作者：Raph Koster

针对MMO游戏《星球大战：旧共和国武士》中战斗记录（游戏邦注：现实战斗行动具体数值的卷轴式文本窗口）消失的情况，有些人为此感到遗憾，因为他们使用该记录来优化操作，将其作为学习的工具。有些玩家刚将其视为文字游戏时代的遗物，或者加速整个系统崩溃的功能，认为战斗记录会影响到游戏的趣味因素，应当去除。

事实上，双方的观点都是正确的。战斗记录只是种反馈的形式。系统给予你的反馈越多，你了解到的进程相关信息也越多，也就越能弄清楚系统的运转方式。这使玩法的优化变得更加简单。

在这个方面，我们首先应当认识到，游戏向你展现的所有东西本质上都只是种反馈的形式。棋盘上棋子的位置，也就是“游戏状态”，是种反馈。从敌人身上跳出的数字是反馈，挥舞的剑展示出的炽热发红效果同样也是反馈。

星球大战:旧共和国武士(from republictrooper.com)

战斗记录(from toroz.com.au)

特定形式的反馈会更为适合特定类型的信息。

比如，通过图表来呈现矢量和物理信息比通过数字来呈现更容易让玩家理解。

反馈还可以提供情感内容，比如使用音效或颜色可以较好地对比呈现事件的好坏。当然，我们还可以通过许多其他的形式，比如间隔时间的长短或特定的颜色等，这些都比文字更容易触发玩家的情感。

某些形式的反馈在信息角度上比其他形式更强，它们较难以过滤或者拥有多种接触用户的渠道，因此难以在众多信息中突显自身信息。

某些形式的反馈要进行具体化，才能够更好地传播给用户。

对于极客群体而言，你使用的反馈形式本质上是信息理论中解决特定问题的方法。

因为从根本上来说，玩家是在努力弄清楚黑盒中的规则，所以必须将某些形式的反馈作为基本内容来呈现，否则玩家会一直按动按钮而且认为游戏毫无进展。在这种情况下，他们会构建起一种心智模型，认为这个黑盒其实是个空盒子，只是外面装饰了些许伪造的按键。记住，玩家与模型间的互动是启发式的。它不一定要非常准确，只需要有合理的预测性能力即可。

但带有不良反馈的深层次系统，玩家依然会觉得它很肤浅。比如，如果你的NPC只有3种呈现情绪的面部动画，那么即便你为其设置12种不同的愤怒层次也无济于事，玩家根本感觉不到。玩家会认为只存在3个层次。更糟糕的情况是，在许多模拟类游戏中，你可能会设计出健全且细节化的模拟世界，但是因为反馈不当无法让玩家看到，或者玩家只是认为它是个很简单的世界。

《星球大战：旧共和国武士》技能系统(from keenandgraev.com)

所以，无论玩家是否会查看战斗记录，你都必须展现已造成伤害的事实，最少要呈现攻击目标的当前状态（比如当前HP）和攻击发生后的状态（比如攻击动作发生后的HP）。这也正是我们经常在游戏中看到计量尺和数字的原因。

然而，单单用计量尺无法准确地呈现伤害来源之类的信息，而战斗记录可以完全细节化地反馈所有的系统深度。这才是那些关心谁对他们造成伤害的人希望保留战斗记录的原因所在，这是种新的信息。

如果你的战斗系统很肤浅，那么可以减少反馈的数量。如果你设计的是深层次战斗系统，那么你的反馈就应当呈现出系统的深度，否则你就要削减系统的深度，因为玩家很难感受到系统深度的存在。在多人战斗场景中，人们显然知道每个人都能够做出动作，但是如果反馈内容不当，他们可能无法分辨出哪些动作与哪个玩家有关。

这个问题自然可以得到解决。如果每次攻击都能在玩家和目标间用带有颜色的线条连起来并在线条上呈现数值，这样你就能准确地知道谁做出了什么动作。事实上，这可以算是种很有价值的呈现信息的方式，其清晰性比快速滚动的战斗记录更好。但是它有两个缺点：它可能会影响炫丽动作效果的呈现；它缺乏记录性，你无法在效果产生后对其进行评估。但是，即便带有这两个问题，依然有办法将其解决。

这也就是战斗记录依然存在于各种游戏中的原因，尽管它的视觉效果并不好，而且很难跟踪，但它们的存在满足了那些有兴趣发掘战斗系统深度的玩家的需求。它们可以被捕获和分析。它们可以被用作源数据流，转变成更多种类的视觉工具。这并不意味着战斗记录是实现上述目标的最佳反馈设计，但是它足以解释为何硬核玩家希望游戏含有战斗记录。

那么，它是否会破坏游戏，游戏是否会因为拥有这些信息而影响到趣味性呢？

以下两句话值得思考：

1、成功游戏的价值在于，它能够在玩家停止玩游戏前教会玩家所有其应该学会的东西。

2、趣味只是个过程，乏味是它最终的宿命。

去除战斗记录而优化游戏趣味性，对我来说这只是种预想中的情况。玩家会合理将系统弄清楚。这并非反馈机制本身的过错，它只是加速或减缓玩家理解系统的进程。趣味性存在于整个发现过程。如果游戏将所有内容都罗列出来呈现给玩家，那么这款游戏的寿命也不会太长久。

游戏邦注：本文发稿于2011年12月8日，所涉时间、事件和数据均以此为准。（本文为游戏邦/gamerboom.com编译，拒绝任何不保留版权的转载，如需转载请联系：游戏邦）

Notes on game feedback

Raph Koster

I was mentioned in a comment on Google+, and ended up writing a little bit about game feedback as a result. So here it is.

The discussion was on the absence of combat logs (scrolling text windows showing you exact numbers for combat actions) in the new SWTOR MMO. Some folks regret the absence, because they use the logs to optimize what they are doing, and use it as a learning tool. Other players find them a legacy of the text mud days, or a feature that hastens the deconstruction of the entire system and therefore damages the fun factor.

Both sides are right, really. Combat logs are just a form of feedback. The more feedback the system gives you, the more information you have for the process of figuring out how the system works. This then makes the process of optimizing play easier (read that as “getting the results you want from a given input”).

The first thing to realize here is that everything the game shows you, really, is a form of feedback. The locations of chess pieces on a board, the “game state,” is a type of feedback. Numbers floating off the enemy are feedback; the glowy effect trailing a swinging sword is also feedback.

Some forms of feedback are better suited for certain types of information than others.

Vector and physics information is more easily grasped when presented graphically than when presented as numbers, for example.

Feedback intended to provide emotional content, such as “this was a good event” versus “a bad event” is typically better conveyed using sound or color. We have a lot of associations with things like major vs minor intervals, specific colors, etc, that are more easily triggered this way than with, say, text.

Some forms of feedback are more susceptible to noise in the signal than others — they are harder to filter, or might have multiple channels coming at the user at once, making it hard to distinguish between multiple simultaneous messages.

Some forms of feedback are best conveyed to users by concretizing them, using analogies that better get across the information.

For the geeks among you, the forms of feedback you use are essentially exercises in solving particular issues in information theory.

Since a player is essentially trying to figure out the rules inside a black box, some form of feedback must be present as the base case, or else the player can press buttons all day long and to them, it looks like nothing is happening whatsoever. In that case, they build a mental model of the black box as being an empty box with some fake buttons on the outside of it. Remember, the player is attempting to arrive at a heuristic for interacting with the model. It does not have to be accurate; it just needs to have reasonable predictive power.

The corollary here is that a deep system with poor feedback will read as shallow to players. For example, it doesn’t matter if your AI NPCs all carefully track twelve levels of anger if they only have three facial animations to display them. Players will instead decode this as three levels. Worse, you can have a situation as in many simulation-based games, where you might have a robust and detailed world simulation that players can’t see, or that feels to them just like a much simply hardcoded state machine (this is the trap that the original ecology in Ultima Online headed into).

OK, so in terms of looking at whether or not to look at combat logs… you have to show the fact that damage is being done at all, of course; that requires at minimum displaying one or both of the current state of the target (aka, its current HP), and the delta that a given action resulted in (aka, the damage done). This is why we tend to see meters (which do state very well, and deltas less so) and floaty numbers (which provide the delta, with higher resolution than a meter does).

But: meters alone have poor granularity and low readability for nuances like source of damage. Combat logs prove full detailed feedback exposing the full depth of the system. And that is why people who care about, say, who did what damage (a whole new type of information) want that feedback.

If you have a shallower combat system you can cap at a shallower amount of feedback. If you have a deep system, your feedback should accommodate revealing that depth or else you may as well cut the depth because people will often literally not be able to tell it is there. In the case of a multiplayer combat scenario, people will obviously know that everyone did something, but they may not be able to tell which delta was associated with which player, leading to arguments over the effectiveness of a given team member.

This can be solved. Every attack could draw a color-coded rope between the player and the target, and tie the delta to that rope, so you could see exactly who did what. It would in fact be a very high-value means of displaying the information, with greater clarity than a rapidly scrolling combat log. But it has two disadvantages: it’s clinical and would work against cool force lightning effects; and it lacks history so you can evaluate it after the fact. But even these things could probably be solved.

This is why combat lags are still around even though they are ugly, and painful to follow in real-time. They satisfice the requirements for players who are invested in investigating the depth of the combat system. They can be captured and analyzed at leisure. They can be used as a raw data flow that can be translated into a variety of more visual tools. This doesn’t mean combat logs are the best feedback design for the purpose, but it explains why the more hardcore (ie depth-invested) players have a desire to have it.

So does it ruin the game and suck the fun out of it to have that sort of information, and the fun “optimized away”?

To pull random perhaps-misremembered quotes out of the book:

“The definition of a successful game is therefore one that teaches everything it has to teach before the player stops playing.”

And

“Fun is a process; boredom is its destination.”

The fun getting optimized out sounds to me like exactly the expected behavior. The players collaboratively figured out the system. This is not the fault of the feedback mechanism per se; all that does is accelerate or reduce the rate of player progress towards their heuristic. Fun is in the discovery process (barring subtleties like replay as meditation, or the joy in perfect execution of a coordinated strategy which is effectively a different game). If it’s all laid out as a schematic for you, it’s over. Move on. It’s time for the next game (which might very well just be a different system within the same game) or MMO. (Source: Raph Koster’s Website)