游戏邦在:
杂志专栏:
gamerboom.com订阅到鲜果订阅到抓虾google reader订阅到有道订阅到QQ邮箱订阅到帮看

以图例解析游戏创意的成型过程

发布时间:2011-09-05 18:22:42 Tags:,,,

作者:Danc

每当我指导刚入行的游戏开发人员时,我就把自己的创意过程画在餐巾纸上,因此浪费了许多餐巾纸。为了防止再次浪费,我就把自己的创意过程图解以正规的形式展现在文本中。

我在给别人讲授创意时,第一句话就是“创意就像吞了好几个乒乓球的蛇。”

但对方通常会一脸茫然,所以我画了下面这幅图:

创意过程(from lostgarden)

创意过程(from lostgarden)

接着,我们可以闲聊一些创意的过程和潜在的问题了:

集体讨论阶段

集体讨论的缺陷

挑选阶段

挑选的失误

循环阶段

循环的问题

集体讨论阶段

我们萌生一个想法时,可能是突发其想,也可能是沉思熟虑。无论是什么,此时你就进入集体讨论或者说是空想的阶段了。这个阶段充满无限的可能和希望。我所说的“集体讨论”是泛指所有能拓展选择数量或可能性的活动。说到集体讨论,大家眼前浮现的画面可能是,一群设计师围坐在房里的白板前、或沉思或讨论,但这只是集体讨论的活动形式之一。

集体讨论的发端小,但会越来越大(from lostgarden)

集体讨论的发端小,但会越来越大(from lostgarden)

集体讨论阶段可以分为以下几个环节:

想法:最初萌生的念头。这些是预实验。比如,天气太热了,能不能发明一种能降温的短裤呢?于是,你在笔记本上胡乱地写下”太阳能短裤“。一个想法就诞生了(游戏邦注:这其实已经不是一个想法了,近来它在西雅图火了)。

思维实验:深入理解想法的实现方式和理论效果。如,你已经有了“太阳能短裤”的概念,接着就是怎么实现的问题。你想到“在短裤内安装一个小扇来降低周围的温度”,这就是一个思维实验。

真实实验:用物理材料制作模型或合用的代码,这样你就可以直接模拟你的想法了。你所做的真实实验是,做一个短裤加金属风扇(加上一个备用绷带)的模型。

交叉实验:完成模型后, 你发现新的可能。刚才我们说发明一种“太阳能短裤”,穿上之后会降温,但通过大量反复实验,我发现直接逃到有空调的咖啡店比这个发明管用多了——这就是新的可能。

进行集体讨论时会涌现了大量实验,有优有劣,且优少劣多(from lostgarden)

进行集体讨论时会涌现了大量实验,有优有劣,且优少劣多(from lostgarden)

集体讨论基本上只是一种实验的预热活动。无论你的想法有多疯狂,它也只不过是一种可能。集体讨论只是让你在这种模糊的可能性上形成一种假定,能否实现还需要更多细节性的设想和模型测试。

如果集体讨论成功了,我们就得到了一个实验组合,其中包括许多不同的开发部分:功能特征、用户案例、使用实例和“某某某周末时YY出来的东西”。

集体讨论阶段存在的最普遍的缺陷就是,创意人员没有进行足够的测试实验。这种错误产生的原因有:

创意人员自以为对想法了如指掌

实验成本高

最初想法的脆弱

创意人员自以为对想法了如指掌

创意人员追求单一的解决方案(from lostgarden)

创意人员追求单一的解决方案(from lostgarden)

既然已经知道了正确答案,为什么还要在昂贵的实验上浪费时间?这种想法的错误之处在于,创意本身是一个反复的过程,需要不断整合各种资源和可能方案,然后才能得到最终结果。创意过程不是你单纯地使用演绎法去推断唯一的正确答案。

做的实验范围太狭窄,那么你的方案就是建立在枯竭的智力和技术力资源之上。这就好比,你设计了一座桥梁,但你做模型实验的唯一材料就是纸。诚然,你还是可以做出桥来。但如果同样有人也设计桥,但他使用了更多的材料和技术,如钢材、混凝土来做实验,你敢说你的结果能和他的相提并论吗?

实验组合的力量何在?看看几个简单的数据就知道了。如果五分之四的实验系统注定失败,而你只做了一次次验,那么你的成功率只有20%。另一方面,如果你做了10次实验,你就有89%的成功率。事实上,你的成功率甚至更高,因为通过前面学习和运用新知识,你可以不断改进实验。

(游戏邦注:如果是知识性实验,那么就让尽可能多的人参与到实验中来,因为参与者越多,意味着可用的相关技术也越多。)

实验成本高

各种尝试都需要付出代价(from lostgarden)

各种尝试都需要付出代价(from lostgarden)

在以上例子中,细心的读者大约会注意到,10次实验意味着付出10次代价。然而付出的额外代价增加了仅仅450%的成功率。所以开发团队干脆来个折中,只做若干次实验。

在此,我提出的解决方案是,降低实验的成本,而不是减少实验次数。你一天当中能做多少事?一小时呢?与其5到10人一队,不如一两个人成组?把注意力放在轻实验和实验之间的迅速周转上,你就可以在集体讨论阶段加入更多次实验。

我惯用的实验“瘦身法”是使用便签式设计说明。因为你的实验必须用便利贴表示,所以你就不得不保证作用域小且容易操作。

初期想法经不起推敲

努力可以加倍,但想法总是有限(from lostgarden)

努力可以加倍,但想法总是有限(from lostgarden)

有时假,你刚想到一个看似绝妙的主意,但总是没办充实它。那么这个想法就是脆弱的——一旦你试图把概念往更有趣的新方向发展,它总是一次又一次地崩坏。

时间久了,你自然而然地就学会如何识别创意过程中产生的想法是否经得起推敲。我的检验方法是围绕相同的想法再想出20或30个变体。如果我能估计到这些变体的乐观前景,那么我就可以断定,我的想法值得进一步投入;否则,我就放弃这个经不起推敲的想法。

人们总是草率地投入数个月,甚至数年的时间,为的就是让某个想法管用。与其浪费大把的时间,不如试试其他让想法更扎实的技巧:

暂时把想法抛到一边:把脆弱的想法暂且搁置一边。随着时间流逝,如果这个想法确实有价值,它自然有办法再次出现在你的创意过程中。在新实验中触发的不同想法,可能正好抵消了对原创意的约束条件。

不久前,我为游戏《Cute God 》编写概念,但没有一个原型成样。这个设计的问题多多,解决办法却很少。我索性暂且不管它。数年后,原来的创意竟然在游戏《Triple Town》中再度现身。你淘汰掉的创意从来就没有消失,它们就像不逢时的种子,暂时钻进肥沃的土壤,等待来年的萌发。

删繁就简:你努力达到的核心实验是什么?请用一个句子或一幅图来表示。然后把你写下的或画下的东西摆在眼前。现在,用集体讨论模拟最简单的实验方式。你应该抛弃所有复杂的系统和限制,直奔核心。

最近,我和一个朋友检验了上述方法。我们用于检验的游戏概念出自一幅铅笔线稿,上面画了许多互相射击的简笔人物。这是一个终极格斗风格的游戏,在多人战斗模式下,你可以在移动各个人物的个别关节。为了检验我们的方法,我们把目光回收到最初的画稿上,扪心自问:“要在玩家的屏幕上呈现这种画面,最简单的方法是什么?””如何利用简笔画人物产生战斗的快感?“我们果断抛弃简笔画人物的娃娃身、复杂的UI和多人模式,最终的想法就成熟了——更容易执行,同时仍然保留了初期创意的精髓。至少,这个实践启发了设计师用新的眼光看待问题和质疑最初的约束条件。

这个过程可以总结为一句话:“如果你的设计太复杂,那就偷懒吧。”

挑选阶段

挑选是创意过程的第二阶段。并非所有实验都是成功的,有些实验的价值立竽见影,有些实验的缺陷则一览无余。

在挑选阶段,你需要扼杀有缺陷的实验,这样才能把注意力双倍集中于好实验上。

挑选意味着浓缩实验,粹取精华(from lostgarden)

挑选意味着浓缩实验,粹取精华(from lostgarden)

这个残酷的过程,似乎没有多少人谈到。我们总是对集体讨论的“发现”时刻大书特书,却对整合曲折的实验过程轻描淡写。这其实是弥天大”误“。

批判性地挑选你的实验是通向成功的创意的必要步骤。挑选是一个浓缩创意、确保项目有始有终的行为,且最终引出更有效力的实验。我作画的时候,总是在一个地方涂画非常多笔,但最终只有一笔是能让人看到的。之前的每一笔都是为最后成功的那笔作铺垫。这种方法让我作画时胸有成足、步步为营。

那些不能组织好“晋级赛”的人,自然没法完成有意义的项目。如何从千千万万个设计师当中辨别出菜鸟设计师?你可以建议砍掉那些看似没什么用处的功能特点,如果他们的狡辩、借口和遣责滔滔不绝,那么,你当场就能断定他们的项目肯定是一团七零八落的无用之物。

优胜劣汰(from lostgarden)

优胜劣汰(from lostgarden)

挑选阶段涉及以下三个环节:

限定挑选标准:你必须知道扼杀的依据,以免“草菅人命”。定义标准的最好方法是,写下成功的案例描述。比如,在《Bunni 2》中,我们说这款游戏是“社交贴图书”。任何不符合描述的想法都有待质疑。

重点培养:增强价值的存在感。例如,在《半条命2》中,有一个实验是关于“重力枪”的概念。当真实原型完成后,玩家们都很喜欢。Valve于是决定进一步完善“重力枪”的概念,然后把它嵌入到整个游戏中。

决定移除对象:如果有些东西不达选定标准,最好趁早把它放到一边。在《Bunni 2》中,我们自己设计了一个战斗系统。然而,我们做好简化版后,我们才意识到这个系统并不符合我们的选定标准,所以我们只好把它晾到一边了。

挑选的问题

在挑选阶段,普遍的问题有以下几种:

没有明确的挑选标准

无实体实验

越多越好

强求圆满

强调问题,忽略重点

没有挑选标准

没有挑选标准(from lostgarden)

没有挑选标准(from lostgarden)

人们通常陷入集体讨论的盲目乐观之中,却忽略达成明确的挑选标准,取而代之的是挑剔或武断。没有明确的标准,即使确实进行了挑选,好创意也只会和坏点子一起被斩杀。

所谓模糊的标准,就是有人跟你说什么好,另一个人跟你说什么不好,然后你对所有人都说有点小差异没什么。不必要的矛盾就是这么产生的。你得下狠心开展严格的讨论会,明确规定项目的标准。我喜欢“挑选标准”这个词,因为它确保你能把实验存好以备他用。明确规定共同的目标,可以避免各自为政、群龙无首。

无论我写什么设计文件,我总是把“目标”写在最前面。虽然目标只占据设计很小的一部分,但它往往是致胜的关键。成功的彼岸四通八达,但只有目标才能保证项目不偏离正确的航线。

无实体实验

没有实体实验,想象是实验成功与否的唯一指标(from lostgarden)

没有实体实验,想象是实验成功与否的唯一指标(from lostgarden)

通常由于害怕实验成本高,创意人员宁可花费更多时间在构思想法和思维实验上。结果是,大家都不怎么做实际的实验了。没有任何实在的东西来批判想法,所以很难客观地运用挑选标准。

解决方案是,尽快原型化你的想法。纸制原型、真人角色扮演等等,总之设法在现实世界中模拟你的游戏。比较恼人的一点是,人们的意识性认知总是合理化身体与世界的互动行为。如果能在可摸可触的系统中直接实验,就可以有效地避免潜意识的干扰。

智慧的唯一来源就是对现实世界的无数次考察,而不是单纯地在头脑中瞎想。只有进行现实实验,你才能看到玩家受阻时的傻笑,感受到“继续点击”这种对话框的粘性,并且在短时间内发现问题所在。

最近,我写了个叫《Panda Poet》的简单的文字游戏。在纸上演示时,这个游戏确实是一目了然。游戏的规则很清楚,所以我可以轻易地在头脑中想象出游戏的乐趣和结果。我要做的就是马上执行项目。

所以我们运行了第一个原型。结果是,这个游戏一点也不好玩。游戏界面基本上是崩坏的;反馈系统也不起作用。玩实体原型的头10分钟,我就发现了大量问题,而在纸上模拟,我能发现的问题只有十分之一吧。《Panda Poet》基本是失败之作,但通过数个实体模型,我们也知道怎么解决问题了——《Panda Poet》大约会在年底问世。

想法越多越好

你觉得做事必须有始有终(from lostgarden)

你觉得做事必须有始有终(from lostgarden)

创建者很容易冒出越多越好的念头,但玩游戏的人可不这么想。他们玩游戏的时间和精力都有限,如果我们在游戏中加入太多额外的“填料”,他们只会觉得那是破坏游戏乐趣的噪音。

想象一下,你画了一只鸭子,然后又随意地添上斑斑点点。不管加上什么,你仍然说那是一只鸭子。但这些斑点却拖累了原来的画面。终端用户更乐意你只给他们一幅精美的图画,但上面只要一只鸭子就够了。

与此类似,如果你不能好好削减实验,你也只不过对创意画蛇添足。

害怕严格地挑选创意的人容易被已支付成本误导。他们认为既然已经为某个创意尽心竭力了,所以那份付出一定要保留下来,不然就没有价值了。他们错就错在把自己的工作看得太“金贵”了。无论什么工作,不可能只因为你努力去做了,它就必定有价值。玩家看不到你怎么拼死拼活,他们能看到的只有混乱、能听到的只有噪音——这就是不对实验精挑细选的结果。

习惯于“扼杀”想法: David McClure曾说过一句伟大的话:“每周消灭一个创意。”带着理智和原则做选择是一门技术,需要不断地练习,这和坚持健身运动是一个道理——熟能生巧。每个创意实验都只是大过程中的一块垫脚石,你是踩着它们走向成功的,而不是小心翼翼地捧着它们走向成功。

客观:以客观的数据为中心的标准。如果你能把结论归纳成数字和指标,你就不会做出情绪化的决断。这项工作可能得费一番工夫,但如果团队成员对某个论题颇有异议,标准的客观性可以起到引导作用。再者,通过现实实验,数据化标准其实并不难。人们如何回应艺术,存在可测定的客观标准。理解这个真相可以促进决断。

怎么做才是“客观的标准”呢?技巧之一是“评分”。我在许多游戏中都放入调查,即让玩家给游戏评分(1到5分)。一个玩家说这个游戏很差劲,可能纯粹是主观的污蔑。但是,如果有一万个人说你的游戏评分为3.1,那就是另一回事了。从历史数据上看,这个评分意味着,你的游戏评分得达到3.9才能赢利或赢口碑。

强求圆满

力求解决所有问题(from lostgarden)

力求解决所有问题(from lostgarden)

人们总是试图解决失败实验中出现的所有问题。这个现象在模块化的项目中尤其常见。创建者罗列了所有问题的特征,然后逐个击破,为的是让各个模块的质量都达到某个水平,从而让整个项目质量达到某个水平。

这种行为的后果惨不忍睹。

强求项目的各个方面都达到圆满,相当于同时追求多个目标,必然会拖累项目的进度。

可惜放缓进度的同时也降低了质量。一棵树的所有枝干尚且不可能长得一般粗,更何况是让一个项目的各个环节都与核心环节平齐呢?同时在项目的不同方向上使力,就像艺术家在作品的各个方面都精雕细刻,结果当然是整体四分五裂、部分各不相干。

强调问题,忽略重点

游戏开发的经验法则是,任何一个实验性功能都需要投入30倍的精力才能使其完善。所以一天完成的原型要用一个月的时间完善。如果你试图把各个方面都磨光,想象一下把钻石的各个面都磨得一样光亮,钻石的表面必然要损失好大一片。

在一款失败的在线游戏上,我就吃过这种教训。这个项目有五个负责人,五个人都以为自己的工作就是指出游戏的不足,然后命令团队解决所有问题。团队只好一个接一个地补缺补漏。这样,这款游戏终于变成一个支离破碎的半成品了。事后看来,如果我们有一套更明确挑选标准,这个项目的整体质量本可以达到更高的水准。

解决方案是,抛开细枝末节,直奔核心。如果你产生价值,你就赢了。如果你做出来的东西润饰不足,但确实有趣,人们也会宽容某些方面的粗糙。

在《Steambirds》中,关卡、任务文本、图形等方面存在一些问题。这些问题差点就浪费我们12个月的努力了。

然而,我们及时改善移动和进攻的核心机制。我们的所有实验,都非常扎实有趣,能给玩家带来极大的快感。最终,其他问题根本就不是问题。正是因为我的抓得了核心重点、抛开细枝末节,所以数月的努力总算没有付诸东流。

循环

头脑风暴和挑选是一个反复的循环过程。这个过程就好像挑苹果,你先剔出烂的,留下不烂的;然后再从不烂的当中挑出新鲜的;再然后从新鲜的当中挑出更新鲜的;直到你最终挑出色香味俱佳的那个。

循环(from lostgarden)

循环(from lostgarden)

各个循环就是劳动价值的积累过程,当达到某个足够的价值量时,你就可以停下了。

循环的问题

循环的问题主要有以下两方面:

·无循环

·没有增加额外价值

无循环

开始-结束(from lostgarden)

开始-结束(from lostgarden)

有些创建者通常只进行一轮集体讨论和挑选。在他们看来,计划就等于集体讨论和初期执行;挑选无异于缩小范围和漏洞修复。经过一轮只有开始到结束的漫长开发期,产品就出炉了。

解决方案是,缩减循环时间。所有机敏的方法论都会要求你缩减循环时间。为了增加循环次数,我提出一个问题:我怎么缩减循环时间,以便在项目中投入更多循环?无论是美术、游戏、UI设计还是程序开发,我都对自己提出相同的要求。结果是,循环越多,结果越令我满意。

没有增加额外价值

开始-重复循环(from lostgarden)

开始-重复循环(from lostgarden)

没有足够的循环的对立面正是无尽地重复,但从未释放额外价值。大家的普遍观念是,没有最好,只有更好。所以在创意的过程中,你很可能陷入死循环的陷阱中,却没有意识到你的产品已经没法更好了。

在此我推荐两种解决方案:

时间表:限定一个发行日期,无论产品如何,该出手时就出手。皮克斯公司的Darla Anderson 曾引言道:“我们没有完成电影,我们只是发布电影。”业已决定的发行日期发挥了强制性作用,确保制作人员在交出成果前尽可能完善作品。另外,制作人员由于时间表的约束,不致于在问题多多的实验上无尽地浪费时间。

发行标准:制定产品可发行的客观目标。对于Flash游戏,我知道评分达到3.9就是发行游戏的时候了。如果我愿意,我也可以再投入一点时间和精力,但价值不大。

结论

所有的思考和插图都可以归纳为非常简短的列表:

集体讨论阶段:制作大量低成本、真实的实验。

挑选阶段:严格地执行商定的挑选标准,以便去其糟粕,取其精华。

循环阶段:重复过程直到得到有意义的价值。

练习:通过在多个项目、项目的多个阶段中不断练习,可以提高头脑风暴、挑选和循环的技术。

既然你看得见,你就直接参考下图吧:

实验-挑选-循环(from lostgarden)

实验-挑选-循环(from lostgarden)

本文为游戏邦/gamerboom.com编译,如需转载请联系:游戏邦

Visualizing the Creative Process

As I coach new developers, I’ve taken to scribbling out the same useful diagram for visualizing the creative process again and again on coffee-ringed napkins.  In order to limit my future abuse of culinary paper wares, I’ve reproduced my images in a more formal fashion in this essay.

The conversation usually starts with the following statement: “Creativity is like a snake swallowing a series of tennis balls.”

And when confused looks inevitably result, I sketch some variant of this odd little picture:

Using this as a starting point, we start chatting about joys and pitfalls of creativity.

The Brainstorming Phase

Failures in brainstorming

The Culling Phase

Failures in culling

Cycling

Failures in cycling

The Brainstorming Phase

We all start with an idea.  It could be a small inspiration or a large insight.  Immediately, you begin a process of brainstorming and daydreaming.  This is a time of infinite possibility and promise.  I use the term brainstorming broadly to include any activities that expand the options or possibilities of a project.  The traditional image of a group of designer types sitting in a room with a whiteboard is indeed part of the brainstorming phase, but it is really only one of a much broader spectrum of activities.

Brainstorming starts out small and expands over time

There are several activities that occur during this phase:

Ideas: Generate new ideas related to your initial insight. These are proto-experiments. A half-jotted note in a notebook “Solar powered underpants!” is an example of an idea. (This is a real idea.  It has been hot in Seattle lately)

Thought experiments: Invest energy in your ideas to understand how they might be built and thinking through the theoretical impact of each idea.  A spec concerning “A Method for inserting a small fan in one’s shorts to reduce ambient temperature” is an example of a thought experiment.

Real world experiments: Build working models from physical materials or usable code so you can experience your idea first hand. A working model of underpants plus a small metal fan (plus a ready supply of bandages) is an example of a real world experiment.

Cross fertilization: As you work through ideas, you see new possibilities.  I’ve discovered through much trial and error that escaping to a coffee shop with air conditioning is immensely more effective than other attempted alternatives.

A multitude of experiments arise during brainstorming

Brainstorming is ultimately the act of kick starting experiments. Even when you dream up a completely off-the-wall idea, you are stating that “There is some potential in this direction.” You’ve formed a postulate at the fuzziest possible level.  Future steps during the brainstorming  involve making more detailed predictions and modeling the results.

When brainstorming is successful, we end up with a portfolio of experiments These go by a variety of different names in software development:  features, user stories, use cases and ‘the thing Bob made while screwing around on the weekend.’

Problems with brainstorming

The single most common flaw during the brainstorming period is that creators do not build enough testable experiments.  This mistake comes in a variety of flavors.

The creator already knows what needs to be done

Experimentation is considered expensive

The original idea is brittle

The creator already knows what needs to be done

Creator pursues a single solution

Why waste time on expensive experiments when the right answer is obvious? The flaw in this thinking is that creativity is an iterative process in which you synthesize the final result from a variety of sources and thousands of potential solutions. It is not purely a deductive process with a single right answer.

When you fail to experiment broadly, you are building your solution from an anemic set of mental and technical resources. It is the equivalent of trying to design a bridge when the only material you’ve tested is paper.  You can certainly build a bridge, but it will not be nearly as good compared to someone who experimented with a broad range of materials and construction techniques including steel or concrete.

To understand the power of a portfolio of experiments, consider some simple statistics.  If 4 out of 5 experimental systems are bound for failure and you create only one experiment, you have a 20% chance of overall success.  On the other hand if you create 10 experiments, you have a 89% chance of finding a success.  In practice, your chance of success is even higher since ideas cross pollinate.  By learning and adapting to your new knowledge you’ll uncover new options that are often far superior than the original set of experiments.

(Note: If the goal of experimentation is hands on knowledge, try including a wide range of participants that can bring a variety of pertinent skills to the table.)
Experimentation is considered expensive

Each individual experiment is expensive

In the example above, a savvy counter of beans might note that 10 experiments is likely to cost 10 times as much as a single experiment.  Yet all this extra money only increases the additive chance of success by a mere 450%.  So the team compromises and invests in a handful of expensive experiments.

The solution here is to use less expensive experiments, not fewer experiments. What can you make in a day? What can you make in an hour? Instead of using teams of 5 to 10, what can you learn with a team of 1or 2?  By focusing on lightweight experimentation and rapid turnover between experiments you can pack more experiments into your brainstorming phase.

One technique I love that keeps experiments small is Post-it note design docs.  Since your experiments must fit on a sticky note, you are forced to keep the scope small and easily implementable.
The original idea is brittle

No matter how hard you try, you only can come up with a handful of ideas

Sometimes you start with an idea that seems brilliant, but as you try to expand upon it, you keep running into walls.  You know you’ve found a brittle idea when you try to bend the concept in interesting new directions and it collapses again and again.

Over time, you get learn to recognize brilliant yet brittle ideas early in the creative process. My test is to try to think up 20 or 30 crazy variations on the idea. If I can imagine that most of those variations would be exciting to build, then I know I’ve discovered a robust idea that is worth investing in further. If I can’t, then I have a brittle idea.

It is too easy to invest months, even years of your life trying to “make it work.” Instead there are a couple techniques for making the idea more robust:

Put the idea aside: The single best thing you can do is to put the brittle idea on the back-burner.  Over time, if the idea is in fact brilliant, it will find its way back into your creative process.  A different perspective, be it brought on by time or new experiences, can be an essential ingredient in softening the idea’s previous constraints.

I wrote up an idea for a game called Cute God a while back, but none of the prototypes really gelled.  It was a design with a thousand problems and very few good solutions. Instead of belaboring the point, I consciously stopped redesigning it.  Years later, some of the original combinatorics ideas found their way into a game called Triple Town that should be released later this year.  Ideas you cull are never really erased.  Instead they turn into fertile soil from which the next generation of ideas are grown.

Radical simplification: What is core experience that you are trying to achieve?   Write a single sentence or draw a picture that represents that experience.  Put that on the wall in front of you. Now brainstorm a dozen different incredibly simple ways of creating the essence of that experience for the user.  Toss out all your complex systems and constraints and start over with just the core.

Recently a friend and I went through this exercise on a game concept that was born from a pencil drawing of dozens of stick figures attacking one another.  The brilliant, yet brittle design was a Toribash-style fighting game where you could move individual joints for each of the stick figures in an epic multiplayer battle.  As an exercise, we went back to the original drawing and asked “What was the simplest way to get that image up on the player’s screen” and “How do we evoke the coolness of a dozen stick figures blasting one another with shotguns.” Out goes the rag doll physics.  Out goes the complex UI.  Out goes the multiplayer.  The resultant idea was robust, easier to implement and still captured the emotional joy of the original inspiration.  At the very least, this exercise helped the designer look at the problem in a new light and question their original (brittle) constraints.

This process can all be summed up “If your design is difficult, cheat.”

Culling Phase

There is an entire second phase of of the creative process called culling.  Inevitably, not all experiments are good experiments.  Some show immediate value and others are plagued by obvious flaws.
During the Culling Phase, you need to kill flawed experiments so you can double down on good ones.

Culling boils down all your experiments to a refined nugget of value

No one really talks much about this unpleasant part of creativity.  We lionize the eureka moments of brainstorming and end up ignoring the agonizing process of trimming and shaping those meandering experiments into something coherent and valuable.  This is a huge mistake.

Critically culling your experiments is an essential step to any successful creative act.  Culling focuses the creative act, ensures projects are finish-able and ultimately yields a more  powerful final experience.  When I start painting, I place down a thousand brush strokes.  But only one stroke appears on top and is visible to the viewer.  Each previous stroke is an experiment that leads me towards that final visible stroke. Once I’ve learned enough from my experiments, I make an informed decision, place the optimal mark and move on.

Those who fail to cull, fail to create meaningful projects.  You can spot a newbie designer from a thousand yards by suggesting they kill a feature that doesn’t seem to be contributing much.  Their nostrils flair and their voice rises. A litany of denials, excuses and accusations pour forth.  And you know immediately that their project is going to be an incoherent piece of crap.  This is a good coaching moment.

Good experiments get more love and the questionable ones get trimmed

Culling is composed of several activities

Determining your culling criteria:  You need to know how you are culling up front so you don’t end up making arbitrary decisions. The single best way I’ve found of defining culling criteria is to write down what success looks like. For example, in Bunni 2, we say that the game is a ‘social stickerbook’.  Anything that doesn’t fit that vision is worth questioning.

Deciding what to invest in further:  Look for opportunities to amplify obvious value in your existing experiments.   For example, in Half Life 2, one experiment was this risky concept called a ‘gravity gun’.  When a real world prototype was made, players loved it. Valved decided to invest further and tried to figure out how that gun could be used throughout the entire game.

Deciding what to remove.  If something doesn’t fit your culling criteria, it is better to cut it early and spend those resources elsewhere.  In Bunni 2, we dabbled with a combat system.  However, after we built a simple version, we realized that it really didn’t fit our culling criteria so we put it on the backburner.

Problems with Culling

There are several common issues that come up during the culling phase

No explicit culling criteria

Experiments are not tangible.

Assuming more is better

Fixing every problem

Focusing on problems not opportunities

Judging features not core experiences.

No culling criteria

Often people get so caught up in the amazing optimism of brainstorming, they fail to agree upon explicit culling criteria.  Instead the particular politics or opinions of the day hold sway and features are randomly invested in.  Culling does occur, but in a haphazard fashion that is just as likely to kill good features as bad.

When you leave your culling criteria vague, you are saying that it is okay for everyone to have slightly different opinions about what is good and what is out. This leads to unnecessary conflicts.  You need to bite the bullet and have the hard conversation about what your shared vision for the project should be.  I like the term ‘culling criteria’ since it ensures buy off from everyone that, yes, you will mothball experiments for the greater good of the project.  The act of explicitly stating a small set of common goals ensures that everyone buys into something bigger than their individual efforts and passions.

I put ‘goals’ at the top of almost every single design document I write.  Though this is the shortest part of the design, it is often the section most critical to success.  Experiments will blossom in dozens of directions, but the goals keep the project on track.

The experiments are not tangible

Mere opinion is the only indicator if an experiment is good or bad

Often due to fear of creating expensive experiments, creative folks spend far too much time in the land of ideas and thought experiments.  Some call this documentation and invest in it like some religious protection from mistakes.  As a result, very few real world experiments are built.  With nothing concrete to react to and judge in a critical fashion, it becomes difficult to apply the culling criteria in an objective fashion.

The solution is to make your ideas real as quickly as possible. Paper prototypes, 24-hour game jams, role-playing with a friend…create your idea in the physical world. To get a bit geeky, a vast portion of conscious cognition emerges as a post-processed rationalization of our body’s physical interactions with the world. Feed your subconscious cognition by creating systems you can touch, see and play with directly.

Instead of merely wearing down a single golden path in your mental thought experiments, you’ll accumulate the wisdom that only comes from a thousand real world observations.  You’ll see players smile when they stumble.  You sense the stickiness of a dialog that asks you to ‘click continue’.  You’ll give yourself the richest possible source of information about the problem space in the shortest amount of time.

Recently I wrote out a design for a simple word game called Panda Poet.  It was completely obvious how the game played on paper.  The rules were crisply defined and I could easily play through the experience in my head and imagine the delight that would result.  All it really needed was a quick implementation and the project would be done.

So we implemented the first prototype. The result was completely unplayable. The interface was fundamentally broken.  The feedback loops were not functional.  In the first 10 minutes of playing a physical prototype, I learned 10 times as much about the problem space than I had learned in the previous days spent designing on paper.  Panda Poet was almost a failure, but by building multiple real world prototypes, we learned enough to salvage the design.  It should be out later this year.

Assuming more is better

If you started something, you feel obligated to release it.

As a creator it is easy to get caught in the belief that more is more.  Yet, the people consuming our creations rarely feel this.  They have limited time and limited mental resources to spend on our work.  Instead, they see all the extra ‘stuff’ as mental noise that actively harms their enjoyment of the project.

Imagine that you have a painting of a duck and you start randomly adding static in the form of blobs from other pictures to it.  You can add a lot of static and still tell it is a duck. But the static detracts from the image.  The end user would likely be much happier if you just gave them a great picture of a duck.

The same thing occurs when you fail to kill experiments. You are actively adding low quality noise to your creation.

Creators that fear rigorously editing their creations suffer from the sunk cost fallacy.  They assume that since they spent effort making something, value will be lost if it is removed.  Often they consider their work ‘precious’. However, there is no inherent utilitarian value in a feature simply because it took a lot of effort to build.  The user doesn’t see your effort.  They only see the messy and imperfect noise that come from not rigorously culling your flawed experiments.

Practice killing your ideas:  David McClure has a great saying, “Kill a feature every week”.  Culling with wisdom and discipline is a skill worth training much like any healthy habit.  You practice and it slowly and steadily gets easier.   You begin to see each feature and experiment as a small step in a much larger process, not a rare and precious thing that must be protected.

Be objective: Another technique is to use objective, data-centric criteria.  By boiling down decisions to numbers and metrics, you give yourself permission to make emotionally difficult decisions.  This can be taken too far, but is particularly useful if you have a group of people that have divergent subjective opinions on a topic.  Again, real world experiments facilitate this approach.  There is an objective measurable reality to how people react to art.    Understand this fact and used it to facilitate your culling.

One implementation of this technique is the use of Fun ratings.  I have a survey built into many of the games I work on that asks users how fun the game is on a scale of 1 to 5.  It is one thing to tell someone that their game sucks.  That statement comes across as a purely subjective and potentially insulting opinion.  It is another thing to have 10,000 people say that your game rates 3.1 out of 5 and to know from historical data that you need to reach 3.9 in order to have a chance at financial or critical success.

Fixing every problem

Every problem with every feature must be solved

A related issue is that creators often attempt to fixed all the problems with their failed experiments.  This is particularly common on projects that are thought of a series of modular features.  The creator lists out the problems with each feature and then methodically fixes each in an attempt to bring the feature and therefore the project up to a reasonable level of quality.

The impact of this technique is painful to witness.

The expense of ‘completing’ the projects blossoms and progress across the board slows to a crawl as multiple objectives are pursued simultaneously.  Adding more resources often only slows down the project more (ala the Mythical Man Month.)

Quality still decreases.  It is rare that each feature is completely aligned to the central value of the project.  You end up with a project that is being pulled in a variety of directions all at once.  The result is like student art where a new artist meticulously polishes every single shape in the drawing, but the end result is a hideously disjointed experience.

A good rule of thumb for game development is that every experimental feature you start takes 30 times as much effort to finish.  So one day prototyping yields one month polishing.  If you try to polish everything, the surface area of your project grows huge in very short order.

I once worked on a failed online game that suffered from this issue.  There were about 5 different project owners, each of which assumed that their job was to point out the flaws in the game and mandate (on threat of death) that the team fix them all.  The team would scramble to plug holes one after another.  In very short order, the game turned into an incoherent mass of half polished features.  In hindsight, a cleaner set of culling criteria that resulted in killing broken features would have resulted in a more focused project of higher overall quality.

The solution is to focus less on the problems and more on the opportunities. You win when you generate value. If you release an unpolished version of a something genuinely interesting and wonderful, it is amazing what people will forgive.

In Steambirds, the levels were tossed together with a semi-random assortment of planes.  The writing on the mission text was highly questionable.  The graphics were one iteration past the initial prototype art.  There was an invisible wall that caused players to die inexplicably.  It could have easily turned into a 12-month project.

However, what we did right was polish the heck out of the core mechanic of movement and attacking.  Of all our experiments, it was a robust and interesting gem that resulted in a powerful user experience.  In the end, by focusing on the heart of the game, none of the other issues really mattered.  We saved months of labor by focusing on what worked and killing or ignoring what didn’t.

Cycling

Brainstorming and culling occurs in an iterative cycle.  In each cycle, you create experiments and then cull back to the valuable core.  Then you repeat.  Each complete cycle spawns a new spurt of ideas and experiments that must be culled in turn.

Each cycle results in accumulating more value for the customers of your labor.  When you’ve generated enough value, you stop.

Cycling problems

There are a couple issues that occur during this process.

Not iterating.

Not delivering value to the customer.

Not iterating

Often creators stop after a single cycle of brainstorming and culling.  I see this quite often in groups that come from a waterfall-centric culture.  Planning is treated as the equivalent of brainstorming and initial implementation.  Culling is treated as scope reduction and bug fixing.  After one long cycle, the product releases.

The solution here is shorter, lower cost cycles.   Any of the various agile methodologies cover this ground extensively.  To facilitate more cycles, I ask the question: How do I decrease cycle time so I can fit more learning cycles into my project? I’ve asked this question for art, for games, for UI design and for application development.  In all situations, the more cycles I can pack in, the happier I am with the end result.

Not delivering value to the customer

The exact opposite of not enough iteration is to continue to iterate indefinitely and never release the value you have accumulated to a wider audience.  There is always room for improvement and always new value to generate.  You can get stuck in the trap of constantly cycling through the creative process and never feeling that your product is good enough to release.

Here are two good solutions I recommend

Timeboxing: Set a release date and release regardless of how far you’ve gotten. Pixar’s Darla Anderson has a great quote that “we don’t finish our films. We just release them” The predetermined release date is a forcing function that ensure they pack in as much value as possible before they are forced to put something out.  It also ensure that you stop working on a flawed experiment.  The emotional distance that comes from releasing can be extremely helpful in realizing that you need to take a break from a particular great white whale that is eating your life.

Release criteria: Another alternative is to set objective goals that trigger a release.   For Flash games, I know that when I’ve hit a 3.9 fun rating, I can release the game.  Certainly, I could invest further, but it really isn’t worth it.

Conclusion

All these thoughts and pictures can be summarized in a very short list.

Brainstorm: Create lots of low cost, real word experiments.

Cull: Rigorously apply agreed upon culling criteria to weed out the weak ideas and reinvest in your most promising experiments.

Cycle: Repeat the process until you generate meaningful value.

Practice: Across multiple projects, practice all stages of the creative process so you constantly improve the myriad of skills involved in brainstorming, culling and cycling.

Or if you are a visual learner, just reference this picture(source:lostgarden


上一篇:

下一篇: