. That’s ridiculous, of course, but I hope its ridiculousness shows that analogy is no crucible for design decisions.</p>
<p>The designer can use analogy in the mode of post-hoc rationalization for any design decision, since people will pick up even tenuous relationships or make up their own where some of lacking. Bugs in the design or programming of Dwarf Fotress dwarf behavior lead to whacky stories created by the player filling in the blanks between one dwarf action and its absurd consequence. If a game has enough analogical connective tissue, the player can fill in the rest, sometimes with <em>more memorable</em> results than if everything were fleshed out by the designer in the first place. It’s a bizarre reward for designers who fail to achieve their aim. Town Management games are in a unique area of design where the designer has to work very hard to just make the behavior of the townspeople inoccuous. Here, in practice, the analogy works against the designer, and the designer gets a false reward when he fails in certain ways. This is a particularly perverse result of analogy-driven design.</p>
<p>On the other side of analogy’s role in design are games where analogy serves as a thin veneer over an abstractly-developed set of rules. This is often seen in Euro-style designer board games like Dominion where analogy plays almost no role except providing some nice art and names to make card concepts like “+2 Draw, +2 Buy” slightly less obtuse. In this process, the designer makes rules that work systemically, then comes up with a reason why the game world would act like that, or fabricates a game world out of whole cloth that may act like that. At its worst, this style of design simply puts arbitrary labels on game elements. In these abstraction-first games, analogy is a powerful tool left unused. Notice that games in this mode typically have very simple rules compared to analogy-driven games. Without the semiotic machinery that a guiding analogy provides, the designer is left to use his brute intellect directly in his grappling with the rules in an effort to get them into a desirable shape.</p>
<p>If only we had better tools to comprehend and intentionally manipulate game systems in the abstract, designers would not have to rely on analogy so heavily to design more complex games. But that’s an enormous problem that I will make attempts to address later. There are plenty of ways we can improve on the modern process of game design incrementally, both through attaining a better understanding of game rules in the abstract, and also through using analogy more expertly and carefully given our knowledge of its limitations, but great power, as a design tool.</p>
<p>Today I’d like to introduce a design tactic which I think leverages the power of analogy while allowing systemic concerns their due consideration. Start a design from the typical ground: pick a game or real life situation that you think is interesting and lay out the skeleton of the game. Let’s say our skeleton is Final Fantasy Tactics. Through playing FFT we’ve noticed that permadeath is present but significantly limited because the designers recognized it’s highly problematic. So let’s start from there: How do we design a tactical RPG where characters don’t die on the battlefield when their HP is reduced to 0? It seems silly at first blush because it so strongly bucks the analogy of 0 HP meaning no life left. We could, and many games do, just return all cahracters to full health between missions without comment, but the design tactic I suggest embraces the analogical difficulty of these inter-mission rules. Let’s come up with an analogical justification for characters being rescued and healed to full. Perhaps Gods, angels, and demons save the characters that gain their favor and whisk them back to home base, healing them along the way. A simple justification, but if we water this flower it will soon bloom: with the analogy established we’ve indirectly suggested more mechanics! Angels and demons are aligned with good and evil, and we can give characters alignments too. Then if they’re faithful to their alignment in their actions in battle, they’ll pay less of a cost for the between-battle healing. Now we have a new property of characters that we can tie into all kinds of other game systems, like character skills, relationships, and missions selection.</p>
<p>Follow the spiral outward from a small improvement, alternating mechanically-necessary change with analogical justification with further elaborating mechanics and you can quickly establish a rich set of properties for the elements and many interactions between them. One small change can precipitate a landslide of systemic alterations to a stale formula. All you need is a skeleton to start with and some imagination. To find the little mechanical tweaks you can make to improve a game, you’ll have to use your intuition for now, but soon you’ll also have a bunch of concepts that I use, as well.</p>In my article on analogy, I hinted that designers design abstract mechanics and then apply analogy to make the system most accessible to players. That is not a representation of how most people design video games today, it was merely a set-up for talking about analogy without diving too deeply into the properties of the abstract systems that underlie games.Analogy2017-02-13T00:00:00+00:002017-02-13T00:00:00+00:00https://evizaer.github.io/analogy<p>Analogy is how the designer bridges the gap between the player and the abstract mechanics of the game. Analogy makes relatable and relevant what would otherwise be a litany of abstractions and seemingly arbitrary relationships.</p>
<p>Games and gameplay are often presented with expressive graphics representing familiar concepts that help the player learn how to play and quickly come to grips with the current state of the system through analogy to the real world. Even when games don’t try to model pieces of life with any accuracy, as in typical combat in a turn-based RPG combat system like Final Fantasy’s, there are myriad graphical symbols used to denote in-game concepts, and the player can more easily relate to the game because of their referential power.</p>
<p>As with any art, knowing your audience is essential when you are designing a game. Picking certain analogies over others, then using the more meaningful terms and symbols that the player will understand, leads to rapid, effective communication from game to player. Good analogies draw the player in quickly, letting them buy into the game’s systems and fluently begin what will hopefully be a long and fruitful process of experimentation and learning. Good analogies do not only make the game easier to learn, they speed up the player’s ability to internalize complex game states through usage of a mnemonic form that’s easy to memorize. The logical structure of the game already does this (<a href="http://theinvisiblegorilla.com/blog/2012/02/15/how-experts-recall-chess-positions/">Chess players remember sensible board positions much better than random ones</a>), and that’s compounded by the additional layer of mnemonic power derived from analogy.</p>
<p>Analogies can convey complex meaning quickly and neatly wrap up many rules in an easy to grok package. In a Civilization game, the tech tree and each individual technology the player can research rely on analogy to convey a relatively complex idea–that of the advancement of civilization granting great capacities to make money, war, and cultural artifacts. Civilization games present a much more limited subset of the implications of this broad idea, but even an unfamiliar player can, just through interpreting this analogy, quickly come to grips with enough of the rules of the system to get started using it. Modern Civilization games have thousands of rules behind the scenes that play into technological advancement, but the player can use the UI and their life experience to make educated guesses at the general form of most of those rules and they’ll gain a significant portion of the knowledge they need to make the decisions they want to make. They can fit the game’s own idiosyncratic interpretation into this broader framework with a lot less effort than they might memorize one new abstract rule atop thirty they’ve already memorized. Compare Civilization games’ rich analogical structure to Chess. Chess has very little rule complexity compared to Civilization, thus its analogical structure needs not be as strongly tied to its mechanics. “Rook”, for example, is not a word familiar to a modern player outside of the world of Chess, but there are so few rules attached to the name–merely two movement rules and castling–that this weak signifier doesn’t handicap the player. This is typical of abstract strategy games. Their rules are remarkably simple and they need to be since their design asthetic of being an analogical desert prevents them from sustaining much complexity without making the interpretational task of the players nigh insurmountable.</p>
<p>Strong analogy has downsides, too. The player’s <a href="https://en.wikipedia.org/wiki/Functional_fixedness">functional fixedness</a> bias makes her more likely to stick to the analogical use of game elements instead of constructing a game-specific model immediately. Multiplied by the many components in a typical strategy game, the implications of the references in a design can mislead a player. Using analogy to convey information quickly is a double-edged sword and must be treated with care.</p>
<p>In short, analogy is the tool the designer uses to package the complex and intricate into a cognitively-friendly form. If a game is about some area of human experience in which a player already has interest, the game can share in the cognitive benefits of that interest. Analogy is incredibly important to making games feel relevant, and to making games playable in the first place. Thus analogy is one of the three foundational aspects of games that designers must keep in mind at all times.</p>Analogy is how the designer bridges the gap between the player and the abstract mechanics of the game. Analogy makes relatable and relevant what would otherwise be a litany of abstractions and seemingly arbitrary relationships.The AVA Paradigm2017-02-12T00:00:00+00:002017-02-12T00:00:00+00:00https://evizaer.github.io/Introduction-to-the-AVA-Paradigm<p>My method of understanding and designing strategy games starts off with a triad of critical concepts which I call the AVA paradigm–a somewhat memorable acronym combining the first letters of Agency, Variety, and Analogy. The interaction of these three concepts lies at the core of strategy game design. They each represent certain values which one must balance when designing any strategy game, and their substance is what attracts players and keeps them playing.</p>
<p>I’ll be posting more detailed discussions of each concept. Each one could have volumes written about it. But these brief discussions at least give us a place to start in our journey towards better strategy game design.</p>
<h3 id="agency">Agency</h3>
<p>Without the player’s actions having some meaning within the game, it’s not worth playing the game. It’s self-evident in strategy game design that the player’s actions must matter.</p>
<h3 id="analogy">Analogy</h3>
<p>Without the outside world to provide substantial meaning to the abstractions of a game, the game seems trivial. Through the power of referential game components, players (and designers!) can more quickly make sense of complex ideas and systems in games.</p>
<h3 id="variety">Variety</h3>
<p>Without enough to try and to do within a game, players quickly grow bored and move on. Variety takes many forms, both profound and insignificant, in games’ designs, and getting a grasp of how a design idea can admit variation without harming its own stability and life-expectancy is a topic of extreme importance to strategy game design.</p>
<h3 id="their-interaction">Their Interaction</h3>
<p>Often games will pay for better expression of one of these concepts by sacrificing another. Historical strategy games eternally struggle with the fact they model historical events and thus must counteract the influence of the player in some way–sacrificing agency at the altar of analogy. Abstract strategy games present situations which find no match in lived experience, and revel in the freedom this grants them–they sacrifice analogy on the altar of variation and often agency. In roguelikes, the player is often at the mercy of elaborate procedural generation systems hosting thousands of items and monsters; the player thus is likely to run into randomly spawned obastacles which he cannot surpass, but, as Tarn Adams says, “losing is fun”–especially when you sacrifice agency at the altar of variety.</p>
<p>In this world of sacrifice and trade-offs, though, I entertain the hope that we can squeeze as much out of all three of these concepts without necessarily sacrificing one for another. It is not a product of good design that the designer picks one of these at the cost of the others–it is a product of great design that the designer takes as little penalty and as great reward from each in spite of and because of the tensions between them. I hope in my explorations on this blog to get us closer to mastering this triad of critical concepts so we can design better games.</p>My method of understanding and designing strategy games starts off with a triad of critical concepts which I call the AVA paradigm–a somewhat memorable acronym combining the first letters of Agency, Variety, and Analogy. The interaction of these three concepts lies at the core of strategy game design. They each represent certain values which one must balance when designing any strategy game, and their substance is what attracts players and keeps them playing.