原文名：Plan Disruption

翻译名：规划扰乱

作者：Ella Hoeppner

已获得原作者翻译、发布授权。

译者：TaoistPunk（游戏学院）

策略游戏（strategy games）中的策略思维可以认为是一种策略规划形成的过程。玩家获取游戏中的信息，然后基于这些信息和对游戏的理解，形成策略规划。当玩家的策略规划被扰乱时，玩家必须去调整自己的规划，而这种调整的过程同时也是一种新的策略规划形成的过程。恰恰是这种策略规划的形成过程以及对达成目标的多种策略途径的权衡，让游戏玩起来更有趣。所以理解游戏结构【注1】是如何对玩家的策略规划的影响是很重要的。

【注1】 【译者】游戏结构：原文为Game structure。笔者对此的理解为，如果将游戏划分为静态、动态的两个部分，其一些客观的事物（如机制、规则、系统、数据、美术资源等）之类的静态事物为Structure（结构），而主观的感受如玩家的体验、情绪或者时间则是动态的，则是非结构的。

导致玩家改变自己的策略规划的主要原因即是游戏中不为玩家所知的信息【注2】对玩家涌现出来了。当新的信息被引入时，玩家之前的策略规划就失效了，随后玩家必须去调整自己的策略规划，甚至有时要想一个新的策略。不同的游戏以不同的频率和不同的范式引入信息，我称之为“信息流”（Information flow）。不同的信息流范式以不同的方式影响玩家形成策略规划的过程，我认为对于策略游戏设计者而言，理解信息流如何影响玩家的策略规划，以及在自己的设计中合理地运用信息流是很重要的。

【注2】【译者】不为玩家所知的信息：原文为Hidden information。hidden一词在中文中直译为被隐藏的，但原文的意思并非是某种主观意志在刻意隐藏什么信息不让玩家看到，而是从玩家的角度出发，主观上不知道的信息。举个例子，当对方玩家有A和B两种进攻策略可以选择时，我方对于敌方玩家的策略是无从得知的，这种“无知”并非是设计师或者对方玩家故意隐藏的，而是一种基于玩家本人的现有信息无法得出有效或者准确结论导致的。故此处翻译不采取直译，而是意译。

不过，不同的信息对信息流的影响的影响是不同的，新涌现的信息对于玩家的扰乱程度是截然不同的。有一些信息对于玩家的影响比其他的信息要高出一些，比方说，在《XCOM》【注3】中，下一个敌人的位置信息就比某一次射击是否命中对玩家的影响更大。信息流即是在特定时空下涌现出来的信息对玩家的影响之和。

【注3】【译者】XCOM：幽浮系列。

一般来说，低影响度信息（Low-impact information）影响低级策略规划（Low-Level Plans），高影响度信息（High-impact information）影响高级策略规划（High-Level Plans）。高级策略规划即是那些更泛、更抽象的目标的规划，而低级策略规划则是那些更具体、更集中的策略规划。

高级策略规划又可以认为是低级策略规划的组合。如果你有一个高级的策略规划，比如“我要干掉那些怪物”，那么你会把这个规划切分成若干个更细致、具体的规划，比如“我要先接近那些怪物”、“我得设法先弱化这些怪物”、“然后我发起攻击”，每一个低级策略规划都可能被一些其他的规划改变或者完全推翻，但是高级策略规划并不一定会随之改变。

在一个游戏中，如果信息流的速度过快，对于玩家来说就很难形成一个长周期策略规划（Long-term plan）。任何长周期策略规划在过快的信息流当中，都会被新涌现的信息给扰乱，最后策略规划就毫无用处了。而对于玩家而言，他们会不再进行长周期的策略规划，进而依靠一种临时性的、短期的策略规划来推进。换言之，游戏转而成为一种战术性质的进程，而是去了战略性的潜力【注4】（特别是在一些非常抽象的战略规划上）。

【注4】【译者】战术性：原文为tactical；战略性，原文为strategic 。前者指在具体环境下的操作对策，后者指宏观的目标方针。

而相反，在一个过慢的信息流的游戏中，玩家的策略规划不怎么会被扰乱，这意味着玩家能对游戏的未来进程进行很长的规划。举个例子，我们可以想象一个游戏，一个展示了全部信息的游戏（Perfect Information Game，下文翻译为全信息游戏），一个纯粹的单人游戏，一切信息都是完整地被玩家知晓。这很像是和一个非常简单、棋路单一的AI对手下象棋的感觉。AI对手要极其简单，简单到玩家几乎太需要过度思考，信息流中玩家没有思考到的信息也对玩家的策略规划毫无影响。这种游戏被Keith Burgun称之为Lookahead contest。玩家会在这种游戏开始，就可以计算出整个游戏的过程，并且一口气规划策略到游戏的结局。策略规划从一开始就已经结束了，导致整个游戏过程显得不那么有意思。整个游戏过程中，没有什么有效的信息涌现出来，玩家的策略规划完全不会受到扰乱，玩家也就没有任何理由要去做任何新的规划了。

信息流速度缓慢的游戏也会遇到同样的问题，但程度相对而言没有那么深重。这种游戏里没有足够的涌现信息去打断玩家的策略规划，大部分的游戏玩法是由一些既成的策略规划组成的，而不是说费尽心思想出一些新的套路去解决游戏里的挑战。

我们有很多种方式去做一个可视的信息流图形，我无法描述或者评论这么多种方案。不过，我认为这其中有一种是值得关注的，我称之为“峰值”信息流图样。

这个图像以时间为横轴，信息流的扰动程度为纵轴，表达出了一个带有峰值的信息流：

举一个很好的例子：2012年重制的《XCOM》和系列续作《XCOM2》。《XCOM》系列的特点就是其信息流的速度相对比较慢，但是会遇到一些高影响信息涌现而形成的信息流峰值。在《XCOM》中。这类峰值在《XCOM》中往往发生在遭遇新的敌人的时候。一般来说，峰值具有足够的能力去干扰所有的低级策略规划，进而让玩家创建新的策略规划。高级策略规划有可能会平安度过这些峰值而不受影响，但低级策略规划的区间往往是介于两个峰值之间的时空当中，而不是直接越过峰值。在《XCOM》中，不管你在什么时候发现了一个新的敌人，你的高级策略规划都不会因此而动摇，比如跑去地图上某个点，阻止敌人伤害特定的目标。而你在面对低级策略规划时，你有大量的涌现信息需要去处理，低级策略规划则会因此而受到扰乱，比如移动你的单位或者决定下一次攻击哪里。

在两个峰值之间的信息流，我称之为静息流【注5】，这种信息流相对较慢，不会对中级策略规划造成太大影响，同时又可以让玩家在两个峰值之间的时段不断地应对一些琐碎的小状况，让整个过程又不至于非常无聊。在《XCOM》中，静息流往往来自射击结算和敌人行动之后。在机制中蕴含的信息也让玩家在两次遇敌之间的时间段里能够根据新涌现信息去调整自己的策略规划。如果《XCOM》没有静息流状态，并且除了敌人的位置之外都是确定的，那么玩家就能够在激活后立即形成一个解决每个敌人的策略规划，由于这个策略规划不需要做任何调整，实际执行起来会让人感觉相当枯燥。

【注5】【译者】静息流：原文为Idle flow。Idle在游戏中的概念常译为待机，其他意为空闲、空转。在文中，Idle flow在下文也翻译了，是一种较慢，但是并不为零的信息流。译者考虑过多种中文词，例如待机流、低熵流、闲置流，甚至考虑过空流（佛了），但都很难达到文中想要表达的意义，故还是采取了一个相对接近的“静息流”。

如果一个游戏有较为稳定恒常的信息流，它的静息流往往比普通状态的信息流更慢，因为即将到来的峰值会妨碍玩家形成更加长远的策略规划。不过，保证静息流不至于过慢依然很重要，不然每个静息区间都会变得像前文所说的全信息游戏，于是所有的问题就变成了：玩的过程是由根据每个信息流峰值的影响所形成的策略规划组成的，以及一些发生在两个峰值之间的新的策略规划。

另一个例子是我自己制作的游戏《Brazen Berry Bonanza》【注6】(下称《BBB》)。

在《BBB》中，你需要一遍保护你的作物不受到其他敌对植物的侵袭，一边种植新的作物。其中我面临的一个早期的设计问题是，如何生成敌对植物。我第一个想到的方案是以一个较短的固定时间间隔，单次生成1个敌对植物。在这个设计中，我发现信息流速度过快了：我几乎无法让我的植物安全地活下去。

然后我试了第二种方案：一波一波地生成多个敌对植物，每一波的间隔比第一个方案中较短的固定时间间隔要更长。结果就是整个游戏的体验大幅改善了。其实从直觉上说我会感觉到这样做更好，但是我却没能给出一个理论上的解释，也正是这个契机让我想到了这篇文章中的“信息流”的概念。

总得来说，我觉得信息流这个方法有两个优点：

如果一个游戏的信息流太快了，把大量的高影响信息集中到峰值处可以有效改善这一问题。你可以假想一个没有峰值流版本的《XCOM》，一个信息涌现更加平缓均匀的版本。不是隔几个回合生成大量的敌人，而是每个回合均匀生成1~2个敌人。想想都不是那么好玩了吧？其实很大程度上就是因为信息流过快了。对于玩家而言，在过快的信息流当中，过饱和的信息扰动让玩家很难形成任何有效的策略规划。而通过分组刷新敌人代替均匀生成，静息的时间段则被分成一块一块的了，对于玩家而言这是很好的规划策略的时间。

这种思路避免了低信息流游戏通过给低级策略规划加上的硬性限制的问题。通常情况下，一个低信息流游戏给了玩家在低级策略规划的时候为长远规划做计划的间隙，同时峰值信息流有让玩家知道了接下来会发重大变故，我必须得去解决当下的问题。过慢信息流的游戏可以通过引入偶发峰值来改善游戏的节奏。

《XCOM》和《BBB》之间最大的不同点在于峰值的位置和静息流的频率，《BBB》相对而言要更加稳定。

在《XCOM》中，当玩家激活一个新的遇怪点时，即会产生一个新的峰值。在《XCOM》中，玩家既可能接连遇到两组敌人，也可能一大段时间完全遇不到敌人，所以《XCOM》本身并不需要那种非常固定恒常的信息流。而《BBB》则相反，峰值发生在敌对植物生成的时候，生成的间隔严格保持1分钟。此外，界面上随时会现实下一波敌人生成的时间倒计时。

这种稳定性的好处在于，玩家会知道他们在什么时候会被扰乱自己的策略规划，也因此，玩家的长周期规划也不会被意外终止。过快的信息流会妨碍玩家产生长周期策略规划，稳定的峰值则会尽量不干扰玩家的长周期策略规划。如果玩家清楚地知道下一次峰值何时发生，他们会自信地去规划策略，一直到被未知的峰值干扰未知。

当然，游戏并不是必然有一个稳定的信息流，《XCOM》即是如此。而《XCOM2》在《XCOM2012》上明显考虑过这一点，不仅在玩家会遇到敌人，还会有敌人呼叫的增援部队（这种增援往往有数个回合的预警），这些都会形成新的信息流峰值。

《XCOM》的静息流同样也不是稳定的。它的静息流主要有射击和敌人移动构成，这两种都是由于敌人行为的偶然性造成的。敌人越多，信息流越快，相反越少则越慢。这意味着静息流在遇敌发生后立刻攀升到最高点（而形成峰值），然后随着敌人被干掉而逐渐降低。而如果玩家又遇到了第二组敌人，那么在第一次遇敌后，玩家将再一次面临一个峰值。

在《BBB》中，静息流源自植物的种植方式，种子和浆果的生成，以及玩家获得的新种子类型，这所有的信息流都是一种规律的状态。这也就意味着静息流的频次基本上是差不多的，以及，作为设计师的我，对游戏流程就有了更好的控制。

如果我们看一下《BBB》的信息流图标，我们会发现基本上各段的峰值分布是一致的；而《XCOM》则看起来不同，峰值流和静息流可能会以不同的频次分布，如下图：

玩家制定策略规划的过程正是策略游戏之所以如此有趣的重要的原因，而信息流则在其中对玩家的策略规划过程有非常重要的影响。对于设计师而言，让信息流的速度保持在既不过快、也不过慢的区间之内确实很重要，但同时，我们更需要时时审视信息流随时间的变化。峰值流的设计是一个非常好的范式，高影响度信息被集中在峰值附近，而峰值之间的信息流则会变得平缓规律。仔细审视游戏中的信息引入方式，以及它们是如何影响到玩家制定计划，对于设计一款策略游戏来说是非常重要的。

感谢 Evizaer 帮助我编辑这篇文章。

The strategic thinking that happens in strategy games can be thought of as a process of forming plans. The player takes in information, and then forms a plan based on that information and their understanding of the game. When a player’s plan is interrupted, they are forced to adapt, and the process of adapting can also be thought of as forming a new plan. The process of forming plans and then learning which types of plans are better or worse at accomplishing the goal of the game is what makes games fun. Thus, it is important to understand how the structure of a game affects the player’s plans.

The main thing that causes players to change their plans is information that had previously been hidden becoming available to the player. When new information is introduced, plans that the player formed previously become outdated, and the player has to adapt their plan, or come up with a completely new one. Different games introduce information at different rates and in different patterns, and I call this property the game’s "information flow". Different information flow patterns have different effects on how players can form plans, and I think that it is important for strategy game designers to actively consider information flow in their games, and how this flow influences the way players make plans.

Impact, and High-Level and Low-Level Plans Not all information affects the flow equally, because the degree to which new information disrupts the player’s plans is not the same for all types of information. Some information has higher impact than other information. For instance, information about the location of the next enemy pod in XCOM is higher impact than the information about whether a particular shot hits or misses. The information flow is defined by the sum of the impact of all the information being introduced in each moment

Generally, low-impact information affects low-level plans, while high-impact information affects high-level plans. High level plans are those that are general and abstract, while low-level plans are more specific and concrete. High-level plans can be thought of as being composed of lower-level plans. If you have a high-level plan of "I will get rid of these monsters", you could accomplish that by several different low-level plans like "I will approach the monsters", "I will cast a weakening spell on the monsters", and "I will attack the monsters", and each of these low-level plans could change or be replaced with something else without getting rid of the overarching, high-level plan of "getting rid of these monsters".

In a game with too fast a flow, it will be impossible for the player to form a long-term plan. Any long-term plan that the player forms will quickly be disrupted by the new information, and will end up being useless. The player will simply stop forming long-term plans, and handle the game on a purely turn-by-turn basis. In other words, the game will be purely tactical, with no potential for any long-term strategic planning (except, perhaps, on a very general and abstract level).

In a game with too slow of a flow the player’s plans won’t be disrupted enough, meaning the player can form plans that reach to far into the future. To demonstrate this, we can consider a game with the no information flow at all. In other words, a single-player game with perfect information. This would be something like a single-player chess game with a very simple, deterministic enemy AI (real chess doesn’t count here, since the future moves of the other player are hidden information) The enemy AI needs to be simple, because if it’s too complex the player won’t be able to consider it very effectively, and effects caused by information the player hasn’t considered are effectively equivalent to new information. This type of game would be what Keith Burgun calls a "lookahead contest", where the player can endlessly calculate into the future, and thus form plans for the entire game from the very start. In a game like this, all planning can be done at the very start of the game, and thus the process of actually "playing" the game isn’t very fun; it’s simply following the plan that you came up with at the start. Since there’s no new information coming in, your plan never gets disrupted, and so there’s never any reason to do any new planning.

A game with a slow, but non-zero information flow will suffer from many of the same problems, but to a lesser degree: there won’t be enough new information interrupting the player’s plans, and much of the gameplay will consist of simply following out plans, instead of coming up with them (the fun part).

There are countless conceivable shapes an information flow could take, so of course I can’t describe or comment on all of them here. However, there is one type of flow that I think is worth focusing on, what I call a "spiky" information flow.

To wrap our heads around variable information flow, it helps to render it as a graph, plotting disruption against time. This graph represents a spiky information flow:

A good example of a game with roughly this type of flow is the 2012 remake of XCOM, and it’s sequel XCOM 2. This type of flow is characterized by having a slow flow most the time, but with occasional spikes, where a lot of high-impact information is introduced. In XCOM, these spikes would be the discovery of new enemy pods. In general, spikes have impact large enough to disrupt the almost all low-level plans, and cause the creation of new plans. High-level plans can potentially survive past the spikes, but the low-level plans are restricted to the periods of time between spikes, not across them. In XCOM, whenever you discover a new enemy pod, your vague, high-level plans like trying to get to a certain point on the map, or trying prevent a particular unit from taking any damage, can stay the same. Your low-level plans, like where you’re going to move your units or what to shoot at next, are completely interrupted, since you have so much new information to deal with.

The flow in between the spikes, which I’ll call the "idle" flow, is there to make sure the player still has to adapt somewhat in between the spikes, but is slow enough to allow mid-level plans to last for a while. In XCOM, the idle flow comes mostly from the results of shots and from enemy movements. The hidden information involved in these mechanics ensures that the gameplay between the activation of pods still requires adaptation. If XCOM had no idle flow, and was deterministic other than the location of enemy pods, the player would be able to form a plan for dealing with each enemy pod right after it activated, and the process of actually carrying out the plan would be boring, since it wouldn’t require any adaptation.

The idle flow can be slower than what might normally be ideal if the game had a constant flow, because the upcoming spikes prevent the player from forming a plan that goes too far into the future. However, it’s still important to make sure the idle flow isn’t too slow, or each idle period will be almost like it’s own perfect information game, with all the problems associated with that: play will consist mostly of following out plans formed right after each spike, with little new plan-formation happening between spikes.

Another example of a game that uses a spiky flow is my own game, Brazen Berry Bonanza.

In Brazen Berry Bonanza, you try to grow certain plants while preventing other, evil plants from growing. One of the early design decisions I faced was how to spawn evil plants. The first and most obvious solution was just to have a single evil plant spawn every once in awhile, at a regular interval. In trying this, I found that the game had the problems that I described as the problems of a game with too fast a flow: it was impossible to form a plan past the immediate future.

I then tried an alternative: instead of spawning one new evil plant at a time, evil plants would spawn several at a time in "waves", with the time between waves being longer than it had been in the one-by-one spawning. The result was immediately and drastically better. Intuitively it seemed to make sense that this was better, but I didn’t have a good theoretical explanation as to why it was better, and it was that situation that lead me to come up with the ideas I’m describing in this article.

In general, I think there are two big advantages to using a spiky information flow:

If a game has too fast of an information flow, changing to a spiky flow can slow down the information flow for the majority of the game by gathering much of the high impact information into the same area. Imagine a version of XCOM without a spiky flow, one in information was introduced more consistently. Instead of the enemies being introduced in large chunks every few turns, one or two enemies would be introduced each turn. This version would be less fun than real XCOM, largely because the information flow would be too fast. It would be difficult to plan for anything past the current turn, since the new enemies would immediately disrupt any plan you made. By having the introduction of new enemies group into occasional chunks instead of being evenly distributed, the time between chunks becomes much easier to plan for.

It avoids the normal problems of a slow flow game by putting a hard limit on how far low-level plans can reach. Normally a game with a slow flow allows the player to form low-level plans too far into the future, but in a spiky flow the player knows there is going to be a huge event coming up that will disrupt all of those plans. Games with too slow of information flows could perhaps be improved using this pattern by introducing occasional spikes to prevent planning that goes too far.

One important difference between XCOM and Brazen Berry Bonanza is that the positioning of the spikes, and the rate of the idle flow, is more consistent in BBB than in XCOM.

In XCOM, a new spike happens when the player activates a new pod. This means that the spikes in XCOM aren’t necessarily very consistently spaced, since it’s possible for the player to discover two pods in very quick succession, or go through a long period of not finding any pods at all. Comparatively, Brazen Berry Bonanza has a very consistent flow. The spikes are waves of new evil plants spawning, which happen on a perfectly consistent interval of around 1 minute. Additionally, there is always a clock displaying the time until the next wave.

The advantage of this consistency is that the player knows exactly when their plans will be disrupted by a spike, and thus won’t ever have a long-term plan unexpectedly interrupted. Just as a flow that is too fast discourages players from planning too far into the future because of constant disruption, the constant threat of a spike can do the same, to a lesser extent. If the player knows exactly when the next spike will happen, like they do in Brazen Berry Bonanza, they’ll be able to confidently plan up until the spike without fear of an unexpected interruption.

Games don’t always need to have perfectly consistent spike-schedules, but I do think that XCOM allows for too much inconsistency. XCOM 2 improved on XCOM2012 in this regard, by having spikes come not only when the player discovers a new pod, but also when a group of reinforcements is called in, which is telegraphed several turns in advance.

The idle flow in XCOM is also inconsistent. It comes mostly from the results of shots and enemy movements, both of which are contingent on the existence of enemies. When there are more enemies, there will be a faster flow, and when there are fewer enemies, there will be a slower flow. This means that the idle flow tends to be at its highest right after a pod is discovered, and decreases as the enemies are killed. If the player discovers two pods in quick succession, the idle flow will be nearly double what it normally is after discovering a single pod.

In Brazen Berry Bonanza, the idle flow comes from the way plants grow, where seeds and berries spawn, and what new types of seeds the player gets, all of which tend to happen very regularly. This means that the idle flow has roughly the same rate at any point, and additionally it means that I, as a designer, have a lot more control in fine-tuning the idle flow than is possible in XCOM.

Brazen Berry Bonanza’s flow diagram would look almost identical to the previous image showing what a general spiky flow looks like, but XCOM would look somewhat different. With the possibility for inconsistency in both the spike-schedule and the idle flow, the real flow diagram of a match of XCOM might look something like this:

The essential thing that makes strategy games fun is the process of forming plans, and the information flow has a huge impact on how the player forms plans. It’s important to make sure the flow as a whole isn’t too fast or too slow, but for a more nuanced view you need to look at the way the flow changes over time. A good pattern to follow is the spiky information flow, in which high-impact information is collected into discrete spikes that happen at regular intervals, with a slow, regular flow of information between the spikes. Carefully considering the way information is introduced, and how that impacts the player’s ability to form plans, is essential to designing a good strategy game.

Thanks to evizaer for helping me edit this article.