万事屋

Dennett走了，Dennett来了

Thu, 30 May 2024 00:00:00 +0000

“国际米原器有多长？”

“一米，不然呢？”

“是也不是。”

“哈？”

“真空中的光速是多少？”

“299,792,458米每秒。”

“你想过一个问题吗，这个数的有效数字有几位？光那么快，光速怎么居然连小数点都没有？”

“唔。。”

“这是因为一米被定义为光在真空中于1/299792458秒内行进的距离，所以光速就是299,792,458米每秒，不多也不少。”

“原来如此，但这跟国际米原器有什么关系？”

“在采用光速定义之前，米曾经是用国际米原器来定义的。”

“这我知道。”

“那国际米原器有多长？”

“一米？不然呢？”

“是也不是。”

“哈？”

“1889年第一届国际计量大会上，人们确实制作了国际米原器并用它来定义米，定义的方法和你想的差不多——0°C时该合金米原器的长度。但在1927年第七届大会上，人们对这个定义做了一些修改，这个修改后的定义一直保持到了米原器定义被废除。这个定义是这样的：‘The unit of length is the meter, defined by the distance, at 0°, between the axes of the two central lines marked on the bar of platinum-iridium kept at the BIPM, and declared Prototype of the meter by the 1st CGPM, this bar being subject to standard atmospheric pressure and supported on two cylinders of at least one centimeter diameter, symmetrically placed in the same horizontal plane at a distance of 571 mm from each other.’ 简单来说，当你在零度和标准大气压下，把国际米原器用两个相距571毫米、直径超过一厘米的圆柱体对称地支撑起来，这时候米原器两端的距离就是一米。”

“温度和压力我理解，俩圆柱体相距571毫米是什么鬼？？”

“这叫艾里点（Airy points），目前还没有中文维基页面哦。不要忘了，米原器是有维度和重量的，假如你用两个东西随便撑着它，它会形变，结果就是两端的截面不再平行，这时候如何测量两个端面间的距离就有歧义了。但假如你在艾里点进行支撑，~~不难证明~~，两个端面依然会是平行且垂直地面的。换句话说，假如你把米原器直接放在桌子上，那它的长度未必就是一米。”

在艾里点支撑的杆虽然弯曲，但端面依然垂直地面

“原来如此，那为什么不直接把米原器做成放桌子上时是一米？”

“虽然没有明说，但第一届大会的定义大概就是得放在平面上用，但那需要一个在一米范围内‘足够’平的桌子，我猜这在制造、操作、定义、交流上都带来了一些困难，所以他们修改了定义。一个只需要在两个支撑点足够平的桌子应该还是容易制造一些。”

“好吧。等等，维基说艾里点的距离是1/√3。那两个圆柱体应该相距577毫米呀，为什么是571？”

“唔好问题。。我也不知道。。当年大会的工作报告是法文的，我浏览了一下但并没有找到57x之类的东西。GPT跟我说这是计量学家们基于经验和微调得到的，似乎可信，不过我没搜到任何来源。如果你找到了什么资料，欢迎告诉我。我和Gemini 3讨论了一下，发现我刚才说的不准确。艾里点有两种，一种是我刚才说的保持端面平行，比较适用于端点尺（End Standard）——如量块；还有一种是可以让尺的投影缩短量最小，这对线纹尺（Line Standard）比较有意义。严格来说这叫贝塞尔点。但在非专业的语境下，人们有时习惯把“为了减小重力误差而精心计算的支撑点”统称为艾里点，因为他是这方面的先驱。米原器是线纹尺，加上没刻度的两端全长102厘米，乘以贝塞尔点的位置0.5594，就是571了。”

以贝塞尔点支撑的梁拥有最大的投影长度

“挺有意思的，但为什么突然聊这个？”

“这是我最近看的一本书学到的，觉得冷僻又有趣，就分享一下。”

“《机械精度基础》，1977年出版翻译的，啥书啊这是，你咋找着的？”

“从一篇哲学论文里，它引用了书里的图片来展示测量精度是如何通过自举/互举持续提高的。”

“哲学论文引用了一本机械精度的书？这作者应该很有趣。”

“没错，他叫丹内特，Daniel Dennett，是我的哲学偶像，我觉得他足以比肩休谟。”

“为什么是休谟？提到哲学家大家想到的不都是什么尼采、黑格尔、康德吗？”

“那是外行，你用英文多搜搜就知道在什么都能分好几派的哲学圈里对休谟的推崇有多么普遍了。”

“好吧，那跟我说说丹内特，你为什么喜欢他？”

“原因很多，他整个人我都很喜欢。首先当然是学识，广博深湛，你从他引用的文献就能看出来，但他的性格也有一些我特别喜欢的地方。一是他对待论敌的态度。学术争论我见多了，他的辩论态度是我见过最好的。他也专门谈过如何有建设性地批评，他称之为拉波波特法则:

你应该非常清楚、生动、不偏不倚地重述对手的想法，使得你的对手说：‘谢谢你，我刚才要是像你这么表述就好了。’

你应该把对方观点中你所同意的部分都列出来，尤其是那种并非被人们广泛接受了的观点。

你应该提到那些从你对手那里学到的东西。

只有完成了以上三点，你才能说一句反驳或批评的话。

他自己也真是这么做的，他对各种对手理论——有些在我看来相当愚蠢——的细节掌握常常到了一种让我心疼的程度。他对对手学说的重述甚至得到过对手的肯定，只要你跟别人进行过争论你一定知道这多么难能可贵。

我个人印象最深的例子是关于古尔德的。古尔德是哈佛大学的生物学教授，写了很多文笔优美、非常流行的进化论‘科普’读物，但他对进化论的表述常常含混不清并且似乎时不时在暗示进化论过时了。这导致内行对他的态度非常复杂，有点 ‘our son of a bitch’ 的味道，比如 Maynard Smith 说：

古尔德的位置相当奇怪，尤其是在美国这边。由于他的文章非常出色，他已被非生物学家视为杰出的进化理论家。与此相反，与我讨论过他的工作的进化生物学家则倾向于把他看作是一个思想混乱到几乎不值得理会的人，但又认为他不应该受到公开批评，因为他至少站在我们这一边反对创世论者。如果不是他给非生物学家描绘了一幅进化论现状的虚假画面，那么这一切都无关紧要。
Gould occupies a rather curious position, particularly on his side of the Atlantic. Because of the excellence of his essays, he has come to be seen by non-biologists as the preeminent evolutionary theorist. In contrast, the evolutionary biologists with whom I have discussed his work tend to see him as a man whose ideas are so confused as to be hardly worth bothering with, but as one who should not be publicly criticized because he is at least on our side against the creationists. All this would not matter, were it not that he is giving non-biologists a largely false picture of the state of evolutionary theory.

我读过很多进化大家对古尔德的批评，有 Maynard 这样嘲讽的，有道金斯在《盲眼钟表匠》里对间断平衡的评论（第九章），也有对古尔德诡辩手法和动机的诛心分析（Joseph Carroll 的 Modern Darwinism and the Pseudo-Revolutions of Stephen Jay Gould），各有各的妙处。但读完之后我都还是会忍不住想找个时间去看看古尔德的原文是什么样的，不是因为我认为他们误解了古尔德或者古尔德也许是对的，我只是觉得他们对古尔德的表述还是不够详尽，我还是不明白为什么这个搅屎棍能有那么大的影响力，直到我读了丹内特对古尔德的处理。《达尔文的危险思想》的第十章都在谈论古尔德，从对拱肩的建筑学细节反驳到对间断平衡各种可能含义的评估，全面而详尽。古尔德说对的地方，丹内特都加以承认，古尔德含混不清的地方，他都尽最大的善意去澄清和加强。从那以后，我再也不想去读古尔德的原文了，因为我知道就算我花时间读完，我对古尔德的理解大概也不会有丹内特阐述的清晰。书中丹内特提到他愿意这么认真阅读古尔德的原因也让我颇有触动：

在许多生物学家看来，古尔德似乎是在主张，间断平衡这个理论讲的是通过大突变实现的戈尔德施密特式物种形成。在他们看来，古尔德正在朝着戈尔德施密特被玷污的名誉努力挥动他那根神奇的历史学家魔杖，想让戈尔德施密特的观点重获青睐。这个传说中的古尔德，正统观念的驳斥者，严重地妨碍了真实的古尔德，最后甚至连他的同事们也按捺不住地要去读读他草草写就的文章。他们难以置信地对他大加调侃，他随后否认自己这篇文章是在给戈尔德施密特的骤变论 (saltationism) 背书——而且说他从没这么干过，大家调侃得更凶了。他们知道他说过什么。但他们真的知道吗？我必须承认，我原本以为他们知道，直到史蒂夫·古尔德坚持要我将他的各类出版物全部查看一遍，亲眼看看他的反对者们是如何硬生生地把他漫画化的。他的话触动了我；没有人比我更清楚，怀疑者们想也不想就顺手给某人的精妙看法贴标签的行为，有多令人泄气（我就是传闻中那个否认人类可以体验颜色和疼痛，并且认为恒温箱会思考的家伙——不信就去问我的批评者们）。因此我查看了一番。

我另一个喜欢丹内特的原因是他的动手能力，他不仅是个哲学家，还是个 craftsman。他有个大农场，从他的农场邻居那，他学会了‘架房子、盖屋顶、装窗户、围篱笆、犁地、砍树、杀鸡、挖蛤蚌、养猪、钓鲑鱼和跳舞’，他的邻居评价他说：‘同样一件事，你不用教他两次。’丹内特很珍视这一评价。

我很羡慕这样的动手能力，买了房子后虽然我也会漆墙弄点简单的水电，也时不时跟你做点木头手工，完工时我确实有成就感，但我对过程还并不享受，也做不了太复杂的东西，希望有一天我也能拥有更多的手艺。”

“听着真是个很有意思的人，那他主要研究什么呢？”

“很多，认知科学、进化论、模因学（memetics）……比如刚才提到的那篇论文，题目叫做《达尔文‘奇怪的推理倒置’ Darwin’s ‘strange inversion of reasoning’》，是他对进化论洞见的精彩提炼，其实也不难理解，比如你手里的手机，功能强大吧，它是怎么来的？”

“苹果生产的？”

“对，怎么生产的？富士康流水线生产的对吧。那生产线的原料是哪来的呢？最终都是机床生产出来的，工业之母嘛。我想说的是那机床是怎么来的？”

“不知道，你说。”

“是精密机床生产出来的。”

“……那精密机床是怎么来的？”

“是超精密机床生产出来的。”

“……”

“你怎么不问超精密机床是怎么来的？”

“你是不是有病。。”

“哈哈不不不，超精密机床是老师傅手刨出来的。”

“哈？”

“或者说最早的精密机床是手刨出来的。原理是这样的，虽然我们不知道怎么把平面刨的特别平，但我们知道怎么检验一个平面是否特别平，这叫做三平面法（three-flat test），就是刨三块平面，两两贴合，假如全部贴合就说明三块都是平的。你可能以为机床的面一定特别光滑，其实它是凹凸不平的，只要加工件放上去是平的就行。这里的要点是：当我们有突变——老师傅随便刨和选择——三平面法，我们就可以得到一个高精度的平面。当然老师傅并不是随机在那刨，他们有他们的技巧，这让突变更定向，但突变加选择正是进化算法的核心。

真实情况当然要更复杂，魔鬼藏在细节中。三平面法其实只适用于正方形平面，而怎么把三块巨大的铸铁平面吊起来相互贴合、怎么避免形变、怎么控制温度，都还有很多学问。但我们不需要一上来就做得很平，我们只需要一个比现有技术更平的平面就算进步，而有了这么一个更平的平面（和其他改进），我们就可以得到更好的吊臂、铸铁、温度计……，回过头来又可以让我们生产更平的平面，精度就是这么互举的。其实假如你是个特别严谨的逻辑学家，你会发现之前谈到的米的定义是不完备的：当你还不知道米是怎么定义的，你怎么判断两个圆柱体是不是相距571毫米？但对进化论者这并不是一个不可捍卫的事，尤其我们还有第一届大会的定义，最初的定义并没有圆柱体这些复杂的东西。我们可以依赖第一届大会定义的米来判断两个圆柱体是不是相距571毫米、直径是不是大于一厘米——这当然不如后来的测量精确，但它足够好，继而我们可以得到更精确的米的定义。换句话说，这个定义本身就是通过之前更粗糙的定义自举的。

“很有意思，但为什么叫达尔文奇怪的推理倒置？”

“嗯，现在回到富士康流水线，光有零件是不够的，零件并不会自己变成手机，我们还需要人，这就带来了另一个角度的追问：厂弟厂妹——或者说人，是怎么来的？一个常见的答案是上帝造的。”

“哈哈哈，我知道你要说什么了！”

“没错，正常情况下大多数人就停留在上帝这个答案上了，但有了前面的铺垫，停留在这就显得毫无道理。我们显然应该像追问机床是怎么来的一样去追问上帝是怎么来的，难道还有一个超精密上帝吗？就算有这在逻辑上真的算进展吗？显然不算。但人的复杂性比机床大得多，更需要一个解释，怎么办呢？对于前达尔文时代的思想家这是个巨大的死结，比如休谟，他认识到这世界的复杂性需要一个解释，但他也知道上帝这条路走不通，可他找不到另一条解释复杂性的路，纠结致死。直到达尔文的出现，人们第一次明白一个复杂的东西可以来源于不那么复杂的东西，复杂性的解释方向被倒置了，这个世界的复杂性不再需要丹内特所说的‘天钩（skyhook）’从天上吊着，而只需要起重机就能从平凡的泥土中慢慢搭建起来，这就是达尔文奇怪的推理倒置。”

“明白了。你之前提到丹内特还研究模因学（memetics），我知道互联网上的梗图叫做 meme，但这居然还是一门学科？”

“嗯，模因是道金斯在《自私的基因》里提出来的，指的是能复制的最小的思想单位。他一开始的提议比较保守笼统，应该预见到了将会遭到的批评。丹内特极大地发展了它，并利用他的哲学功底对术语进行了清理，澄清了许多常见的误解和非难。模因这概念可能有点抽象，但如丹内特举过的例子，贝多芬《第五交响曲》的前四个音符——当当当当，就是一个模因。首先它能在大脑间传播——刚才你的脑子里肯定响起了那个旋律，这就是模因在大脑间复制，其次这组音符不能再拆得更小了。当然我们更熟悉的模因就是单词。如丹内特所说：‘单词就是能发声的模因 Words are memes that can be pronounced’。”婴儿六岁之前，平均每天要被7个新单词模因感染。互联网上的梗图显然也是模因，所以英文里梗图就叫做 meme。当然如果只是单纯地把这些东西赋予一个新的名字并没有多大意思，模因学的期待是我们可以通过类比模因和基因来获得更多的洞见，比如丹内特举过一个模因连锁的例子，这个例子描述的并不是什么新东西，在个体视角下我们可能会用锚定效应之类的心理学名词来描述这类现象，但模因学提供了一个新的角度来思考它。这个角度能走多远、能带来多少新意、值不值得都很有争议，但丹内特是坚定的支持者。”

“好吧，你说服我了，我相信丹内特真的很厉害，但这反而让我对另一个问题更加困惑：为什么我感觉从没听说过他？”

“那是因为你这方面的信息主要来自于中文世界——现代文明的边缘地带。举个例子，前阵子 OpenAI 和马斯克撕逼，把以前的邮件都爆了出来，中文媒体兴奋地报道了一遍，但对下面这封邮件，我没看到任何人问

当 Ilya 在跟马斯克探讨开源 AGI 的利弊时，他发的那个 http://slatestarcodex.com/2015/12/17/should-ai-be-open 是啥？
为什么在这样的讨论中，他唯一的引用不是论文，不是新闻报道，不是某位 AI 大神的推特，而是一个博客？
为什么他居然还认为马斯克应该已经看过了？
从马斯克的回复来看，他确实也看过了。这个博客到底是何方神圣？

其实这个博客（SSC）的作者是个精神科医生，在理性主义社区声望极高。几年前作者跟纽约时报起了冲突还炸出一堆重量级粉丝，比如 Paul Graham。他博客的文章出现在英语世界的各种地方，甚至法院判词都引用过。所以 Ilya 和马斯克读过它并且作为讨论的起点非常合理，他们没读过才是难以想象的，但你在中文世界听说过它吗，除了我的朋友圈？

这还是在紧追不舍的 AI 领域，其他领域就更别提了。

丹内特的情况也是类似的，在国外知识分子里他大名鼎鼎，你从 DeepMind 的 Demis Hassabis 接受采访提到他时那尊敬的口吻就不难看出来。

Lex Fridman：前段时间谷歌的一个工程师认为某个语言模型是有感知的，你遇到过有感知的语言模型吗？如果一个系统出现了“感知”，你怎么理解这种情况？
Demis Hassabis：我不觉得目前世界上的任何一个 AI 系统是有意识或者有感知的，这是我每天与 AI 互动的真实感受。所谓感知，更多是我们大脑自己的投射，由于那是一个语言模型，与智慧息息相关，所以人们就很容易把系统拟人化。这也是为什么我认为图灵测试有缺陷，因为它建立于人的反应和判断上。我们应该和顶尖的哲学家谈谈意识，比如 Daniel Dennett 和 David Chalmers，以及其他对意识有深刻思考的人。目前意识还没有公认的定义，如果让我来说的话，我觉得意识的定义是，信息得到处理时带来的感觉。

丹内特去世时 Demis 也第一时间发表了悼念。在中文圈丹内特这几年也慢慢火了起来，好几本书都翻译出版了，比如《达尔文的危险思想》去年终于出来了，一本1995年的畅销书。。”

“等等，他去世了？”

“嗯，昨天。”

“难怪我觉得你昨天情绪不太好，是因为这个吗？”

“嗯。”

“那你不打算写篇文章纪念一下吗？”

“有点想，但我又觉得我分量不够。辉格应该会写的。”

“……你觉不觉得 Dennett 这名字挺好的，很典雅很平静的感觉。”

“你这么一说还真是。”

“我们不正愁孩子叫什么吗？”

“你这么一说还真是。。不过这是姓，我看看…… Google 说虽然少见，不过也可以用作名。那就它吧，我越想越觉得不错。”

“为什么越想越觉得不错？”

“第一挺独特的，我们中国家长还是希望名字有特色一点；第二是对一位我们喜爱的故人的纪念，也符合英文习惯；第三丹内特一生致力于阐述进化论哲学，我觉得以一个新生命的诞生来纪念他再合适不过了。从基因之河的角度看，一艘满载着智人基因的船于 2024/4/19 沉没了，几周后一艘几乎一样的船浮了上来。”

“怎么会几乎一样？”

“你忘了人类和黑猩猩的基因相似度都有98.8%吗？”

“哦对。”

“模因之河的角度就完全不同了，一艘满载着人类顶级思想模因的巨轮沉没了，之后一艘空空的小舢板浮了上来，除了船艏涂着一样的名字，它们目前还没太多相似之处。

不过你觉得未来会怎样？这艘小船会长得多大？他会因为我们给他的名字而更容易被丹内特留下的的思想模因感染吗（名字模因对思想模因的连锁增强）？丹内特说：‘A scholar is just a library’s way of making another library 学者就是图书馆自我复制的渠道。’你觉得他将来会去复制一座什么样的图书馆呢？”

“我不知道，不过我很期待。”

为什么金斯伯格反对罗诉韦德案

Fri, 24 Jun 2022 00:00:00 +0000

今天最高法院正式推翻了罗诉韦德案，很多人都在哀嚎从此女性不能堕胎了，都是最高法院老白男们的错，假如我们“臭名昭著”的露丝•金斯伯格大法官（Notorious RBG）还在就好了~~（或者假如她没那么恋栈，在奥巴马任内退下来而不是死在川普任上也行，当然这话不能大声说）~~。但很多当代女权主义者不了解的是，她们的偶像，伟大的女权主义大法官，金斯伯格，恰恰，是反对罗诉韦德案的……

在判决草稿泄漏后女权组织 "Ruth Sent Us (露丝派我们来的)" 在网上公布了六位支持推翻罗案的大法官的家庭住址并组织大家去抗议

1993年初，金斯伯格在纽约法学院的麦迪逊讲座做了一个演讲，从她自己几十年的法律从业经验谈了她对美国法律制度的理解，非常精彩。在演讲的第二部分她论证了为什么过于激进的判决反而会阻碍社会的进步，“经验告诉我们，教义的四肢如果成形的太快，可能会被证明是不稳定的”，在她看来，罗案就是这么一个判决。罗案“不够审慎”的判决超出了最高法院的职责，“没有邀请立法者进行对话。相反，它似乎完全是为了把球从立法者的场地上拿走”，结果反而团结了反对者，打断了原本正在自由化的堕胎法规进程，“相当长的一段时间内成功地将立法潮流转向相反的方向”。这个充满洞见的演讲给她带来了意想不到的麻烦，因为就在几周后，克林顿打算提名她当大法官了。谁能想到，这样一位履历完美，为女权事业做出巨大贡献的人，在提名大法官时遇到的最大障碍，甚至是唯一障碍，恰恰来源于女权团体呢？历史真是有趣。

不管你同不同意金斯伯格的论证，你应该都会同意她反对的理由完全不涉及女性“应不应该”有权利堕胎。当然以金斯伯格的身份，没人敢说她不支持女性堕胎。但老白男阿利托就没这种幸运了。问题是，假如你去读一下阿利托撰写的多数意见，里面提到女性不应该堕胎了吗？他说的不是“我们的决定是，将堕胎权问题交还给这些（州）立法机构，它允许（赞同和反对）堕胎的双方妇女通过影响公众舆论、游说立法者、投票和竞选公职，来寻求影响立法进程。妇女并非没有选举权或政治权利。值得注意的是，登记投票和进行投票的女性比例，一直高于男性的比例。”吗？哦原来你没读过判决书啊……真不愧是你呢。

我并不是说如果金斯伯格还活着她会一块推翻罗案。从她晚年的访谈和对 Gonzales v. Carhart 的强烈反对来看，她大概会以尊重先例为由表示异议。毕竟罗案已经存在了近50年，所以阿利托在多数意见里也以极大的篇幅回应了这一观点。但不管怎样，它显然都与女性是否有权堕胎无关。

事实上罗案是一个极富教育意义的例子，有助于人们理解和思考司法分支的角色、法律的演化、规则的层次等等问题。女性是否有权堕胎是它最不相关的角度，但却是人们最热衷的。

回顾罗案这五十年，我们能学到什么呢？是保守派法院真讨厌？还是冲在社会运动最前面的史上最进步的沃伦法院其实耽误了我们五十年的时间，让一个本可以早早解决的事情延宕至今？不同人会有不同的看法，但假如你无法理解史上最杰出的女权主义大法官为什么也反对罗诉韦德案，那你的看法有价值的概率大概等于0吧？

为了让更多人能讨论到点子上，我借助 DeepL 草翻了一下金斯伯格的演讲，毕竟就像辉格说的：”宪政体制并不是文化中性的，许多守护着美国人自由的美德——比如对持枪权的热爱与执着——，在美国以外很少得到同情，世人对宪政这棵果树结出的果实大流口水，却常常对果树之根和它深植于其中的文化土壤懵然无知甚或嗤之以鼻，这样你就很难相信，他们仅仅依靠自己也同样能把果树种活养好。“

——译文的分割线——

Speaking in a Judicial Voice 以司法之声发言

Introduction 导言

The Madison Lecture series has exposed and developed two main themes: human rights and the administration of justice, particularly in our nation’s federal courts.¹ My remarks touch on both themes; I will speak first about collegiality in style, and next, about moderation in the substance of appellate decisionmaking. My views on these matters reflect experiences over a span of three decades. They have been shaped from my years as a law teacher beginning in the 1960s, through the 1970s when I helped to launch the American Civil Liberties Union’s Women’s Rights Project, and most recently during the nearly thirteen years I have had the good fortune to serve on the United States Court of Appeals for the District of Columbia Circuit. What I hope to convey about courts, I believe, is in line with the founders’—Madison’s and Hamilton’s—expectation. As a preface, I will comment on that expectation.

麦迪逊系列讲座揭示并发展了两大主题：人权和司法行政，尤其是在我们国家的联邦法院。¹这两个主题在我的发言中都将涉及；首先我将谈论合议制的风格，然后谈论上诉决策实质上的温和性。我对这些问题的看法反映了三十年来的经验。这些观点的形成源于我从20世纪60年代开始的法律教师生涯，到20世纪70年代我帮助发起美国公民自由联盟的妇女权利项目，以及最近我有幸在美国哥伦比亚特区巡回上诉法院任职的近13年期间。我相信，我希望传达的关于法院的内容与国父——麦迪逊和汉密尔顿——的期望是一致的。作为序言，我将对这种期望进行评论。

James Madison’s forecast still brightens the spirit of federal judges. In his June 1789 speech introducing to Congress the amendments that led to the Bill of Rights, Madison urged:

If [a Bill of Rights is] incorporated into the Constitution, independent tribunals of justice will consider themselves in a peculiar manner the guardians of those rights; they will be an impenetrable bulwark . . . naturally led to resist every encroachment upon rights . . . stipulated for in the Constitution by the declaration of rights.²

詹姆斯-麦迪逊的预测仍然照亮着联邦法官的精神。在1789年6月的演讲中，麦迪逊向国会介绍了导致《权利法案》的修正案，他敦促说:

“如果[权利法案]被纳入宪法，独立的司法法庭将以一种特别的方式认为自己是这些权利的监护人；他们将成为一个坚不可摧的堡垒……自然会抵制每一个侵犯权利的行为……在宪法中通过权利宣言规定的权利。”²

Today’s independent tribunals of justice are faithful to that “original understanding” when they adhere to traditional ways courts have realized the expectation Madison expressed.

今天的独立司法法庭在坚持法院实现麦迪逊所表达的期望的传统方式时，忠实于这种 “原始理解”。

In The Federalist No. 78, Alexander Hamilton said that federal judges, in order to preserve the people’s rights and privileges, must have authority to check legislation and acts of the executive for constitutionality.³ But he qualified his recognition of that awesome authority. The judiciary, Hamilton wrote, from the very nature of its functions, will always be “the least dangerous” branch of government, for judges hold neither the sword nor the purse of the community; ultimately, they must depend upon the political branches to effectuate their judgments.⁴ Mindful of that reality, the effective judge, I believe and will explain why in these remarks, strives to persuade, and not to pontificate. She speaks in “a moderate and restrained” voice,⁵ engaging in a dialogue with, not a diatribe against, coequal departments of government, state authorities, and even her own colleagues.

在《联邦党人》第78篇中，亚历山大-汉密尔顿说，为了维护人民的权利和特权，联邦法官必须有权检查立法和行政部门的行为是否符合宪法。 ³但他对这一令人敬畏的权力进行了限定。汉密尔顿写道，从其职能的本质来看，司法部门将永远是政府中 “最不危险的 “部门，因为法官既没有掌握社会的剑，也没有掌握社会的钱包；最终，他们必须依靠政治部门来实现他们的判决。⁴ 考虑到这一现实，我相信，并将在这些评论中解释为什么，有效的法官会努力说服，而不是武断地说教。她以 “温和而克制的 “声音说话，⁵ 与政府的同级部门、州政府当局，甚至她自己的同事进行对话，而不是对他们进行抨击。

I spoke of the founders’ “original understanding” a moment ago, and that expression, as I comprehend it, bears clarification in this preface. In his 1987 foreword to The Evolving Constitution, the second collection of Madison Lectures, Norman Dorsen stressed, as Chief Justice John Marshall did in 1819, that our fundamental instrument of government is an evolving document, “an instrument ‘intended to endure for ages to come.’ ”⁶ Professor Dorsen quoted Chief Justice Charles Evans Hughes’ 1934 rejection of the notion that “the great clauses of the Constitution must be confined to the interpretation which the framers, with the conditions and outlook of their time, would have placed upon them.”⁷ That understanding, as Professor Dorsen commented, has been and should remain common ground.⁸

我刚才谈到了建国者的 “原始理解”，而我所理解的这一表达方式值得在本序言中加以澄清。诺曼-多森在1987年为《麦迪逊讲座》第二辑《不断发展的宪法》撰写的前言中强调，正如首席大法官约翰-马歇尔在1819年所做的那样，我们的政府基本文书是一份不断发展的文件，”这份文书’旨在为未来的时代而存在’。”⁶ 多森教授引用了首席大法官查尔斯-埃文斯-休斯1934年反对的观点，即 “宪法的重要条款必须局限于制定者在当时的条件和前景下对其做出的解释。”⁷ 正如多森教授所评论的那样，这种理解一直是而且应该是共同的基础。⁸

In the recent decade and more of bicentennial celebrations, Supreme Court Justice Thurgood Marshall reminded us that while the Constitution’s endurance is indeed something to celebrate, the framers had a distinctly limited vision of those who counted among “We the People.”⁹ Qualified voters when the nation was new bore more than a passing resemblance to the framers: the franchise was confined to property-owning adult white males, people free from dependence on others, and therefore considered trustworthy citizens, not susceptible to influence or control by masters, overlords, or supervisors.¹⁰ In 1787, only five of the thirteen states had abolished slavery, women did not count as part of the franchise-holding, politically active community in any state, and wealth qualifications severely limited voter eligibility even among white males.¹¹ In correspondence with a friend about the qualifications for voting in his home state of Massachusetts, patriot and second president John Adams elaborated:

在最近十多年和更多的二百周年纪念活动中，最高法院法官瑟古德-马歇尔提醒我们，虽然宪法的持久性确实值得庆祝，但制宪者对那些属于 “我们人民 “的人的看法明显有限。⁹ “国家刚成立时的合格选民与制宪者比较相似：选举权仅限于拥有财产的成年白人男性，他们不依赖他人，因此被认为是值得信赖的公民，不容易受到主人、领主或监督者的影响或控制。 ¹⁰ 1787年，13个州中只有5个州废除了奴隶制，妇女在任何州都不能算作拥有选举权、也不是政治上活跃的群体，而财富要求甚至在白人男性中也严重限制了选民资格。¹¹爱国者和第二任总统约翰-亚当斯在与一位朋友的通信中就其家乡马萨诸塞州的投票资格进行了阐述：

[I]t is dangerous to open so fruitful a source of controversy and altercation as would be opened by attempting to alter the qualifications of voters; there will be no end of it. New claims will arise; women will demand a vote; lads from twelve to twenty-one will think their rights not enough attended to; and every man who has not a farthing, will demand an equal voice with any other, in all acts of state. It tends to confound and destroy all distinctions, and prostrate all ranks to one common level. ¹²

试图改变选民的资格，会引起如此多的争论和争吵，这是很危险的；这种争论和争吵不会停止的。新的要求会出现；妇女会要求有投票权；12至21岁的年轻人会认为他们的权利没有得到足够的重视；每个没有钱的人都会要求在所有国家行为中享有与其他任何人平等的发言权。它倾向于混淆和破坏所有的区别，并使所有的等级降到一个共同的水平。¹²

Our second President notwithstanding, equalizing voices and destroying rank distinctions have been dominant concerns in recent generations and, as one would expect, the focus of several Madison Lectures.’ ¹³ Although the word “equal,” or “equality,” in relation to individual rights does not even appear in the original U.S. Constitution or in the first ten amendments that compose the Bill of Rights, ¹⁴ the equal dignity of individuals ideal is part of our constitutional legacy, even of the pre-Civil War original understanding, in this vital sense. The founding fathers rebelled against the patriarchal power of kings and the idea that political authority may legitimately rest on birth status. Their culture held them back from fully perceiving or acting upon ideals of human equality and dignity. Thomas Jefferson, for example, when President, told his Secretary of the Treasury: “The appointment of a woman to [public] office is an innovation for which the public is not prepared, nor am I. ‘ ¹⁵ But the founders stated a commitment in the Declaration of Independence to equality and in the Declaration and the Bill of Rights to individual liberty. Those commitments had growth potential. As historian Richard Morris has written, a prime portion of the history of the U.S. Constitution is the story of the extension (through amendment, judicial interpretation, and practice) of constitutional rights and protections to once-excluded groups: to people who were once held in bondage, to men without property, to Native Americans, and to women. ¹⁶

尽管第二任总统如是说，但平等的声音和摧毁等级的区别一直是近几代人的主要关注点，正如人们所期望的那样，这也是几个麦迪逊讲座的重点。¹³尽管与个人权利有关的 “平等的”或 “平等 “一词甚至没有出现在美国宪法原文或构成《权利法案》的前十条修正案中，¹⁴但个人的平等尊严理想是我们宪法遗产的一部分，尤其重要的是，它甚至是内战前的原始理解。开国元勋们反抗国王的宗法权力以及政治权力可以合法地建立在出生地位上的想法。他们的文化使他们无法充分认识到人类平等和尊严的理想，也无法据此行事。例如，托马斯-杰斐逊在担任总统时，对他的财政部长说 “任命妇女担任[公共]职务是一项创新，公众没有做好准备，我也没有做好准备。”¹⁵但创始人在《独立宣言》中表明了对平等的承诺，在《宣言》和《权利法案》中表明了对个人自由的承诺。这些承诺具有增长潜力。正如历史学家理查德-莫里斯（Richard Morris）所写的那样，美国宪法历史的主要部分是（通过修正案、司法解释和实践）将宪法权利和保护扩展到曾经被排斥的群体的故事：曾经被奴役的人、没有财产的人、美洲原住民和妇女。¹⁶

I. Collegiality in Appellate Decisionmaking 上诉判决的合议性

I turn now to the first of the two topics this lecture addresses—the style of judging appropriate for appellate judges whose mission it is, in Hamilton’s words, “to secure a steady, upright, and impartial administration of the laws.”¹⁷ Integrity, knowledge, and, most essentially, judgment are the qualities Hamilton ascribed to the judiciary.¹⁸ How is that essential quality, judgment, conveyed in the opinions appellate judges write? What role should moderation, restraint, and collegiality play in the formulation of judicial decisions? As background, I will describe three distinct patterns of appellate opinion-casting: individual, institutional, and in-between.¹⁹

我现在谈谈本讲座所讨论的两个主题中的第一个——适合上诉法官的审判风格，用汉密尔顿的话说，上诉法官的使命是 “确保稳定、正直和公正的法律管理。”¹⁷ 正直、知识，以及最重要的——判断力，是汉密尔顿赋予司法机构的品质。¹⁸ 上诉法官撰写的意见中如何传达这一基本品质——判断力？节制、克制和合议制在司法判决的制定中应发挥什么作用？作为背景，我将描述上诉法院意见铸造的三种不同模式：个人、机构和介于两者之间。¹⁹

The individual judging pattern has been characteristic of the Law Lords, who serve as Great Britain’s Supreme Court. The Lords sit in panels of five and, traditionally, have delivered opinions seriatim, each panel member, in turn, announcing his individual judgment and the reasons for it.²⁰

个人审判模式一直是上议院贵族法院——英国的最高法院的特点。上议院以五人小组的形式开庭，传统上是逐一发表意见，每个小组成员轮流宣布他的个人判决和理由。²⁰

In contrast to the British tradition of opinions separately rendered by each judge as an individual, the continental or civil law traditions typified and spread abroad by France and Germany call for collective, corporate judgments. In dispositions of that genre, disagreement is not disclosed. Neither dissent nor separate concurrence is published. Cases are decided with a single, per curiam opinion generally following a uniform, anonymous style.²¹

与英国的传统，即每个法官作为个人单独提出意见相反的是以法国和德国为代表，并传播到国外的大陆法系或民法法系传统。它们要求集体的、团体的判决。在这种类型的处置中，分歧是不被披露的。既不发表异议，也不发表单独的赞同意见。案件的判决通常采用统一的匿名风格，以单一的法庭意见为准。²¹

Our Supreme Court, when John Marshall became Chief Justice, made a start in the institutional opinion direction. Marshall is credited with establishing the practice of announcing judgments in a single opinion for the Court.²² The Marshall Court, and certainly its leader, had a strong sense of institutional mission, a mission well served by unanimity. Marshall was criticized, in those early days, for suppressing dissent. Thomas Jefferson complained: “An opinion is huddled up in conclave, perhaps by a majority of one, delivered as if unanimous, and with the silent acquiescence of lazy or timid associates, by a crafty chief judge, who sophisticates the law to his own mind, by the turn of his own reasoning.”²³

当约翰-马歇尔成为首席大法官时，我们的最高法院开始向机构意见的方向上起步。马歇尔被誉为建立了以单一意见宣布法院判决的做法。²² 马歇尔法院，当然还有其领导人，具有强烈的机构使命感，这种使命感通过一致意见得到很好的实现。在早期，马歇尔曾因压制异议而受到批评。托马斯-杰斐逊抱怨说：“狡猾的首席法官把在秘密会议中也许只是以一票胜出的观点包装成一致意见，在懒惰或胆小的同事的默许下，把自己的想法和推理包装成法律”²³ 。

But even Marshall, during his long tenure as Chief Justice, ultimately dissented on several occasions and once concurred with a separate opinion.²⁴ We continue in that middle way today. Our appellate courts generally produce a judgment or opinion for the court. In that respect, we bear some resemblance to the highly institution-minded civil law judges, although our judges individually claim authorship of most of the opinions they publish. In tune with the British or common law tradition, however, we place no formal limit on the prerogative of each judge to speak out separately.

但即使是马歇尔，在他长期担任首席大法官期间，最终也多次提出异议，并有一次以单独意见表示赞同。²⁴我们今天依然采用这样的中间道路。我们的上诉法院一般都会为法院做出判决或意见。在这一点上，我们与具有高度机构意识的大陆法系法官有一些相似之处，不过我们的法官会对他们发表的大多数意见以个人名义署名。然而，为了与英国或普通法传统保持一致，我们对每个法官单独发表意见的特权没有正式限制。

To point up the difference between individual and institutional modes of judging, I have drawn upon a 1989 letter from a civilian jurist.²⁵ The letter came from a member of the Conseil d’Etat, the illustrious body created by Napoleon that still serves, among other functions, as Supreme Administrative Court for France. The conseiller who wrote to me had observed, together with several of his colleagues, an appellate argument in the District of Columbia Circuit. The appeal was from a criminal conviction; the prime issue concerned the Fifth Amendment’s double jeopardy ban.²⁶ When the case was decided, I sent our French visitors copies of the slip sheet. It contained the panel’s judgment, and three opinions, one per judge. I paraphrase the conseiller’s reaction:

为了指出个人判断模式和机构判断模式之间的区别，我参考了一位民法法系法学家1989年的一封信。²⁵这封信来自于法国最高行政法院（Conseil d’Etat）的一位成员，这个由拿破仑创建的杰出机构，除其他职能外，仍然是法国的最高行政法院。给我写信的这位委员曾与他的几位同事一起观察了哥伦比亚特区巡回法院的一次上诉辩论。该上诉是针对一项刑事定罪的；主要问题涉及第五修正案的一罪不二审²⁶ 。它包含了陪审团的判决，以及三个意见，每个法官一个。这里我转述这位委员的反应：

The way the decision is given is surprising for us according to our standards. The discussion of theory and of the meaning of precedents is remarkable. But the divided opinions seem to me very far from the way a judgment should issue, particularly in a criminal case. The judgment of a court should be precise and concise, not a discourse among professors, but the order of people charged to speak in the name of the law, and therefore written with simplicity and clarity, presenting short explanations. A judgment that is too long indicates uncertainty. At the same time, it is very impressive for me to see members of a court give to the litigants and to the readers the content of their hesitations and doubts, without diminishing the credibility of justice, in which the American is so confident.²⁷

按照我们的标准，判决的方式对我们来说是令人惊讶的。对理论和先例的意义的讨论是引人注目的。但在我看来，这些分歧意见与判决书应该发布的方式相去甚远，尤其是在刑事案件中。法院的判决应该是精确和简洁的，不是教授之间的讨论，而是受命以法律的名义发言的人的命令，因此应该写得简单明了，提出简短的解释。一份太长的判决书表明了不确定性。同时，看到法院的成员向诉讼当事人和读者介绍他们的犹豫和疑虑的内容，同时又不削弱美国人非常自信的司法的可信度，这给我留下了非常深刻的印象²⁷ 。

The conseiller seems at first distressed, even appalled, at our readiness to admit that legal judgments (including constitutional rulings) are not always clear and certain. In his second thought, however, the conseiller appears impressed, touched with envy or admiration, that our system of justice is so secure, we can tolerate open displays of disagreement among judges about what the law is. ²⁸

对于我们愿意承认法律判决（包括宪法判决）并不总是清晰和确定的，这位顾问起初似乎感到不安，甚至感到震惊。然而，在他的第二个想法中，顾问似乎对我们的司法系统如此安全，我们可以容忍法官之间对法律是什么的公开表现出的分歧印象深刻，并带有羡慕或钦佩之情。²⁸

But overindulgence in separate opinion writing may undermine both the reputation of the judiciary for judgment and the respect accorded court dispositions. Rule of law virtues of consistency, predictability, clarity, and stability may be slighted when a court routinely fails to act as a collegial body.²⁹ Dangers to the system are posed by two tendencies: too frequent resort to separate opinions and the immoderate tone of statements diverging from the position of the court’s majority.

但是，过度沉迷于单独撰写意见，可能会破坏司法机关的判决声誉和对法院判决的尊重。如果法院经常不作为一个合议机构行事，那么一致性、可预测性、明确性和稳定性等法治美德就会受到轻视。²⁹ 对该系统构成危险的是两种倾向：过于频繁地诉诸单独意见，以及在不同意多数意见时不恰当的语气。

Regarding the first danger, recall that “the Great Dissenter,” Justice Oliver Wendell Holmes, in fact dissented less often than most of his colleagues.³⁰ Chief Justice Harlan F. Stone once wrote to Karl Llewellyn (both gentlemen were public defenders of the right to dissent): “You know, if I should write in every case where I do not agree with some of the views expressed in the opinions, you and all my other friends would stop reading my separate opinions.”³¹ In matters of statutory interpretation, Justice Louis D. Brandeis repeatedly cautioned: “It is more important that the applicable rule of law be settled than that it be settled right.” “This is commonly true,” Brandeis continued, “even where the error is a matter of serious concern, provided correction can be had by legislation.”³² Revered constitutional scholar Paul A. Freund, who clerked for Justice Brandeis, recalled Justice Cardozo’s readiness to suppress his dissent in common law cases (the Supreme Court had more of those in pre-Erie³³ days), so that an opinion would come down unanimous.³⁴

关于第一种危险，请回忆一下，”伟大的异议者 “奥利弗-温德尔-霍姆斯大法官事实上比他的大多数同事更少提出异议。³⁰首席大法官哈兰-F-斯通曾经写信给卡尔-卢埃林（两位先生都是异议权的公开捍卫者）。”你知道，如果我在每个案件中都写上我不同意意见中表达的一些观点，你和我所有的其他朋友都会停止阅读我的独立意见。”³¹ 在法律解释问题上，路易斯-D-布兰代斯大法官一再告诫。”适用的法律规则被确定下来比它被正确地确定更重要”。”这通常是正确的“，布兰代斯继续说，”即使错误是一个严重的问题，只要可以通过立法来纠正就行。”³² 著名的宪法学者Paul A. Freund曾是布兰代斯大法官的书记员，他回忆说，卡多佐大法官在普通法案件（最高法院在伊利案³³之前的日子里有更多这样的案件）中随时准备压制他的反对意见，以便法院能够产生一致意见。

Separate concurrences and dissents characterize Supreme Court decisions to a much greater extent than they do court of appeals three-judge panel decisions. In the District of Columbia Circuit, for example, for the statistical year ending June 1992, the court rendered 405 judgments in cases not disposed of summarily; over 86 percent of those decisions were unanimous.³⁵ During that same period, the Supreme Court decided 114 cases with full opinions; only 21.9 percent of the decisions were unanimous.³⁶ A reality not highlighted by a press fond of separating Carter from Reagan/Bush appointees³⁷ accounts in considerable measure for this difference: the character of cases heard by courts of appeals combines with our modus operandi to tug us strongly toward the middle, toward moderation and away from notably creative or excessively rigid positions.³⁸ (The tug is not so strong, however, as to make a proposal I recently advanced acceptable. At a meeting of U.S. court of appeals judges in February 1993, I suggested that when panels are unanimous, the standard practice should be to issue the decision per curiam, without disclosing the opinion writer. That would encourage brevity, I thought, and might speed up dispositions. Few of the judges in attendance found the idea appealing.)

在最高法院的判决中，单独的同意意见和反对意见比上诉法院三法官小组的判决要多得多。以哥伦比亚特区巡回法院为例，在截至1992年6月的统计年度中，该法院对未简易处理的案件作出了405项判决；其中86%以上的判决是一致的。 ³⁵ 在同一时期，最高法院对114个案件作出了完整的意见，其中只有21.9%的判决是一致的。 ³⁶ 一个没有被那些喜欢把卡特和里根/布什任命的人分开的媒体所强调的现实³⁷在很大程度上解释了这种差异：上诉法院审理的案件的特点与我们的工作方式相结合，强烈地将我们拉向中间，走向温和，远离明显的创造性或过于僵化的立场。³⁸（不过这种拉力并没有大到让我最近的一个提议获得通过。在1993年2月的一次美国上诉法院法官会议上，我建议，当陪审团一致同意时，标准做法应该是按陪审团的意见发布判决，而不披露意见撰写人。我认为这将鼓励简明扼要，并可能加快处理速度。在场的法官中，很少有人认为这个想法有吸引力）。

Concerning the character of federal cases, unlike the Supreme Court, courts of appeals deal far less frequently with grand constitutional questions than with less cosmic questions of statutory interpretation or the rationality of agency or district court decisions. In most matters of that variety, as Justice Brandeis indicated, it is best that the matter be definitively settled,³⁹ preferably with one opinion. Furthermore, lower court judges are bound more tightly by Supreme Court precedent than is the High Court itself.

关于联邦案件的特点，与最高法院不同，上诉法院处理宏大的宪法问题的频率远远低于处理法律解释或机构或地区法院判决合理性等较小的问题。正如布兰代斯大法官所指出的，在大多数此类问题中，最好是明确地解决这个问题，³⁹最好是有一个意见。此外，下级法院的法官受最高法院判例的约束比最高法院本身更严格。

Turning to the way we operate, I note first that no three-judge panel in a circuit is at liberty to depart from the published decision of a prior panel; law of the circuit may be altered only by the court en banc.⁴⁰ To ensure that each panel knows what the others are doing, the District of Columbia Circuit, and several other federal circuit courts of appeals, circulate opinions to the full court, once approved by a panel, at least a week in advance of release.⁴¹

谈到我们的运作方式，我首先注意到，一个巡回法院的任何三位法官组成的合议庭都不能随意背离前一个合议庭公布的判决；巡回法院的法律只能由法院全院庭审来改变。⁴⁰ 为了确保每个合议庭都知道其他合议庭在做什么，哥伦比亚特区巡回法院和其他几个联邦巡回上诉法院会在合议庭同意后，至少在发布前一周向全法院分发意见。⁴¹

Second, in contrast to district judges, who are the real power holders in the federal court system—lords of their individual fiefdoms from case filing to first instance final judgment—no single court of appeals judge can carry the day in any case. To attract a second vote and establish durable law for the circuit, a judge may find it necessary to moderate her own position, sometimes to be less bold, other times to be less clear.⁴² We can listen to and persuade each other in groups of three more effectively than can a larger panel.

其次，与地区法院法官相反——地区法院法官是联邦法院系统中真正的权力持有者，从案件提交到一审最终判决，他们是辖区的主人——没有一个上诉法院的法官可以在任何案件中主导一切。为了吸引第二票并为巡回法院制定持久的法律，法官可能会发现有必要缓和自己的立场，有时不那么大胆，有时不那么明确。⁴² 比起多人合议庭，三人合议庭能更有效地互相倾听和说服。

On the few occasions each year when we sit en banc—in the District of Columbia Circuit, all twelve of us when we are full strength—I can appreciate why unanimity is so much harder to achieve in Supreme Court judgments. Not only do the Justices deal much more often with constitutional questions, where, in many cases, only overruling or constitutional amendment can correct a mistake. In addition, one becomes weary after going round the table on a first ballot. It is ever so much easier to have a conversation—and an exchange of views on opinion drafts—among three than among nine or twelve.⁴³

每年在我们全体法官开庭的少数情况下——在哥伦比亚特区巡回法院，我们全部12人都是精神饱满，我可以理解为什么在最高法院的判决中，一致意见如此难以实现。大法官们不仅更经常地处理宪法问题，在许多情况下，只有推翻或修改宪法才能纠正一个错误。此外，在第一轮投票后，人们会感到疲惫。三个人之间的对话和对意见草案的意见交流比九个或十二个人之间的对话要容易得多。⁴³

In writing for the court, one must be sensitive to the sensibilities and mind-sets of one’s colleagues, which may mean avoiding certain arguments and authorities, even certain words.⁴⁴ Should institutional concerns affect the tone of separate opinions, when a judge finds it necessary to write one?

在为法院写作时，必须对同事的情感和心态保持敏感，这可能意味着要避免某些论点和权威，甚至是某些词语。⁴⁴那么当法官认为有必要写单独意见时，对机构的关注是否应该影响其语气？

I emphasize first that dissents and separate concurrences are not consummations devoutly to be avoided. As Justice William J. Brennan said in thoughtful defense of dissents: “None of us, lawyer or layman, teacher or student, in our society must ever feel that to express a conviction, honestly and sincerely maintained, is to violate some unwritten law of manners or decorum.”⁴⁵ I question, however, resort to expressions in separate opinions that generate more heat than light. Consider this sample from an April 1991 District of Columbia Circuit decision. The dissenter led off: “Running headlong from the questions briefed and argued before us, my colleagues seek refuge in a theory as novel as it is questionable. Unsupported by precedent, undeveloped by the court, and unresponsive to the facts of this case, the . . . theory announced today has an inauspicious birth.”⁴⁶ That spicy statement, by the way, opposed an en banc opinion in which all of the judges concurred, except the lone dissenter.

我首先强调，反对意见和单独同意意见不是圆满案件所要虔诚避免的。正如威廉-J-布伦南大法官在为反对意见进行深思熟虑的辩护时所说。”在我们的社会中，无论是律师还是普通人，无论是教师还是学生，都不应该觉得表达一个诚实和真诚的信念是违反了一些不成文的礼仪或礼节的法律。”⁴⁵我质疑的是在单独意见中使用容易引发情绪而不是洞见的表达方式。考虑一下1991年4月哥伦比亚特区巡回法院判决中的这个样本。持反对意见的人开始说 “我的同事们从摆在我们面前的问题中一溜烟跑出来，在一个既新颖又有问题的理论中寻求庇护。没有先例的支持，没有法庭的发展，也没有对本案的事实作出反应，今天宣布的……理论有一个不详的诞生。”⁴⁶顺便说一下，这个辛辣的声明反对的是全员庭审的意见——即除了唯一的反对者，所有法官都同意的意见。

It is “not good for public respect for courts and law and the administration of justice,” Roscoe Pound decades ago observed, for an appellate judge to burden an opinion with “intemperate denunciation of [the writer’s] colleagues, violent invective, attributi[on]s of bad motives to the majority of the court, and insinuations of incompetence, negligence, prejudice, or obtuseness of [other judges].”⁴⁷ Yet one has only to thumb through the pages of current volumes of United States Reports and Federal Reporter Second to come upon condemnations by the score of a court or colleague’s opinion or assertion as, for example, “folly,”⁴⁸ “ludicrous,”⁴⁹ “outrageous,”⁵⁰ one that “cannot be taken seriously,”⁵¹ “inexplicable,”⁵² “the quintessence of inequity,”⁵³ a “blow against the People,”⁵⁴ “naked analytical bootstrapping,”⁵⁵ “reminiscent . . . of Sherman’s march through Georgia,”⁵⁶ and “Orwellian.”⁵⁷

几十年前，罗斯科-庞德就指出，如果一个上诉法官在其意见书中加入 “对[作者]的同事进行无礼的谴责、激烈的谩骂、将不良动机归咎于法院的大多数人，并影射[其他法官]的无能、疏忽、偏见或愚钝，这对公众的尊重是不利的。 “⁴⁷然而，人们只需翻翻目前的《美国报告》和《联邦报告》第二卷，就会发现对法院或同事的意见或主张的谴责，例如，”愚蠢”⁴⁸、”可笑”⁴⁹、”无耻”⁵⁰、”不能认真对待”⁵¹、 “无法解释”⁵²、 “不公平的典范”⁵³ 、”对人民的打击”⁵⁴、 “赤裸裸的自圆其说”、⁵⁵ “让人想起…谢尔曼在乔治亚州的行军”⁵⁶ 和 “奥威尔式”⁵⁷。

“[L]anguage impugning the motives of a colleague,” Senior Third Circuit Judge Collins J. Seitz recently commented, may give momentary satisfaction to the separate opinion writer, but “does nothing to further cordial relationships on the court.”⁵⁸ Judge Seitz counseled “waiting a day”—I would suggest even a week or two—“before deciding whether to send a biting response.”⁵⁹

第三巡回法院资深法官 Collins J. Seitz 最近评论说，”指责同事动机的愤怒可能会让独立意见撰写者感到一时的满足，但 “对促进法院的友好关系毫无帮助。”⁵⁸ Seitz 法官建议 “等待一天”——我建议甚至一周或两周——”再决定是否发出尖锐回应”。⁵⁹

The most effective dissent, I am convinced, “stand[s] on its own legal footing”;⁶⁰ it spells out differences without jeopardizing collegiality or public respect for and confidence in the judiciary. I try to write my few separate opinions each year as I once did briefs for appellees—as affirmative statements of my reasons, drafted before receiving the court’s opinion, and later adjusted, as needed, to meet the majority’s presentation. Among pathmarking models, one can look to Justice Curtis’ classic dissent in the Dred Scott case,⁶¹ and, closer to our time, separate opinions by the second Justice John Marshall Harlan.⁶²

我相信，最有效的反对意见是 “站在自己的法律立场上”；⁶⁰它阐明了分歧，但又不损害同事关系或公众对司法机构的尊重和信任。我试图像我曾经为被上诉人写辩护状那样，每年写我的几个独立意见——作为我的理由的肯定性陈述，在收到法院的意见之前起草，然后根据需要调整，以满足多数人的陈述。作为典范性的路标，我们可以参考柯蒂斯大法官在 Dred Scott 案中的经典异议，⁶¹以及更接近我们这个时代的大法官约翰-马歇尔-哈兰二世的单独意见⁶² 。

Taking a comparative sideglance, I find instructive the March 5, 1992, judgment of the Supreme Court of Ireland in the case of Attorney General v. X.⁶³ The case involved a fourteen-year-old girl who, it was alleged, had been raped by the father of a school friend and had become pregnant. She and her parents had gone to England to secure an abortion. But they promptly returned home when notified that the attorney general had obtained an order from the High Court (the court of first instance) in Ireland enjoining their travel and its purpose. At issue was a clause of the Constitution of Ireland that read: “The State acknowledges the right to life of the unborn and, with due regard to the equal right to life of the mother, guarantees in its laws to respect, and, as far as practicable, by its laws to defend and vindicate that right.”⁶⁴

从比较的角度来看，我认为爱尔兰最高法院1992年3月5日对司法部长诉X 一案的判决具有指导意义。⁶³ 该案涉及一个14岁的女孩，据称她被一个同学的父亲强奸并怀孕。她和她的父母去了英国以确保堕胎。但当他们被告知总检察长已从爱尔兰高等法院（初审法院）获得命令，禁止他们的旅行及其目的时，他们立即回国。争论的焦点是爱尔兰宪法中的一项条款，内容如下 “国家承认未出生婴儿的生命权，并在适当考虑母亲的平等生命权的情况下，在其法律中保证尊重，并在可行的情况下，通过其法律捍卫和维护这一权利”⁶⁴ 。

In fact, no implementing laws had been passed, so the courts were called upon to interpret the Constitution directly. The Supreme Court, composed of five judges, voted four to one to set aside the High Court’s February 17, 1992, injunction.⁶⁵ Each judge spoke separately, but the majority agreed that, in view of the documented “real and substantial” risk that the girl would take her own life, termination of her pregnancy was permissible, even in Ireland itself. In so ruling, the Chief Justice referred to precedent calling upon judges to bring to bear on their judgments the instruction in the Constitution’s preamble that the fundamental instrument of government was adopted by the people “to promote the common good, with due observance of prudence, justice and charity so that the dignity and freedom of the individual might be assured.”⁶⁶ Those concepts and judicial interpretations of them, the Chief Justice said, “may gradually change or develop as society changes and develops.”⁶⁷

事实上，当时还没有通过任何相关的施行法律，所以法院被要求直接解释宪法。由五名法官组成的最高法院以四比一的投票结果撤销了高等法院1992年2月17日的禁令。⁶⁵ 每位法官都分别发言，但大多数人都同意，鉴于有文件证明该女孩有 “真实和实质性 “的自杀风险，即使在爱尔兰，也应允许终止其妊娠。在这个判决中，首席大法官提到了一些先例，这些先例要求法官将宪法序言中的指示带到他们的判决中，即人民建立政府的根本原因是为了”促进共同利益，并适当遵守审慎、正义和仁慈，从而确保个人的尊严和自由。”⁶⁶ 首席大法官说，这些概念和对它们的司法解释，”可能随着社会的变化和发展逐渐改变或发展。⁶⁷

The dissenting Justice spent no energy characterizing his colleagues’ opinions as “activist” or “imperial.”⁶⁸ He simply stated affirmatively his view that the evidence did not justify overturning the injunction.⁶⁹ “Suicide threats,” he reasoned, “can be contained.”⁷⁰ “The choice,” he said, was “between the certain death of the unborn life and a feared substantial danger . . . but no degree of certainty of the mother’s death by way of self-destruction.”⁷¹ The Constitution’s “equal right” provision, he concluded, required the judiciary to prevent the certain death, not the one that could be guarded against.

反对的大法官没有花精力把他的同事的意见说成是 “司法能动主义 “或 “帝国主义”。⁶⁸ 他只是肯定地指出他的观点，即证据并不足以支持推翻禁令。⁶⁹ “自杀威胁，”他推断，”可以被控制。 “⁷⁰ “他说，”我们是在未出生胎儿的必然死亡和担心母亲会自杀的实质性危险的恐惧之间做选择……但母亲自杀的可能性是不确定的。”⁷¹ 他总结说，宪法的 “平等权利 “条款要求司法部门阻止肯定的死亡，而不是可以防范的死亡。

I did not select this example as a springboard to comparison of positions on access to abortion under constitutional prescriptions and legal regimes here and abroad.⁷² I chose Attorney General v. X only to demonstrate that even in the most emotion-laden, politically sensitive case, effective opinion writing does not require a judge to upbraid colleagues for failing to see the light or to get it right.⁷³

我选择这个例子并不是为了比较美国和外国宪法和法律制度对堕胎的立场。⁷²我选择司法部长诉X案只是为了证明，即使在最充满感情、政治敏感的案件中，有效的意见写作并不要求法官指责同事没有领悟或没有得到正确的答案⁷³ 。

Concerned about the erosion of civility in the legal profession, the Seventh Circuit, commencing in the fall of 1989, conducted a “study and investigation into litigation practices and the attending relationships among lawyers, among judges, and between lawyers and judges.”⁷⁴ The Final Report of the committee in charge of the study, released in June 1992, urges judges to set a good example by staying on the high ground. Specifically, the report calls on judges to avoid “disparaging personal remarks or criticisms, or sarcastic or demeaning comments about another judge,” and instead to “be courteous, respectful, and civil in opinions, ever mindful that a position articulated by another judge generally is the result of that judge’s earnest effort to interpret the law and the facts correctly.”⁷⁵ To that good advice, one can say “amen.”

由于担心法律界的文明程度受到侵蚀，第七巡回法院从1989年秋季开始，对 “诉讼惯例以及律师之间、法官之间、律师与法官之间的关系进行了研究和调查。”⁷⁴负责这项研究的委员会于1992年6月发表的最后报告敦促法官树立良好的榜样，站在高处。具体而言，该报告呼吁法官避免 “贬低个人言论或批评，或讽刺或贬低另一位法官”，而是 “在发表意见时要礼貌、尊重和文明，永远记住另一位法官所阐述的立场通常是该法官认真努力正确解释法律和事实的结果。”⁷⁵ 对于这个好建议，人们可以说 “阿门”。

II Measured Motions in Third Branch Decisionmaking 第三权判决中的审慎提议

Moving from the style to the substance of Third Branch decisionmaking, I will stress in the remainder of these remarks that judges play an interdependent part in our democracy. They do not alone shape legal doctrine but, as I suggested at the outset, they participate in a dialogue with other organs of government, and with the people as well.⁷⁶ “Judges do and must legislate,” Justice Holmes “recognized without hesitation,” but “they can do so,” he cautioned, “only interstitially; they are confined from molar to molecular motions.”⁷⁷ Measured motions seem to me right, in the main, for constitutional as well as common law adjudication. Doctrinal limbs too swiftly shaped, experience teaches, may prove unstable.⁷⁸ The most prominent example in recent decades is Roe v. Wade.⁷⁹ To illustrate my point, I have contrasted that breathtaking 1973 decision with the Court’s more cautious dispositions, contemporaneous with Roe, in cases involving explicitly sex-based classifications,⁸⁰ and will further develop that comparison here.

谈论完了第三权的风格，我接下来将讨论第三权的实质，在这篇评论的剩余部分，我将强调法官在我们的民主中发挥着相互依赖的作用。他们并不单独塑造法律学说，但正如我在一开始所建议的那样，他们参与了与政府其他机构以及与人民的对话。⁷⁶“法官确实而且必须立法，”霍姆斯法官 “毫不犹豫地承认”，但他告诫说：”他们只能填隙式地做；他们被限制在小的动议中。”⁷⁷在我看来，对宪法和普通法的判决应该审慎，这大致是对的。经验告诉我们，教义的四肢如果成形的太快，可能会被证明是不稳定的。⁷⁸ 近几十年来最突出的例子是罗诉韦德案。⁷⁹ 为了说明我的观点，我曾将1973年罗案那令人震惊的判决与同时期法院在涉及明确的基于性别分类的案件中更谨慎的判决进行了对比，⁸⁰我将在这里进一步探讨这一对比。

The seven-to-two judgment in Roe v. Wade⁸¹ declared “violative of the Due Process Clause of the Fourteenth Amendment” a Texas criminal abortion statute that intolerably shackled a woman’s autonomy; the Texas law “excepted from criminality only a life-saving procedure on behalf of the pregnant woman.”⁸² Suppose the Court had stopped there, rightly declaring unconstitutional the most extreme brand of law in the nation, and had not gone on, as the Court did in Roe, to fashion a regime blanketing the subject, a set of rules that displaced virtually every state law then in force.⁸³ Would there have been the twenty-year controversy we have witnessed, reflected most recently in the Supreme Court’s splintered decision in Planned Parenthood v. Casey?⁸⁴ A less encompassing Roe, one that merely struck down the extreme Texas law and went no further on that day, I believe and will summarize why, might have served to reduce rather than to fuel controversy.

在罗诉韦德案⁸¹中，最高法院以七比二的判决宣布德克萨斯州的刑事堕胎法规”违反了第十四修正案的正当程序条款”，不可容忍地束缚了妇女的自主权；德克萨斯州的法律 “仅将代表孕妇的生命拯救程序排除在犯罪之外。 “⁸² 假设法院到此为止，正确地宣布全国最极端的法律违宪，而不是像法院在罗案中那样，制定一个涵盖该主题的制度，一套取代当时生效的几乎所有州法律的规则。 ⁸³ 我们是否会有二十年的争议，就像最近在最高法院对宾州东南部计划生育组织诉凯西案的分裂判决中体现的那样？ ⁸⁴ 我认为，并将论证，如果罗案的范围较小，只是推翻了得克萨斯州的极端法律，而没有进一步拓展，可能会起到减少而不是加剧争议的作用。

In the 1992 Planned Parenthood decision, the three controlling Justices accepted as constitutional several restrictions on access to abortion that could not have survived strict adherence to Roe.⁸⁵ While those Justices did not closely consider the plight of women without means to overcome the restrictions, they added an important strand to the Court’s opinions on abortion—they acknowledged the intimate connection between a woman’s “ability to control her reproductive life” and her “ability . . . to participate equally in the economic and social life of the Nation.”⁸⁶ The idea of the woman in control of her destiny and her place in society⁸⁷ was less prominent in the Roe decision itself, which coupled with the rights of the pregnant woman the free exercise of her physician’s medical judgment.⁸⁸ The Roe decision might have been less of a storm center⁸⁹ had it both homed in more precisely on the women’s equality dimension of the issue and, correspondingly, attempted nothing more bold at that time than the mode of decisionmaking the Court employed in the 1970s gender classification cases.

在1992年的宾州东南部计划生育组织诉凯西案判决中，三位持最多数意见的大法官接受了对获得堕胎的若干限制，这些限制在严格遵守罗案的情况下是不可能存在的。⁸⁵ 虽然这些大法官没有仔细考虑没有办法克服这些限制的妇女的困境，但他们为法院关于堕胎的意见增加了一个重要部分——他们承认妇女 “控制其生殖生活的能力 “与她 “平等参与国家经济和社会生活的能力 “之间存在密切联系。 “⁸⁶ 妇女掌握自己的命运和在社会中的地位的想法⁸⁷ 在罗案判决本身中并不突出，它与孕妇自由行使医生的医疗判断的权利结合在一起。⁸⁸ 如果罗案的判决能更准确地定位在妇女平等方面，并且没有采用比法院在1970年代其他性别分类案件中更激进的决策模式，它可能就不会成为一个风暴中心⁸⁹ 。

In fact, the very Term Roe was decided, the Supreme Court had on its calendar a case that could have served as a bridge, linking reproductive choice to disadvantageous treatment of women on the basis of their sex. The case was Struck v. Secretary of Defense;⁹⁰ it involved a captain the Air Force sought to discharge in Vietnam War days. Perhaps it is indulgence in wishful thinking, but the Struck case, I believe, would have proved extraordinarily educational for the Court and had large potential for advancing public understanding. Captain Susan Struck was a career officer. According to her commanding officer, her performance as a manager and nurse was exemplary.⁹¹ Captain Struck had avoided the drugs and the alcohol that hooked many service members in the late 1960s and early 1970s,⁹² but she did become pregnant while stationed in Vietnam. She undertook to use, and in fact used, only her accumulated leave time for childbirth. She declared her intention to place, and in fact placed, her child for adoption immediately after birth. Her religious faith precluded recourse to abortion.⁹³

事实上，就在罗案判决的同一工作庭期，最高法院的日程表上有一个案件可以作为桥梁，将生殖选择与基于性别的不利于妇女的待遇联系起来。该案是斯特拉克诉国防部长案；⁹⁰该案涉及空军在在越战时期寻求逼迫一名上尉退伍。也许这只是我一厢情愿，但我相信斯特拉克案会被证明对法院具有极大的教育意义，并有可能促进公众理解。苏珊-斯特拉克上尉是一名职业军官。据她的指挥官说，她作为管理人员和护士的表现堪称典范。⁹¹ 斯特拉克上尉在1960年代末和1970年代初避免了让许多军人迷恋的毒品和酒精，⁹² 但她在驻扎在越南时确实怀孕了。她承诺使用，事实上也只使用了她累积的休假时间来生孩子。她宣布打算在孩子出生后立即将其送去收养，也确实这么做了。她的宗教信仰不允许她堕胎。⁹³

Two features of Captain Struck’s case are particularly noteworthy. First, the rule she challenged was unequivocal and typical of the time. It provided: “A woman officer will be discharged from the service with the least practicable delay when a determination is made by a medical officer that she is pregnant.”⁹⁴ To cover any oversight, the Air Force had a backup rule: “The commission of any woman officer will be terminated with the least practicable delay when it is established that she . . . has given birth to a living child while in a commissioned officer status.”⁹⁵

斯特拉克上尉的案件有两个特点特别值得注意。首先，她所质疑的规则是明确的，是当时的典型。它规定 “当医务人员确定一名女军官怀孕时，她将在最短的时间内被开除。”⁹⁴为了保证没有任何遗漏，空军还有一条备用规则：“如果确定任何女军官……在担任军官期间生了一个孩子，将在最短的时间内终止她的职务。”⁹⁵

A second striking element of Captain Struck’s case was the escape route available to her, which she chose not to take. Air Force regulations current at the start of the 1970s provided: “The Air Force Medical Service is not subject to State laws in the performance of its functions. When medically indicated or for reasons involving medical health, pregnancies may be terminated in Air Force hospitals . . . ideally before 20 weeks gestation.”⁹⁶

斯特拉克上尉案件的第二个引人注目的因素是有其他的解决方案（堕胎），但她拒绝了。20世纪70年代初的空军条例规定。”空军医务处在履行其职能时不受国家法律的约束。当有医学指征或涉及医疗健康的原因时，可以在空军医院终止妊娠……最好是在妊娠20周之前”⁹⁶。

Captain Struck argued that the unwanted discharge she faced unjustifiably restricted her personal autonomy and dignity; principally, however, she maintained that the regulation mandating her discharge violated the equal protection of the laws guarantee implicit in the Fifth Amendment’s Due Process Clause.⁹⁷ She urged that the Air Force regime differentiated invidiously by allowing males who became fathers, but not females who became mothers, to remain in service and by allowing women who had undergone abortions, but not women who delivered infants, to continue their military careers.⁹⁸ Her pleas were unsuccessful in the lower courts, but on October 24, 1972, less than three months before the Roe decision, the Supreme Court granted her petition for certiorari.⁹⁹

斯特拉克上尉辩称，她所面临的她不喜欢的解雇不合理地限制了她的个人自主权和尊严；然而，她主要坚持认为，规定她被解雇的条例违反了第五修正案的正当程序条款中隐含的法律平等保护保障。⁹⁷ 她敦促说，空军的制度允许成为父亲的男性继续服役，但不允许成为母亲的女性继续服役，允许接受堕胎的妇女继续其军事生涯，但不允许分娩的妇女继续其军事生涯，这是一种令人反感的区别对待。⁹⁸ 她的请求在下级法院没有成功，但在1972年10月24日，在罗案判决前不到三个月，最高法院决定审理她的诉讼请求。⁹⁹

At that point the Air Force decided it would rather switch than fight. At the end of November 1972, it granted Captain Struck a waiver of the once unwaivable regulation and permitted her to continue her service as an Air Force officer. The solicitor general promptly and successfully suggested that the case had become moot.¹⁰⁰

这时，空军决定打官司不如换政策。1972年11月底，空军批准斯特拉克上尉豁免曾经不可豁免的规定，允许她继续作为空军军官服役。首席检察官迅速并成功地提出，该案已变得毫无意义。¹⁰⁰

Given the parade of cases on the Court’s full calendar, it is doubtful that the Justices trained further attention on the Struck scenario. With more time and space for reflection, however, and perhaps a female presence on the Court, might the Justices have gained at least these two insights? First, if even the military, an institution not known for avant-garde policy, had taken to providing facilities for abortion, then was not a decision of Roe’s muscularity unnecessary? Second, confronted with Captain Struck’s unwanted discharge, might the Court have comprehended an argument, or at least glimpsed a reality, it later resisted—that disadvantageous treatment of a woman because of her pregnancy and reproductive choice is a paradigm case of discrimination on the basis of sex?¹⁰¹ What was the assumption underlying the differential treatment to which Captain Struck was exposed? The regulations that mandated her discharge were not even thinly disguised. They declared, effectively, that responsibility for children disabled female parents, but not male parents, for other work—not for biological reasons, but because society had ordered things that way.¹⁰²

鉴于法院日程表上的案件繁多，大法官们是否对斯特拉克案进行了进一步的关注值得怀疑。然而，如果有更多的时间和空间进行思考，以及或许还会有女性出庭，大法官们可能至少会获得以下两点启示？首先，如果连军队这个并不以前卫政策著称的机构都开始为堕胎提供设施，那么罗案的决定岂不是没有必要？第二，面对斯特拉克上尉的意外退役，法院可能已经理解了一个论点，或者至少瞥见了一个现实，它后来抵制了这个论点——因为妇女怀孕和生育选择而受到的不利待遇是基于性别的歧视的典范案例？ ¹⁰¹ 斯特拉克上尉所面临的差别待遇的基础假设是什么？规定她被开除的条例甚至没有任何掩饰的意思。他们有效地宣布，对孩子的责任使母亲不能从事其他工作，而父亲则不会——不是因为生理原因，而是因为社会已经这样安排。¹⁰²

Captain Struck had asked the Court first to apply the highest level of scrutiny to her case, to hold that the sex-based classification she encountered was a “suspect” category for legislative or administrative action.¹⁰³ As a fallback, she suggested to the Court an intermediate standard of review, one under which prescriptions that worked to women’s disadvantage would gain review of at least heightened, if not the very highest, intensity.¹⁰⁴ In the course of the 1970s, the Supreme Court explicitly acknowledged that it was indeed applying an elevated, labeled “intermediate,” level of review to classifications it recognized as sex-based.¹⁰⁵

斯特拉克上尉首先要求法院对她的案件适用最高级别的审查，认为她遇到的基于性别的分类是立法或行政行动的 “可疑 “类别。 ¹⁰³ 作为退路，她向法院提出了一个中间审查标准，根据这个标准，对妇女不利的规定即使不适用最高强度的审查，也至少是高强度的审查。¹⁰⁴ 在1970年代，最高法院明确承认，它确实对它认为是基于性别的分类适用一个高强度的、被称为 “中间 “的审查水平。¹⁰⁵

Justice O’Connor carefully traced that development in last year’s Madison Lecture,¹⁰⁶ and I will recall it only summarily. Until 1971, women did not prevail before the Supreme Court in any case charging unconstitutional sex discrimination.¹⁰⁷ In the years from 1971 to 1982, however, the Court held unconstitutional, as violative of due process or equal protection constraints, a series of state and federal laws that differentiated explicitly on the basis of sex.¹⁰⁸

奥康纳大法官在去年的麦迪逊讲座中仔细追溯了这一发展，¹⁰⁶我只想简要回顾一下。直到1971年，在指控违宪的性别歧视的任何案件中，妇女都没有在最高法院胜诉。¹⁰⁷然而，在1971年至1982年期间，法院认为一系列明确基于性别的州和联邦法律违反了正当程序或平等保护的约束，因此是违宪的。¹⁰⁸

The Court ruled in 1973, for example, that married women in the military were entitled to the housing allowance and family medical care benefits that Congress had provided solely for married men in the military.¹⁰⁹ Two years later, the Court held it unconstitutional for a state to allow a parent to stop supporting a daughter once she reached the age of 18, while requiring parental support for a son until he turned 21.¹¹⁰ In 1975, and again in 1979, the Court declared that state jury-selection systems could not exclude or exempt women as a class.¹¹¹ In decisions running from 1975 to 1980, the Court deleted the principal explicitly sex-based classifications in social insurance¹¹² and workers’ compensation schemes.¹¹³ In 1981, the Court said nevermore to a state law designating the husband “head and master” of the household.¹¹⁴ And in 1982, in an opinion by Justice O’Connor, the Court held that a state could not limit admission to a state nursing college to women only.¹¹⁵

例如，法院在1973年裁定，军队中的已婚妇女有权享受国会只为军队中的已婚男子规定的住房津贴和家庭医疗福利。¹⁰⁹两年后，法院认为，如果一个州允许父母在女儿年满18岁时停止抚养她，那要求父母抚养儿子直到他21岁就是违宪的。 ¹¹⁰ 1975年，以及1979年，法院宣布各州的陪审团选拔制度不能将妇女作为一个类别排除或豁免。¹¹¹ 在1975年至1980年的判决中，法院删除了社会保险¹¹²和工人赔偿计划¹¹³中主要的明确的基于性别的分类。1981年，最高法院不再允许州法认定丈夫为家庭 “户主和主人 “。¹¹⁴1982年，在奥康纳大法官主笔的判决书中，法院认为一个州不能将一所州立护理学院的入学名额只限于女性。¹¹⁵

The backdrop for these rulings was a phenomenal expansion, in the years from 1961 to 1971, of women’s employment outside the home,¹¹⁶ the civil rights movement of the 1960s and the precedents set in that struggle,¹¹⁷ and a revived feminist movement, fueled abroad and in the United States by Simone de Beauvoir’s remarkable 1949 publication, The Second Sex.¹¹⁸ In the main, the Court invalidated laws that had become obsolete, retained into the 1970s by only a few of the states.¹¹⁹ In a core set of cases, however, those dealing with social insurance benefits for a worker’s spouse or family,¹²⁰ the decisions did not utterly condemn the legislature’s product. Instead, the Court, in effect, opened a dialogue with the political branches of government. In essence, the Court instructed Congress and state legislatures: rethink ancient positions on these questions. Should you determine that special treatment for women is warranted, i.e., compensatory legislation because of the sunken-in social and economic bias or disadvantage women encounter, we have left you a corridor in which to move.¹²¹ But your classifications must be refined, adopted for remedial reasons, and not rooted in prejudice about “the way women (or men) are.”¹²² In the meantime, the Court’s decrees removed no benefits; instead, they extended to a woman worker’s husband, widower, or family benefits Congress had authorized only for members of a male worker’s family.¹²³

这些判决的背景是，在1961年至1971年期间，妇女在家庭以外就业的人数显著增加，¹¹⁶ 1960年代的民权运动以及在这场斗争中确立的先例，¹¹⁷ 以及由西蒙娜-德-波伏瓦1949年出版的《第二性》¹¹⁸在国外和美国复兴的女权运动。主要而言，法院宣布那些已经过时、只有少数几个州还保留到1970年代的法律无效。¹¹⁹ 然而，在一系列案件的核心，那些涉及工人配偶或家庭的社会保险福利的案件，¹²⁰ 判决并没有完全谴责立法机构的立法。相反，法院实际上是与政府的政治部门展开了对话。从本质上讲，法院指示国会和州立法机构：重新思考对这些问题的古老立场。如果你们确定有必要对妇女进行特殊处理，即针对妇女在社会和经济方面遇到的偏见或劣势的补偿性立法，那我们为你们留下了可行的通道。¹²¹ 但你们的分类必须细致，应该是出于补救的原因而采用，而不是植根于对 “女性（或男性）就是这样 “的偏见。¹²² 同时，法院的法令没有取消任何福利；相反，它们将国会只授权给男性工人家庭成员的福利扩展到女性工人的丈夫、鳏夫或家庭。¹²³

The ball, one might say, was tossed by the Justices back into the legislators’ court, where the political forces of the day could operate. The Supreme Court wrote modestly, it put forward no grand philosophy;¹²⁴ but by requiring legislative reexamination of once-customary sex-based classifications, the Court helped to ensure that laws and regulations would “catch up with a changed world.”¹²⁵

可以说，大法官们把球扔回了立法者的场地，在那里，当时的政治力量可以运作。最高法院写得很谦虚，没有提出任何宏大的理念；¹²⁴ 但通过要求立法机构重新审查曾经习惯的基于性别的分类，法院帮助确保法律和法规能够 “跟上变化的世界”¹²⁵ 。

Roe v. Wade,¹²⁶ in contrast, invited no dialogue with legislators. Instead, it seemed entirely to remove the ball from the legislators’ court. In 1973, when Roe issued, abortion law was in a state of change across the nation. As the Supreme Court itself noted, there was a marked trend in state legislatures “toward liberalization of abortion statutes.”¹²⁷ That movement for legislative change ran parallel to another law revision effort then under way—the change from fault to no-fault divorce regimes, a reform that swept through the state legislatures and captured all of them by the mid-1980s.¹²⁸

相比之下，”罗诉韦德 “案¹²⁶ 没有邀请立法者进行对话。相反，它似乎完全是为了把球从立法者的场地上拿走。1973年，当罗案判决时，全国的堕胎法正处于转变状态。正如最高法院本身所指出的，各州立法机构有一个明显的趋势，即 “堕胎法规的自由化”。¹²⁷ 这一立法变革运动与当时正在进行的另一项法律修订工作一起推进——从过错离婚制度到无过错离婚制度的转变，这一离婚制度改革席卷了各州立法机构，并在20世纪80年代中期将其全部占领。¹²⁸

No measured motion, the Roe decision left virtually no state with laws fully conforming to the Court’s delineation of abortion regulation still permissible.¹²⁹ Around that extraordinary decision, a well-organized and vocal right-to-life movement rallied and succeeded, for a considerable time, in turning the legislative tide in the opposite direction.

罗案的判决不够审慎，它使得几乎所有州的法律都不符合最高法院允许的对堕胎的划定。¹²⁹ 围绕着这一特殊的决定，一场组织严密、声势浩大的生命权运动集结起来，并在相当长的一段时间内成功地将立法潮流转向相反的方向。

Constitutional review by courts is an institution that has been for some two centuries our nation’s hallmark and pride.¹³⁰ Two extreme modes of court intervention in social change processes, however, have placed stress on the institution. At one extreme, the Supreme Court steps boldly in front of the political process, as some believe it did in Roe.¹³¹ At the opposite extreme, the Court in the early part of the twentieth century found—or thrust—itself into the rear guard opposing change, striking down, as unconstitutional, laws embodying a new philosophy of economic regulation at odds with the nineteenth century’s laissez-faire approach.¹³² Decisions at both of these poles yielded outcries against the judiciary in certain quarters. The Supreme Court, particularly, was labeled “activist” or “imperial,” and its precarious position as final arbiter of constitutional questions was exposed.¹³³

法院的宪法审查是一种制度，大约两个世纪以来一直是我们国家的标志和骄傲。¹³⁰ 然而，法院干预社会变革进程的两种极端模式给该制度带来了压力。在一个极端，最高法院大胆地走到了政治进程的前面，正如一些人认为它在罗案中所做的那样。¹³¹在另一个极端，法院在20世纪早期发现——或者说把自己变成了——反对变革的后卫，把体现了与19世纪自由放任的经济监管新理念的法律视为违宪而予以驳回。¹³² 这两个极端的判决都在某些方面引起了对司法机构的强烈不满。尤其是最高法院，被贴上了 “能动主义者 “或 “帝国主义 “的标签，其作为宪法问题最终仲裁者的不稳定地位被暴露出来。¹³³

I do not suggest that the Court should never step ahead of the political branches in pursuit of a constitutional precept. Brown v. Board of Education,¹³⁴ the 1954 decision declaring racial segregation in public schools offensive to the equal protection principle, is the case that best fits the bill. Past the midpoint of the twentieth century, apartheid remained the law-enforced system in several states, shielded by a constitutional interpretation the Court itself advanced at the turn of the century—the “separate but equal” doctrine.¹³⁵

我并不是说，法院在追求宪法戒律的过程中永远不应该走在政治部门的前面。布朗诉教育委员会案¹³⁴，即1954年宣布公立学校的种族隔离违反了平等保护原则的判决，是最符合该法案的案例。过了二十世纪中期，种族隔离仍然是几个州的法律执行制度，受到法院自己在世纪之交提出的宪法解释——”隔离但平等 “原则的保护。¹³⁵

In contrast to the legislative reform movement in the states, contemporaneous with Roe, widening access to abortion, prospects in 1954 for state legislation dismantling racially segregated schools were bleak. That was so, I believe, for a reason that distances race discrimination from discrimination based on sex. Most women are life partners of men; women bear and raise both sons and daughters. Once women’s own consciousness was awakened to the unfairness of allocating opportunity and responsibility on the basis of sex, education of others—of fathers, husbands, sons as well as daughters—could begin, or be reinforced, at home.¹³⁶ When blacks were confined by law to a separate sector, there was no similar prospect for educating the white majority.¹³⁷

与各州的立法改革运动形成鲜明对比的是与罗案同时期的扩大堕胎机会的运动，1954年各州立法拆除种族隔离的学校的前景是暗淡的。我认为，之所以如此，是因为种族歧视与基于性别的歧视有一定的区别。大多数女性是男性的生活伴侣；女性既生育又抚养儿子和女儿。一旦妇女自己的意识被唤醒，意识到基于性别分配机会和责任的不公平性，对其他人——父亲、丈夫、儿子和女儿——的教育就可以从家庭开始，或者得到加强。¹³⁶当黑人被法律限制在一个单独的部门时，并不存在类似的自动教育多数白人的机会。¹³⁷

It bears emphasis, however, that Brown was not an altogether bold decision. First, Thurgood Marshall and those who worked with him in the campaign against racial injustice, carefully set the stepping-stones leading up to the landmark ruling.¹³⁸ Pathmarkers of the same kind had not been installed prior to the Court’s decision in Roe.¹³⁹ Second, Brown launched no broadside attack on the Jim Crow system in all its institutional manifestations. Instead, the Court concentrated on segregated schools;¹⁴⁰ it left the follow-up for other days and future cases. A burgeoning civil rights movement—which Brown helped to propel—culminating in the Civil Rights Act of 1964,¹⁴¹ set the stage for the Court’s ultimate total rejection of Jim Crow legislation.

然而，值得强调的是，布朗案并不是一个完全大胆的决定。首先，瑟古德-马歇尔和那些在反对种族不公正的运动中与他合作的人，精心设置了通往这一里程碑式判决的垫脚石。.¹³⁸在法院对罗案作出判决之前则还没有类似的路标。¹³⁹ 其次，布朗案没有对吉姆克劳制度的所有制度性表现进行广泛攻击。相反，法院专注于种族隔离的学校；¹⁴⁰ 它把后续工作留给了其他日子和未来的案件。蓬勃发展的民权运动——布朗案帮助推动了这一运动——最终形成了1964年的《民权法案》，¹⁴¹ 为法院最终完全拒绝吉姆-克劳法奠定了基础。

Significantly, in relation to the point I just made about women and men living together, the end of the Jim Crow era came in 1967, thirteen years after Brown: the case was Loving v. Virginia,¹⁴² the law under attack, a state prohibition on interracial marriage. In holding that law unconstitutional, the Court effectively ruled that, with regard to racial classifications, the doctrine of “separate but equal” was dead—everywhere and anywhere within the governance of the United States.¹⁴³

重要的是，就我刚才提出的关于男女同居的观点而言，吉姆-克劳时代的结束是在1967年，即布朗案的13年之后：案件是Loving诉弗吉尼亚，¹⁴²受到攻击的法律是州政府禁止异族通婚的规定。在认定该法律违宪的同时，法院有效地裁定，在种族分类方面，”隔离但平等”的理论已经死亡——在美国的任何地方和管理范围内。¹⁴³

The framers of the Constitution allowed to rest in the Court’s hands large authority to rule on the Constitution’s meaning; but the framers, as I noted at the outset, armed the Court with no swords to carry out its pronouncements. President Andrew Jackson in 1832, according to an often-told legend, said of a Supreme Court decision he did not like: “The Chief Justice has made his decision, now let him enforce it.”¹⁴⁴ With prestige to persuade, but not physical power to enforce, with a will for self-preservation and the knowledge that they are not “a bevy of Platonic Guardians,”¹⁴⁵ the Justices generally follow, they do not lead, changes taking place elsewhere in society.¹⁴⁶ But without taking giant strides and thereby risking a backlash too forceful to contain, the Court, through constitutional adjudication, can reinforce or signal a green light for a social change. In most of the post-1970 gender-classification cases, unlike Roe, the Court functioned in just that way. It approved the direction of change through a temperate brand of decisionmaking, one that was not extravagant or divisive. Roe, on the other hand, halted a political process that was moving in a reform direction and thereby, I believe, prolonged divisiveness and deferred stable settlement of the issue. The most recent Planned Parenthood decision¹⁴⁷ notably retreats from Roe¹⁴⁸ and further excludes from the High Court’s protection women lacking the means or the sophistication to surmount burdensome legislation.¹⁴⁹ The latest decision may have had the sanguine effect, however, of contributing to the ongoing revitalization in the 1980s and 1990s of the political movement in progress in the early 1970s, a movement that addressed not simply or dominantly the courts but primarily the people’s representatives and the people themselves. That renewed force, one may hope, will—within a relatively short span—yield an enduring resolution of this vital matter in a way that affirms the dignity and equality of women.¹⁵⁰

宪法制定者允许最高法院拥有对宪法含义进行判决的巨大权力；但正如我在一开始指出的那样，宪法制定者没有给法院配备执行其声明的剑。根据一个经常被提及的传说，1832年安德鲁-杰克逊总统在谈到他不喜欢的一项最高法院判决时说 “首席大法官已经做出了他的决定，现在让他来执行吧。”¹⁴⁴大法官有说服力，但没有执行力，他们有自我保护的意愿，也知道自己不是 “一群柏拉图式的监护人”，¹⁴⁵大法官通常会跟随而不是领导社会上其他地方发生的变化。 ¹⁴⁶即使不冒着难以遏制的反弹的风险采取大踏步的行动，法院仍可以通过宪法判决来加强或发出社会变革的绿灯。在1970年后的大多数性别分类案件中，与罗案不同，法院正是以这种方式发挥作用。它通过一种温和的判决批准了变革的方向，这种决策不是不切实际的或分裂的。另一方面，罗案停止了一个正朝着改革方向发展的政治进程，因此，我认为，它延长了分歧，推迟了问题的稳定解决。最近的宾州东南部计划生育组织诉凯西案的判决¹⁴⁷明显地从罗案¹⁴⁸中退了出来，并进一步将缺乏克服繁琐立法的手段或能力的妇女排除在高等法院的保护之外。¹⁴⁹然而，最新的判决可能产生了令人振奋的效果，即促进了20世纪80年代和90年代正在进行的政治运动的振兴，这一运动不是简单或主要针对法院，而是主要针对人民的代表和人民本身。人们希望，这种新的力量将在相对较短的时间内，以肯定妇女尊严和平等的方式，对这一重要问题作出持久的解决。¹⁵⁰

Conclusion 结论

To sum up what I have tried to convey in this lecture, I will recall the counsel my teacher and friend, Professor Gerald Gunther, offered when I was installed as a judge. Professor Gunther had in mind a great jurist, Judge Learned Hand, whose biography Professor Gunther is just now completing. The good judge, Professor Gunther said, is “openminded and detached . . . heedful of limitations stemming from the judge’s own competence and, above all, from the presuppositions of our constitutional scheme; th[at] judge . . . recognizes that a felt need to act only interstitially does not mean relegation of judges to a trivial or mechanical role, but rather affords the most responsible room for creative, important judicial contributions.”¹⁵¹

为了总结我在这次演讲中试图表达的内容，我将回顾我的老师和朋友杰拉尔德-冈瑟教授在我被任命为法官时提出的建议。冈瑟教授想到了一位伟大的法学家——勒恩德·汉德法官，冈瑟教授刚刚完成了他的传记。冈瑟教授说，好的法官是 “心胸开阔，超然物外……注意来自法官自身能力的限制，尤其是来自我们宪法制度的预设；该法官……认识到，感觉到只需在间隙中行事并不意味着将法官降格为琐碎或机械的角色，而是为创造性的重要司法贡献提供了最负责任的空间”¹⁵¹ 。

See Norman Dorsen, Foreword to The Evolving Constitution, at x (Norman Dorsen ed., 1987). ↩ ↩²
Annals of Congress 457 (Joseph Gales ed., 1789), quoted in Chapman v. California, 386 U.S. 18, 21 n.4 (1967). While Madison originally doubted the efficacy of bills of rights, he eventually joined Jefferson in recognizing the value of “the legal check [a declaration of rights] puts into the hands of the judiciary.” 虽然麦迪逊最初对权利法案的效力表示怀疑，但他最终与杰斐逊一样，承认”[权利宣言]对司法机关的法律检查 “的价值。Letter from Thomas Jefferson to James Madison (Mar. 15, 1789), in 14 The Papers of Thomas Jefferson 659 (Julian P. Boyd ed., 1958); see also Maeva Marcus, The Adoption of the Bill of Rights, 1 Win. & Mary Bill of Rts. J. 115, 117-19 (1992). ↩ ↩²
See The Federalist No. 78, at 466 (Alexander Hamilton) (Clinton Rossiter ed., 1961). ↩ ↩²
见前注，第465页。 ↩ ↩²
I borrow language here from Professor Brainerd Currie’s guide to analysis of choice-of-law cases in which the policies of two states are in apparent conflict. 我在这里借用布雷纳德-库里教授的语言来分析当两个州的政策明显冲突时该如何选择法律。See Brainerd Currie, The Disinterested Third State, 28 Law & Contemp. Probs. 754, 757 (1963) (“[T]he conception should be re-examined, with a view to a more moderate and restrained interpretation both of the policy and of the circumstances in which it must be applied to effectuate the forum’s legitimate purpose.应该重新审查这一概念，以便对政策和为实现论坛的合法目的而必须适用的情况作出更温和和克制的解释”); see also Herna Hill Kay, A Defense of Currie’s Governmental Interest Analysis, 215 Recueil des Cours 10, 68-73 (1989-III). ↩ ↩²
N. Dorsen, supra note 1, at xii (quoting McCulloch v. Maryland, 17 U.S. (4 Wheat.) 316, 415 (1819)). ↩ ↩²
同前注，(引用Home Bldg. & Loan Ass’n v. Blaisdell, 290 U.S. 398, 443 (1934)). ↩ ↩²
同前注。 ↩ ↩²
See Thurgood Marshall, Reflections on the Bicentennial of the United States Constitution, 101 Harv. L. Rev. 1, 2 (1987). ↩ ↩²
See Richard B. Morris, The Forging of the Union, 1781-1789, at 162-93 (1987); Linda K. Kerber, “Ourselves and Our Daughters Forever”: Women and the Constitution, 1787- 1876, This Constitution: A Bicentennial Chron., Spring 1985, at 25. ↩ ↩²
See Deborah Jones Merritt, What’s Missing from the Bill of Rights?, 1991 U. IM. L Rev. 765, 766-69. ↩ ↩²
Letter from John Adams to James Sullivan (May 26, 1776), in 9 The Works of John Adams 378 (Charles Francis Adams ed., 1854). ↩ ↩²
See, e.g., Arthur Goldberg, Equality and Governmental Action, in The Evolving Constitution 25 (Norman Dorsen ed., 1987); J. Skelly Wright, Public School Desegregation, in id. at 44; Abe Fortas, Equal Rights-for Whom?, in id. at 85; Thurgood Marshall, Group Action in the Pursuit of Justice, in id. at 85. ↩ ↩²
See Merritt, supra note 11, at 765; R. Morris, supra note 10, at 162-63. ↩ ↩²
Letter from Thomas Jefferson to Albert Gallatin (Jan. 13, 1807), in I The Writings of Albert Gallatin 328 (Henry Adams ed., 1960). Jefferson, who declared it self-evident “that all men are created equal,” also expressed this once prevailing view: “Were our State a pure democracy.., there would yet be excluded from our deliberations.., women, who, to prevent depravation of morals and ambiguity of issues, should not mix promiscuously in the public meetings of men.”宣布 “人人生而平等 “是不言而喻的杰斐逊，也表达了这种曾经盛行的观点。”如果我们的国家是一个纯粹的民主国家……，那么在我们的讨论中还会排除妇女，为了防止道德的堕落和问题的含糊不清，她们不应该在男人的公共会议中乱混。 Letter from Thomas Jefferson to Samuel Kercheval (Sept. 5, 1816), in 10 The Writings of Thomas Jefferson 46 n.1 (Paul L. Ford ed., 1899). ↩ ↩²
R. Morris, supra note 10, at 193. ↩ ↩²
The Federalist No. 78, supra note 3, at 465. ↩ ↩²
同前注，at 465, 471. ↩ ↩²
1 have earlier addressed this topic in Ruth Bader Ginsburg, Remarks on Writing Separately, 65 Wash. L. Rev. 133 (1990). See also Ruth Bader Ginsburg, Styles of Collegial Judging, 39 Fed. Bar News & J. 199 (1992). These remarks borrow from, revise, and build upon my prior lectures. ↩ ↩²
See Louis Blom-Cooper & Gavin Drewry, Final Appeal: A Study of the House of Lords in Its Judicial Capacity 81-82, 523 (1972); see also Alan Paterson, The Law Lords 109-10 (1982) (noting that the Lords no longer routinely deliver five separate opinions). ↩ ↩²
It has been said of the French tradition that the ideal judgment is “considered all the more perfect for its concise and concentrated style, so that only experienced jurists are able to understand and admire it”关于法国的传统，有人说，理想的判决书 “因其简洁和集中的风格而被认为更加完美，因此只有经验丰富的法学家才能理解和欣赏它。 Ren6 David & John E.C. Brierley, Major Legal Systems in the World Today 129 (2d ed. 1978). ↩ ↩²
See Karl M. ZoBell, Division of Opinion in the Supreme Court: A History of Judicial Disintegration, 44 Cornell L.Q. 186, 193 (1959); see also G. Edward White, The Working Life of the Marshall Court, 1815-1835, 70 Va. L. Rev. 1, 36-47 (1984). ↩ ↩²
Letter from Thomas Jefferson to Thomas Ritchie (Dec. 25, 1820), in 10 The Writings of Thomas Jefferson 169, 171 (Paul L. Ford ed., 1899). ↩ ↩²
See ZoBell, supra note 22, at 196 & n.57. ↩ ↩²
I first used this illustration in Ruth Bader Ginsburg, On Muteness, Confidence, and Collegiality: A Response to Professor Nagel, 61 Colo. L. Rev. 715, 718 (1990). ↩ ↩²
United States v. Rosenberg, 888 F.2d 1406 (D.C. Cir. 1989). ↩ ↩²
Letter from B. Ducamin, President, Section of Finances, Conseil d’Etat, to Judge Ruth Bader Ginsburg (Dec. 15, 1989) (on file with the New York University Law Review). ↩ ↩²
But cf. L. Blom-Cooper & G. Drewry, supra note 20, at 81 (observing that, as an exception to the individual opinion tradition, separate opinions in English criminal appeals are disfavored and may be presented only when the presiding judge so authorizes观察到，作为个人意见传统的一个例外，英国刑事上诉中的单独意见是不受欢迎的，只有在主审法官授权的情况下才能提出). ↩ ↩²
See Robert W. Bennett, A Dissent on Dissent, 74 Judicature 255, 258-59 (1991). ↩ ↩²
See ZoBell, supra note 22, at 202. ↩ ↩²
Walter F. Murphy, Elements of Judicial Strategy 62 (1964) (quoting Letter from then Justice Stone to Karl Llewellyn (Feb. 4, 1935)). ↩ ↩²
Burnet v. Coronado Oil & Gas Co., 285 U.S. 393, 406 (1932) (Brandeis, J., dissenting) (citation omitted); see also Di Santo v. Pennsylvania, 273 U.S. 34, 42 (1927) (Brandeis, J., dissenting). ↩ ↩²
Erie R.R. Co. v. Tompkins, 304 U.S. 64 (1938) (holding that federal courts must apply state law except in matters governed by the federal Constitution or by Acts of Congress). ↩ ↩²
Colloquy, Proceedings of the Forty-Ninth Judicial Conference of the District of Columbia Circuit (May 24, 1988) (comment of Paul A. Freund), reprinted in 124 F.R.D. 241, 336, 347 (1989). ↩
See Summary of Annual Report of D.C. Circuit Opinions for Statistical Year July 1. 1991 - June 30, 1992 (the 405 figure includes 291 opinions published in the Federal Reporter series and 114 unpublished opinions issued in cases resolved after oral argument; separate concurrences and dissents numbered 55). ↩ ↩²
See The Supreme Court, 1991 Term-The Statistics, 106 Harv. L. Rev. 378, 380 (1992). For the 68 memorandum orders, however, the unanimity rate was 91.2%. See id. ↩ ↩²
See, e.g., George Archibald, Free Hill Mailings to Future Districts Banned by Court, Wash. Times, July 31, 1992, at A3 (“A three-judge panel of the U.S. Court of Appeals for the District of Columbia split 2-1…. Judges Laurence H. Silberman, a Reagan appointee, and A. Raymond Randolph, appointed by President Bush, voted to overturn” the district court’s ruling; “Judge Patricia M. Wald, a Carter appointee, voted [to affirm].”); Philip J. Hilts, Judge Overturns Federal Seizure of Abortion Pill, N.Y. Times, July 15, 1992, at Al (“In the increasingly political battle over [the] RU486 [abortion pill], the decision favoring the woman, who brought the drug into the country to test the ban, came from a judge who had been appointed by President Jimmy Carter. Later, his order was stopped by a panel of three judges… : John M. Walker, President Bush’s cousin and a Bush appointee, and Frank X. Altimari and Daniel J. Mahoney, both appointed by President Ronald Reagan.”); Karen Riley, Mayor to Flout Court Ruling, Wash. Times, May 9, 1992, at Al (“[‘Mayor Sharon Pratt Kelly] said the Bush and Reagan administrations had packed the federal bench with judges who are in ‘retreat’ on civil rights issues”; she “threatened to defy” a unanimous federal appeals court decision “by two Carter-appointed judges and a judge appointed by President Bush” striking down the District’s minority contracting program.); Cindy Rugeley, Abortion Fight Now Heads to Legislature, Houston Chron., June 30, 1992, at All (“‘President Bush has changed his opinion on abortion and so it’s not surprising to see the Supreme Court-a majority of whom have been appointed by President Bush or Reagan-ignoring its own precedent and changing its opinion on a woman’s right to choose.’ “) (quoting Texas Lieutenant Governor Bob Bullock); Will DeKalb Students Win?, Atlanta J. & Const., April 2, 1992, at A18 (editorial) (People seeking “to end the last vestiges of segregation in American schools” face “a federal judiciary dominated by conservatives appointed by Presidents Reagan and Bush”; “[t]hose judges are likely to be more sympathetic to school officials arguing for a return of local control than to minority students seeking remedies to the lingering effects of segregation.”). But see, e.g., Mary Deibel, Supreme Surprises, Star Trib. (Minneapolis), July 5, 1992, at 14A (“Of the three dozen cases in which the administration advocated a position, it lost 20 times, often because of the votes of the five justices appointed by Bush and his predecessor, Ronald Reagan.”). ↩ ↩²
See J. Woodford Howard, Jr., Courts of Appeals in the Federal Judicial System 189-221 (1981); Ruth Bader Ginsburg, The Obligation to Reason Why, 37 U. Fla. L. Rev. 205, 212, 216 (1985). ↩ ↩²
See note 32 and accompanying text supra. ↩ ↩²
See Ginsburg, supra note 38, at 215 & n.47. If a panel opinion plainly has not stood the test of time, the court can abbreviate the en banc process. See id. at 215 nA8. ↩ ↩²
Under current District of Columbia Circuit practice, judgments that will not be published in the Federal Reporter series, as well as decisions scheduled for publication, are circulated to the full court before release to the public.根据哥伦比亚特区巡回法院的现行做法，不会在《联邦报告》系列中发表的判决以及计划发表的判决，在向公众发布之前都会向全体法院传阅。 See D.C. Cir. R. 36(a)(2), (c). ↩ ↩²
On the check exerted by colleagues, Chief Justice William H. Rehnquist has described the practice of one of his predecessors, Chief Justice Charles Evans Hughes: He approached his own decisions with his usual meticulous care, turning out innumerable drafts in order to be certain of the most correct and precise language. But he had no particular pride of authorship, and if in order to secure a vote he was forced to put in some disconnected or disjointed thoughts or sentences, in they went and let the law schools concern themselves with what they meant.他以其一贯的谨慎态度对待自己的决定，为了确定最正确和最精确的语言，他翻出了无数的草稿。但他对自己的著作没有特别的自豪感，如果为了确保投票，他不得不把一些不连贯或不连贯的想法或句子写进去，那就写进去，让法学院自己去关注它们的意思。William tL Rehnquist, Chief Justices I Never Knew, 3 Hastings Const. LQ. 637, 643 (1976) (quoting Edwin McElwain, The Business of the Supreme Court as Conducted by ChiefJustice Hughes, 63 Harv. L. Rev. 5, 19 (1949)). ↩ ↩²
Cf. Jon 0. Newman, The Second Circuit Review, 1987-1988 Term-Foreword: In Banc Practice in the Second Circuit, 1984-1988, 55 Brook. L. Rev. 355, 369-70 (1989) (“[I]f we were confronting one another frequently each year as members of an in banc court, I believe there would be at least some risk to the extremely high level of civility that now pervades our relationships both in the decision-making and opinion-writing phases of our work.”如果我们每年都频繁地与全员庭审成员相互对峙，我相信那至少会在一定程度上威胁到我们目前工作中决策和意见撰写阶段极高的文明水平。). ↩ ↩²
See Frank M. Coffin, The Ways of a Judge: Reflections from the Federal Appellate Bench 181-88 (1980). ↩ ↩²
William J. Brennan, Jr., In Defense of Dissents, 37 Hastings L.J. 427, 437 (1986). ↩ ↩²
Coalition for the Preservation of Hispanic Broadcasting v. FCC, 931 F.2d 73, 80 (D.C. Cir. 1991) (Mikva, C.J., dissenting); see also id. at 84 (court’s decision is “driven more by evasion than logic”; “nonsensical results will flow from [it]”). ↩ ↩²
Roscoe Pound, Cacoethes Dissentiendi: The Heated Judicial Dissent, 39 A.B.A. J. 794, 795 (1953). For more recent commentary, see Brenda Jones Quick, Whatever Happened to Respectful Dissent?, A.B.A. J., June 1991, at 62. ↩ ↩²
R.A.V. v. St. Paul, 112 S. Ct. 2538, 2560 (1992) (White, J., concurring in the judgment) (“I join the judgment, but not the folly of the opinion.”). ↩ ↩²
Lee v. Weisman, 112 S. Ct 2649, 2679, 2681, 2685 (1992) (Scalia, J., dissenting) (describing Court’s opinion as “oblivious to our history,” “incoherent,” “nothing short of ludicrous,” and “a jurisprudential disaster”). ↩ ↩²
Planned Parenthood v. Casey, 112 S. Ct. 2791, 2875 (1992) (Scalia J., concurring in the judgment in part and dissenting in part) (“I must… respond to a few of the more outrageous arguments in today’s opinion, which it is beyond human nature to leave unanswered.”). ↩ ↩²
Webster v. Reproductive Health Servs., 492 U.S. 490, 532 (1989) (Scalia, J., concurring in part and concurring in the judgment) (“Justice O’Connor’s assertion… that a fundamental rule of judicial restraint’ requires us to avoid reconsidering Roe, cannot be taken seriously.”); cf. Payne v. Tennessee, 111 S. CL 2597, 2622 (1991) (Marshall, J., dissenting) (“majority cannot sincerely expect anyone to believe [its assertion]”); McCleskey v. Zant, I11 S. Ct. 1454, 1481 (1991) (Marshall, J., dissenting) (“It is difficult to take [the majority’s] reasoning seriously.”); Maislin Indus., U.S., Inc. v. Primary Steel, Inc., 497 U.S. 116, 139 (1990) (Stevens, J., dissenting) (“Even wearing his famous blinders, old Dobbin would see through the tired arguments the Court accepts today.”). ↩ ↩²
Coleman v. Thompson, 111 S. Ct. 2546, 2574 (1991) (Blackmun, J., dissenting) (describing majority’s distinction as “inexplicable” and its conception as “entirely unprecedented”). ↩ ↩²
同前注，at 2578. ↩ ↩²
Morgan v. Illinois, 112 S. CL 2222, 2242 (1992) (Scalia, J., dissenting) (Today … the Court strikes a further blow against the People in its campaign against the death penalty.”). ↩ ↩²
Central States Motor Freight Bureau v. ICC, 924 F.2d 1099, 1112 (D.C. Cir. 1991) (Silberman, J., dissenting) (majority’s suggestion “is naked analytical bootstrapping”). ↩ ↩²
Synovus Fin. Corp. v. Board of Governors of the Fed. Reserve Sys., 952 F.2d 426, 437 (D.C. Cir. 1991) (Silberman, J., dissenting) (“The majority’s opinion… is reminiscent for Civil War buffs of Sherman’s march through Georgia. Principles of administrative law are brushed aside like Johnston and Hood’s army. Our precedents are overturned like Georgia plantations.”). For a restrained and moderate reply from a South Carolinian, see id. at 437 n.8 (Henderson, J.) (“With all respect to our colleague in dissent, to equate the legal issues in this case with a Civil War campaign manifests not only a misunderstanding of those issues but also a lack of appreciation for a wrenching event in our country’s history.”). ↩ ↩²
Planned Parenthood v. Casey, 112 S. CL 2791, 2882 (1992) (Scalia, J., concurring in the judgment in part and dissenting in part) (“[T]o portray Roe as the statesmanlike ‘settlement’ of a divisive issue, a jurisprudential Peace of Westphalia that is worth preserving, is nothing less than Orwellian.”); County of Allegheny v. ACLU, 492 U.S. 573, 678 (1989) (Kennedy, J., concurring in the judgment in part and dissenting in part) (“Court lends its assistance to an Orwellian rewriting of history”); FCC v. League of Women Voters, 468 U.S. 364, 417 n.10 (1984) (Stevens, J., dissenting) (majority’s argument “would be laughable were it not so Orwellian”); United Steelworkers v. Weber, 443 U.S. 193, 219-21 (1979) (Rehnquist, J., dissenting) (Court has behaved like “Orwellian speaker” who, in mid-sentence, “‘switched from one line to the other’ “) (quoting George Orwell, 1984, 181-82 (1949)). ↩ ↩²
Collins J. Seitz, Collegiality and the Court of Appeals, 75 Judicature 26, 27 (1991). ↩ ↩²
同前注 ↩ ↩²
同前注 ↩ ↩²
Dred Scott v. Sandford, 60 U.S. (19 How.) 393 (1856), cited as a model of “reasoned discussion” in Pound, supra note 47, at 797. ↩ ↩²
See, e.g., Welsh v. United States, 398 U.S. 333, 344 (1970) (Harlan, J., concurring). ↩ ↩²
The various High Court and Supreme Court opinions in this case are reprinted in The Attorney General v. X and Others (Sunniva McDonagh ed., 1992). ↩ ↩²
Ireland Const. art. 40.3.3 (inserted following enactment of the Eighth Amendment of the Constitution Act, 1983). Following a referendum on November 25, 1992, two sentences were added to Article 40.3.3: “This subsection shall not limit freedom to travel between the State and another state”; “This subsection shall not limit freedom to obtain or make available, in the State, subject to such conditions as may be laid down by law, information relating to services lawfully available in another state.” ↩ ↩²
See Opinion of Costello, J. (Ir. H. Ct.) (Feb. 17, 1992), reprinted in The Attorney General v. X and Others, supra note 63, at 9. ↩ ↩²
Opinion of Finlay, CJ. (Ir. S.C.) (Mar. 5, 1992) (quoting McGee v. Attorney General, [1974] I.R. 284, 318 (Walsh, J.)), reprinted in The Attorney General v. X and Others, supra note 63, at 47, 59. ↩ ↩²
同前注。 (quoting State (Healy) v. Donoghue, [1976] I.R. 326, 347 (O’Higgins, C.J.)); cf. text accompanying notes 6-8 supra. ↩ ↩²
Cf., e.g., Planned Parenthood v. Casey, 112 S. Ct. 2791, 2882 (1992) (Scalia, J., concurring in the judgment in part and dissenting in part) (“The Imperial Judiciary lives.”). ↩ ↩²
See Opinion of Hederman, J. (Ir. S.C.) (Mar. 5, 1992), reprinted in The Attorney General v. X and Others, supra note 63, at 69, 83. ↩ ↩²
同前注 ↩ ↩²
同前注 ↩ ↩²
See generally Mary Ann Glendon, Abortion and Divorce in Western Law (1987); Donald P. Kommers, Abortion and Constitution: United States and West Germany, 25 Am. J. Comp. L. 255 (1977). ↩ ↩²
Dissents might concede, for example, more often than they do, that “[t]he majority’s argument is by no means implausible.” Hubbard v. EPA, 949 F.2d 453, 469 (D.C. Cir. 1991) (Wald, J., concurring in part and dissenting in part). ↩ ↩²
Final Report of the Committee on Civility of the Seventh Federal Judicial Circuit 3 (1992). ↩ ↩²
同前注，at 7A; see also Interim Report of the Committee on Civility of the Seventh Federal Judicial Circuit 3, 13, 39-42 (1991). ↩ ↩²
See generally Louis Fisher, Constitutional Dialogues: Interpretation as Political Process (1988). Recent commentary on court-legislature communication includes Shirley S. Abrahamson & Robert L. Hughes, Shall We Dance? Steps for Legislators and Judges in Statutory Interpretation, 75 Minn. L. Rev. 1045 (1991); Robert A. Katzmann, Bridging the Statutory Gulf Between Courts and Congress: A Challenge for Positive Political Theory, 80 Geo. L.J. 653 (1992); Deanell Reece Tacha, Judges and Legislators: Renewing the Relationship, 52 Ohio St. L.J. 279 (1991). ↩ ↩²
Southern Pac. Co. v. Jensen, 244 U.S. 205, 221 (1917) (Holmes, J., dissenting) ↩ ↩²
The Supreme Court’s post-1970 decisions on alienage as a “suspect” category are illustrative. Compare Graham v. Richardson, 403 U.S. 365, 372 (1971) (invalidating state legislation denying public assistance benefits to resident aliens, Court declared that “classifications based on alienage, like those based on nationality or race, are inherently suspect [under equal protection principles] and subject to close judicial scrutiny”) (footnotes omitted) with Cabell v. Chavez-Salido, 454 U.S. 432, 436 (1982) (upholding citizenship requirement for a state’s probation officers, Court commented that alienage cases “illustrate a not unusual characteristic of legal development: broad principles are articulated, narrowed when applied to new contexts, and finally replaced when the distinctions they rely upon are no longer tenable”)最高法院在1970年后关于外国人作为 “嫌疑人 “类别的决定可以说明问题。比较Graham v. Richardson, 403 U.S. 365, 372 (1971)(宣布拒绝向居留外国人提供公共援助福利的州立法无效，法院宣布 “基于外国人的分类，就像基于国籍或种族的分类一样，[根据平等保护原则]本质上是可疑的，应受到严格的司法审查”(脚注省略)与Cabell v. Chavez-Salido, 454 U.S. 432, 436 (1982) (支持对一个州的缓刑犯的公民身份要求，法院评论说，外国人案件 “说明了法律发展的一个常见的特点：广泛的原则被阐明，在适用于新的环境时被缩小，最后在它们所依赖的区别不再成立时被取代。） ↩ ↩²
410 U.S. 113 (1973). ↩ ↩²
Ruth Bader Ginsburg, Some Thoughts on Autonomy and Equality in Relation to Roe v. Wade, 63 N.C. L. Rev. 375 (1985) ↩ ↩²
Justices White and Rehnquist dissented. ↩ ↩²
Roe, 410 U.S. at 164 (emphasis in original). ↩ ↩²
In a companion case, Doe v. Bolton, 410 U.S. 179 (1973), the Court, again 7-2, held unconstitutional several provisions of Georgia’s abortion law. The Georgia statute, enacted in 1968, had moved a considerable distance from the Texas extreme. It was based on the American Law Institute’s Model Penal Code formulation, and resembled reformed laws then in force in about one-fourth of the states. The Court might have deferred consideration of Doe v. Bolton pending its disposition of Roe; indeed, the Court might have awaited the Fifth Circuit’s resolution of an appeal taken by Georgia to the intermediate appellate court instead of ruling immediately on plaintiffs’ direct appeal from a three-judge district court decision holding in substantial part for plaintiffs.在 “无名氏诉博尔顿 “案中，法院再次以7比2认定佐治亚州的堕胎法中的若干条款违宪。佐治亚州的法规于1968年颁布，与德克萨斯州的极端情况相差甚远。它是基于美国法律协会的《示范刑法典》制定的，与当时约四分之一的州所实施的改革后的法律相似。法院可能在处理罗案时推迟了对”无名氏诉博尔顿 “案的审议；事实上，法院可能有意等到第五巡回法院解决乔治亚州向中级上诉法院提出的上诉，而不是立即对原告就地区法院三名法官的判决提出的直接上诉作出判决，该判决在很大程度上支持原告。 See Doe, 410 U.S. at 187 & n.8. ↩ ↩²
112 S. Ct. 2791 (1992). ↩ ↩²
见前注 at 2841-43 (Stevens, J., concurring in part and dissenting in part) (maintaining that 24-hour delay requirement and counseling provisions conflicted with Court precedent); id. at 2846, 2850-52 (Blackmun, J., concurring in part, concurring in the judgment in part, and dissenting in part) (maintaining that counseling, 24-hour delay, and parental consent provisions conflicted with Court precedent). ↩ ↩²
同前注， at 2809. On this point, the controlling Justices-Justices O’Connor, Kennedy, and Souter-spoke for the Court. ↩ ↩²
See generally Rachel N. Pine & Sylvia A. Law, Envisioning a Future for Reproductive Liberty: Strategies for Making the Rights Real, 27 Harv. C.R.-C.L L. Rev. 407 (1992). ↩ ↩²
See Roe v. Wade, 410 U.S. 113, 164-65 (1973) (“abortion decision … must be left to the medical judgment of the pregnant woman’s attending physician堕胎决定……必须留给孕妇的主治医生进行医学判断”; “decision [in Roe] vindicates the right of the physician to administer medical treatment according to his professional judgment[罗案]的决定维护了医生根据其专业判断实施医疗的权利”). ↩ ↩²
See Paul A. Freund, Storms over the Supreme Court, 69 A.B.A. J. 1474, 1480 (1983) ↩ ↩²
409 U.S. 947 (granting certiorari in 460 F.2d 1372 (9th Cir. 1971)), remanded for consideration of mootness, 409 U.S. 1071 (1972); see also Note, Pregnancy Discharges in the Military: The Air Force Experience, 86 Harv. L. Rev. 568 (1973). ↩ ↩²
See Appendix to Brief for Petitioner at 34a, Struck (No. 72-178) (Memorandum of Colonel Max B. Bralliar, May 14, 1971, recommending waiver of discharge for Captain Struck) ↩ ↩²
See Brief for Petitioner at 67-69 & n.70, Struck (No. 72-178). ↩ ↩²
See id. at 3-5, 56. ↩ ↩²
Air Force Regulation 36-12(40), set out in relevant part in Brief for Petitioner at 2-3, Struck (No. 72-178); see also Struck, 460 F.2d at 1374. ↩ ↩²
Struck, 460 F.2d at 1374. ↩ ↩²
Appendix to Brief for Petitioner at 22a, Struck (No. 72-178) (quoting Air Force policy on therapeutic abortion, contained in Air Force Regulation 169-12(C2) (Sept. 23, 1970)). On his second full day in office, President Clinton ended a total ban on abortions at U.S. military facilities, imposed during the 1980s, and ordered that abortions be permitted at such facilities if paid for with non-Department of Defense funds.克林顿总统在他上任的第二天，就结束了1980年代对美国军事设施内堕胎的全面禁止，并下令如果用非国防部的资金支付，则允许在这些设施内堕胎。See Memorandum on Abortions in Military Hospitals, Jan. 22, 1993, 29 Weekly Comp. Pres. Doc. 88 (Jan. 25, 1993). ↩ ↩²
As earlier observed, see text accompanying note 14 supra, the original Constitution and the Bill of Rights contain no equality guarantee. Since 1954, however, the Supreme Court has attributed to the fifth amendments due process clause an equal protection principle regarding federal action corresponding to the fourteenth amendment’s equal protection clause controlling state action. 如前所述，见上文附注14，原始宪法和权利法案不包含平等保障。然而，自1954年以来，最高法院将联邦行动的平等保护原则涵盖于于第五修正案的正当程序条款下，与第十四修正案中控制州行动的平等保护条款相对应。See Boiling v. Sharpe, 347 U.S. 497, 499 (1954) (initial recognition); cf. Weinberger v. Wiesenfeld, 420 U.S. 636, 638 n.2 (1975) (“Court’s approach to Fifth Amendment equal protection claims has always been precisely the same as to equal protection claims under the Fourteenth Amendment 法院对第五修正案的平等保护要求的处理方法一直与根据第十四修正案提出的平等保护要求完全相同”). ↩ ↩²
See Struck, 460 F.2d at 1380 (Duniway, J., dissenting); Brief for Petitioner at 8, 54-55 Struck (No. 72-178). The Air Force had asserted that the purpose of its pregnancy discharge regulation was to “encourage” birth control 空军宣称，其怀孕退伍条例的目的是 “鼓励 “节育. Brief for Respondents in Opposition to Certiorari at 11, Struck (No. 72-178). In response, Captain Struck observed, inter alia, that the .’encouragement’ [was] directed at females only”: “A man serves in the Air Force with no unwarranted governmental intrusion into the matter of his sexual privacy or his decision whether to beget a child. The woman serves subject to ‘regulation’; her pursuit of an Air Force career requires that she decide not to bear a child.” 对此，Struck上尉特别指出，”‘鼓励’[是]只针对女性的”。”一个男人在空军服役，政府不会无端干涉他的性隐私或他是否生孩子的决定。妇女在服役时受到’管制’；她对空军事业的追求要求她决定不生孩子”。 Brief for Petitioner at 54, 55, Struck (No. 72-178). ↩ ↩²
Struck, 409 U.S. 947, 947 (1972). ↩ ↩²
See Memorandum for the Respondents Suggesting Mootness (Dec. 1972). Struck (No. 72-178); Struck, 409 U.S. at 1071 (remanding for consideration of mootness) ↩ ↩²
See Reva Siegel, Reasoning from the Body: A Historical Perspective on Abortion Regulation and Questions of Equal Protection, 44 Stan. L. Rev. 261 (1992). ↩ ↩²
Cf. Weinberger v. Wiesenfeld, 420 U.S. 636, 653 (1975) (holding unconstitutional, as a violation of the equal protection principle, the denial to a widowed father of child-in-care social security benefits Congress had provided solely for widowed mothers 国会只向寡妇而不向鳏夫提供育儿社会保障福利的规定违反了平等保护原则，因此被认定违宪). ↩ ↩²
See Brief for Petitioner at 26, Struck (No. 72-178) (“[T]he regulation applied to petitioner establishes a suspect classification for which no compelling justification can be shown.”). ↩ ↩²
Id. (citing Bullock v. Carter, 405 U.S. 134, 144 (1972), as precedent for “an intermediate standard” under which the challenged classification would be “‘closely scrutinized’ “). ↩ ↩²
See Craig v. Boren, 429 U.S. 190, 197 (1976) (sex-based classification would not be sustained if merely rationally related to a permissible government objective; defender of classification would be required to show a substantial relationship to an important objective 基于性别的分类，如果只是与政府允许的目标有合理的联系，则不会得到支持，性别分类的辩护人需要证明与重要的目标有实质性的关系); see also Mississippi Univ. for Women v. Hogan, 458 U.S. 718, 724 (1982). ↩ ↩²
Sandra Day O’Connor, Portia’s Progress, 66 N.Y.U. L. Rev. 1546 (1991). ↩ ↩²
The turning-point case was Reed v. Reed, 404 U.S. 71 (1971). Reed involved a youth from Idaho who had committed suicide while in his father’s custody, the “mother’s preference” regarding custody having endured only while the boy was “of tender years.” The boy’s mother and father, long separated, had each applied to be the administrator of their son’s property. The Idaho court appointed the father under a state statute that provided: as between persons “equally entitled to administer, males must be preferred to females.” Id. at 73 (quoting Idaho Code § 15-314 (1948)). The Court unanimously ruled that the statute denied to the mother the equal protection of the laws guaranteed by the fourteenth amendment. 转折点案件是Reed诉Reed。该案涉及一个来自爱达荷州的青年，他在父亲监护时自杀，关于监护权的 “母亲偏好 “只在该男孩 “年幼 “时持续存在。男孩的母亲和父亲长期分居，各自申请成为其儿子的财产管理人。爱达荷州法院根据该州的一项法规判决父亲为管理人，该法规规定：在 “同样有权管理的人之间，男性必须优先于女性”。同上，第73页（引用爱达荷州法典第15-314条（1948年））。法院一致裁定，该法规剥夺了第十四修正案所保障的对母亲的平等法律保护。 ↩ ↩²
See Wendy W. Williams, Sex Discrimination: Closing the Law’s Gender Gap, in The Burger Years: Rights and Wrongs in the Supreme Court 1969-1986, at 109 (Herman Schwartz ed., 1987); see also Ruth Bader Ginsburg, The Burger Court’s Grapplings with Sex Discrimination, in The Burger Court: The Counter-Revolution that Wasn’t 132 (Vincent Blasi ed., 1983). ↩ ↩²
Frontiero v. Richardson, 411 U.S. 677, 688 (1973). ↩ ↩²
Stanton v. Stanton, 421 U.S. 7, 17 (1975). ↩ ↩²
Taylor v. Louisiana, 419 U.S. 522, 525 (1975) (invalidating law restricting service by women to volunteers); Duren v. Missouri, 439 U.S. 357, 360 (1979) (invalidating law allowing “any woman” to opt out of jury duty). ↩ ↩²
Weinberger v. Wiesenfeld, 420 U.S. 636, 639 (1975) (extending to widowers social security benefits Congress had provided for widows); Califano v. Goldfarb, 430 U.S. 199, 201-02 (1977) (same); Califano v. Westcott, 443 U.S. 76, 85 (1979) (extending to unemployed mothers public assistance benefits Congress had provided solely for unemployed fathers). ↩ ↩²
Wengler v. Druggists Mut. Ins. Co., 446 U.S. 142, 147 (1980). ↩ ↩²
Kirchberg v. Feenstra, 450 U.S. 455, 461 (1981). ↩ ↩²
Mississippi Univ. for Women v. Hogan, 458 U.S. 718, 723 (1982) ↩ ↩²
This expansion reflected a new reality: in the 1970s, for the first time in the nation’s history, the “average” woman in the United States was experiencing most of her adult years in a household not dominated by childcare requirements. That development, Columbia University professor of economics Eli Ginzberg observed, may be “the single most outstanding phenomenon of our century.”这种扩张反映了一个新的现实：在20世纪70年代，美国的 “普通 “妇女在国家历史上第一次在一个不受育儿要求支配的家庭中经历了她的大部分成年时期。哥伦比亚大学经济学教授伊莱-金兹伯格指出，这一发展可能是 “本世纪最突出的一个现象”。 Jean A. Briggs, How You Going to Get ‘Em Back in the Kitchen? (You Aren’t.), Forbes, Nov. 15, 1977, at 177 (quoting comment by Eli Ginzberg), ↩ ↩²
See, e.g., Brief for Appellant at 12-13, Reed v. Reed, 404 U.S. 71 (1971) (No. 70-4) (urging Court not to repeat “the mistake” of Plessy v. Ferguson, 163 U.S. 537 (1896)-which had upheld a state statute requiring railway companies to provide, inter alia, separate, but equal, accommodations for blacks and whites-and to rank sex-based classifications with the recognized suspect classifications). ↩ ↩²
Simone de Beauvoir, The Second Sex (1949). ↩ ↩²
For example, the male preference at issue in Reed v. Reed, described at note 107 supra, had been repealed, but not retroactively, before the Supreme Court heard the case; the categorical exemption of women from jury service had been largely abandoned in state systems by the time the Court heard Duren v. Missouri, described at note 111 supra. ↩ ↩²
See the Wiesenfeld and Goldfarb cases cited in note 112 supra. ↩ ↩²
See Califano v. Webster, 430 U.S. 313 (1977) (upholding classification, effective from 1956 to 1972, establishing more favorable social security benefit calculation for retired female workers than for retired male workers). ↩ ↩²
Ruth Bader Ginsburg, Some Thoughts on Benign Classification in the Context of Sex, 10 Conn. L. Rev. 813, 823 (1978). ↩ ↩²
See Ruth Bader Ginsburg, Some Thoughts on Judicial Authority to Repair Unconstitutional Legislation, 28 Clev. St. L. Rev. 301, 310-12 (1979). ↩ ↩²
Notably too, the equal rights or sex equality advocates of the 1970s urged no elaborate theory. They did argue that by enshrining and promoting the woman’s “natural” role as selfless homemaker, and correspondingly emphasizing the man’s role as provider, the state impeded both men and women from pursuit of the opportunities and styles of life that could enable them to break away from familiar stereotypes. The objective, however, was not “assimilationist” in the sense of accepting “man’s world” and asking only that self-regarding, economically advantaged women be allowed to enter that world and play by men’s rules. The endeavor was, instead, to remove artificial barriers to women’s aspiration and achievement; if women became political actors in numbers, it was thought, they could then exercise their will and their judgment to help make the world and the rules fit for all humankind. 值得注意的是，20世纪70年代的平等权利或性别平等倡导者们也没有敦促任何详细的理论。他们确实认为，通过奉行和促进妇女作为无私的家庭主妇的 “自然 “角色，并相应地强调男子作为供养者的角色，国家阻碍了男子和妇女追求能够使他们摆脱熟悉的定型观念的机会和生活方式。然而，其目的并不是接受 “男人的世界”，只要求允许自律的、有经济优势的妇女进入这个世界并按男人的规则行事的 “同化主义”。相反，我们的努力是为了消除对妇女的愿望和成就的人为障碍；如果妇女成为大量的政治行动者，人们认为她们就可以行使她们的意志和判断力，帮助使世界和规则适合全人类的发展。See Ruth Bader Ginsburg & Barbara Flagg, Some Reflections on the Feminist Legal Thought of the 1970s, 1989 U. Chi. Legal F. 9, 17-18; cf. Henna Hill Kay, The Future or Women Law Professors, 77 Iowa L. Rev. 5, 18 (1991) (“The future of women law professors is not to adapt to legal education by being ‘one of the boys,’ but to transform the enterprise so that all of its participants are equal members of the same team.女法学教授的未来不是通过成为’男孩之一’来适应法律教育，而是改造这项事业，使其所有参与者都是同一团队的平等成员。”). ↩ ↩²
Williams, supra note 108, at 123. This brand of review has been aptly called “judicial enforcement of constitutional accountability.” Guido Calabresi, The Supreme Court, 1990 Term-Foreword: Antidiscrimination and Constitutional Accountability (What the BorkBrennan Debate Ignores), 105 Harv. L. Rev. 80, 103-08 (1991). ↩ ↩²
410 U.S. 113 (1973). ↩ ↩²
Id. at 140; see also Ginsburg, supra note 80, at 385 & n.81. ↩ ↩²
See Herma Hill Kay, Equality and Difference: A Perspective on No-Fault Divorce and its Aftermath, 56 U. Cin. L. Rev. 1, 4-14, 26-55 (1987); see also Ginsburg, supra note 80, at 380 & n.36. ↩ ↩²
See Vincent Blasi, The Roofless Activism of the Burger Court, in The Burger Court: The Counter Revolution that Wasn’t 198, 212 (Vincent Blasi ed., 1983) (Roe “burst upon the constitutional scene with very little in the way of foreshadowing or preparation.”罗案 “在几乎没有预示或准备的情况下突然出现在宪法舞台上”。); Geoffrey C. Hazard, Jr., Rising Above Principle, 135 U. Pa. L Rev. 153, 166 (1986) (“By making such an extensive change, the Court [in Roe] foreclosed the usual opportunities for assimilation [and] feedback… that are afforded in a decisional process involving shorter and more cautious doctrinal steps.通过做出如此大的改变，法院[在Roe案中]排除了同化[和]反馈的通常机会……这些机会是在涉及较短和较谨慎的理论步骤的决策过程中所提供的。”). ↩ ↩²
See generally Louis Henkin, The Age of Rights 141-80 (1990). ↩ ↩²
Cf. Archibald Cox, Direct Action, Civil Disobedience, and the Constitution, in Civil Rights, the Constitution, and the Court 2, 22-23 (1967) (“[S]harp changes in the law depend partly upon the stimulus of protest.”). ↩ ↩²
See, e.g., Lochner v. New York, 198 U.S. 45 (1905) (state maximum hours regulation for bakery employees, covering men and women alike, held unconstitutional). But cf. Muller v. Oregon, 208 U.S. 412 (1908) (upholding maximum hours legislation for women only). ↩ ↩²
Cf. Calabresi, supra note 125, at 86 (typing bold court intervention as the “judicial supremacy” model of constitutional review). ↩ ↩²
347 U.S. 483 (1954) ↩ ↩²
See Plessy v. Ferguson, 163 U.S. 537, 540 (1896). ↩ ↩²
See Ginsburg & Flagg, supra note 124, at 18. ↩ ↩²
See United States v. Carolene Products Co., 304 U.S. 144, 152 n.4 (1938) (suggesting heightened judicial scrutiny of legislation disadvantageous to “discrete and insular minorities,” i.e., classifications tending “seriously to curtail the operation of those political processes ordinarily to be relied upon to protect minorities”建议加强对不利于 “独立和孤立的少数群体 “的立法的司法审查，即倾向于 “严重削弱通常用来保护少数群体的政治进程的运作 “的分类。); cf. Owen M. Fiss, Groups and the Equal Protection Clause, 5 Phil. & Pub. Aff. 107, 152 (1976) (stressing situation of blacks as “a numerical minority” and “their economic status, their position as the perpetual underclass”强调黑人作为 “数量上的少数 “和 “他们的经济地位，他们作为永久的下层社会的地位 “的情况。). ↩ ↩²
See Richard Kiuger, Simple Justice 256-84 (1976) (chronicling the eforts of Marshall and others in connection with Sipuel v. Board of Regents of the Univ. of Oklahoma, 332 U.S. 631 (1948); Shelley v. Kraemer, 334 U.S. 1 (1948); Sweatt v. Painter, 339 U.S. 629 (1950); and McLaurin v. Oklahoma St. Regents for Higher Educ., 339 U.S. 637 (1950)); Jack Greenberg, Litigation for Social Change: Methods, Limits and Role in Democracy, 29 The Record of the Ass’n of the Bar of the City of New York 320, 327-34 (1974) (discussing the “litigation campaign” preceding Brown) ↩ ↩²
Compare The Orison S. Marden Lecture in Honor of Justice Thurgood Marshall, 47 The Record of the Ass’n of the Bar of the City of New York 227, 254 (1992) (comments of Constance Baker Motley) (“[No civil action was ever initiated under [Marshall’s] leadership unless it was part of an overall strategy …. No major legal thrust was made without months if not years of careful legal research and planning such as occurred in the early voting cases, teacher salary cases, restrictive covenant cases, interstate travel cases as well as the school desegregation cases.[在[马歇尔]的领导下，除非是整体战略的一部分，否则从未发起过民事诉讼….。没有经过数月甚至数年的仔细法律研究和规划，就不会有重大的法律行动，例如在早期投票案、教师工资案、限制性契约案、州际旅行案以及学校取消种族隔离案中发生的情况。“) with Blasi, supra note 129, at 212 (Roe “could not plausibly [be] justiftied] .. as the working out of a theme implicit in several previous decisions.”罗案 “不能合理地[被]证明是有道理的。］ ……作为对以前几项决定中隐含的主题的落实。”). ↩ ↩²
The Court relied on the psychological harm, empirically documented, that segregated schools caused black children. See 347 U.S. at 493-94 & 494 n.1l. ↩ ↩²
Pub. L. No. 88-352, 78 Stat. 241 (codified as amended at 28 U.S.C. § 1447, 42 U.S.C. §§ 1971, 2000a-2000h-6 (1988 & Supp. 111990)). ↩ ↩²
388 U.S. 1 (1967). ↩ ↩²
The legislative reapportionment cases of the early 1960s present a second notable instance of the Court confronting blocked political processes. Before the 1960s, many state legislatures arranged their districts in ways that diluted the voting power of urban voters. Under precedent then in place, legal objections to these malapportioned schemes were not justiciable in federal court. 20世纪60年代初的关于立法机构重新分配的案件是法院面对受阻的政治进程的第二个值得注意的例子。在20世纪60年代之前，许多州的立法机构以稀释城市选民投票权的方式安排其选区。根据当时的先例，对这些分配不当的方案提出的法律异议不能在联邦法院审理。 See Colegrove v. Green, 328 U.S. 549 (1946). In Baker v. Carr, 369 U.S. 186 (1962), this changed: the Supreme Court declared challenges to malapportioned schemes justiciable and thereby opened the way for their invalidation by federal court decree. As one leading commentator on the reapportionment cases observed:这种情况发生了变化：最高法院宣布，对分配不当的计划提出的挑战是可以审理的，从而为通过联邦法院的法令宣布其无效开辟了道路。正如一位关于重新分配案件的主要评论员所言： The ultimate rationale to be given for Baker v. Carr and its numerous progeny is that when political avenues for redressing political problems become dead-end streets, some judicial intervention in the politics of the people may be essential in order to hare any effective politics. In Tennessee, [for example,] at the time its legislative composition was challenged in Baker, there was a history of several years of unsuccessful state court litigation and unsuccessful efforts for corrective legislation.贝克诉卡尔案及其众多后裔的最终理由是，当解决政治问题的政治途径成为死胡同时，为了获得任何有效的政治，对人民的政治进行一些司法干预可能是必不可少的。在田纳西州，[例如，]在贝克案中挑战其立法构成时，有几年州法院诉讼不成功的历史，以及纠正立法不成功的努力。Robert G. Dixon, Jr., Democratic Representation: Reapportionment in Law and Politics 8 (1968) (emphasis in original). ↩ ↩²
The decision in the legend is Worcester v. Georgia, 31 U.S. (6 Pet.) 515 (1832). ↩ ↩²
Learned Hand, The Bill of Rights 73 (1958). ↩ ↩²
Cf. Archibald Cox, The Role of the Supreme Court: Judicial Activism or Self-Restraint?, 47 Md. L. Rev. 118, 124-25 (1987) (though the “style of interpretation” of Chief Justice Marshall’s Court “was active and creative,” that Court, “[i]n expanding national power[,] … was moving in step with the dominant trend in the political branches”尽管首席大法官马歇尔的法院的 “解释风格 “是积极和创造性的，”但该法院 “在扩大国家权力方面[，]……与政治部门的主导趋势保持同步”). ↩ ↩²
Planned Parenthood v. Casey, 112 S. Ct. 2791 (1992). ↩ ↩²
Three years before its Planned Parenthood decision, the Court had come close to overruling Roe. See Webster v. Reproductive Health Servs., 492 U.S. 490 (1989), discussed in Sylvia A. Law, Abortion Compromise-Inevitable and Impossible, 1992 U. Ill. L. Rev. 921, 923-26. ↩ ↩²
The hostile reaction to Roe has hit most heavily women who are most vulnerable-“the poor, the unsophisticated, the young, and women who live in rural areas.”对罗案的敌对反应对最脆弱的妇女打击最大——”穷人、不成熟的人、年轻人和生活在农村地区的妇女”。 Law, supra note 148, at 931; see also Ginsburg, supra note 80, at 383-85. ↩ ↩²
Indicative of the changed political climate, President Clinton, on his second full day in office, January 22, 1993, signed five Memoranda terminating abortion-related restraints imposed in the 1980s. 克林顿总统在他上任的第二天，即1993年1月22日，签署了五份备忘录，终止了1980年代实施的与堕胎有关的限制措施，这表明了政治气候的变化。 See 29 Weekly Comp. Pres. Doc. 87-89 (Jan. 25, 1993) (Memorandum for the Secretary of Health and Human Services, on Federal Funding of Fetal Tissue Transplantation Research, Memorandum for the Secretary of Health and Human Services, on the Title X [of the Public Health Services Act] “Gag Rule”; Memorandum for the Acting Administrator of the Agency for International Development, on AID Family Planning Grants/Mexico City Policy; Memorandum for the Secretary of Defense, on Privately Funded Abortions at Military Hospitals; Memorandum for the Secretary of Health and Human Services, on Importation of RU-486). Cf. Law, supra note 148, at 931-32 (setting out opposing assessments and commenting that “[o]nly time will tell”). ↩ ↩²
Professor Gerald Gunther Speaks at Investiture of Judge Ruth Ginsburg in Washington, D.C., The Colum. Law Alumni Observer, Dec. 31, 1980, at 8. ↩ ↩²

公平的多样性

Fri, 19 Jun 2020 00:00:00 +0000

考虑一个预测犯人是否会再次犯罪的模型，假如我们追踪被打分的嫌犯在之后两年的表现发现：

1.那些没有再犯的白人，24%的人当初被认为会再犯，而没有再犯的黑人里45%的人当初被认为会再犯；在真的再犯的白人里，48%的人当初被认为不会再犯，在真的再犯的黑人里，28%的人当初被认为不会再犯。你觉得这个模型公平吗？如果不公平，歧视了谁？

2.如果预测会再犯的白人里59%真的再犯了，预测会再犯的黑人里63%再犯；预测不会再犯的白人里71%没有再犯，预测不会再犯的黑人里65%没有再犯。你觉得这模型公平吗？如果不公平，歧视了谁？

没什么意外的话我猜你的答案是1歧视了黑人，2还大致公平。我不在乎你的答案是否跟我一样，我在乎的是它是否跟你自己一样，因为这两个描述说的是同一个模型。

怎么会这样？让我们画几个图看看。（图重绘自 LessWrong）

对每个正方形来说，方块1代表再犯并且模型也预测会再犯，方块2代表没有再犯但模型预测会再犯，3代表再犯但模型预测不会再犯，4代表没有再犯模型也预测不会再犯。左边的正方形代表白人，右边代表黑人，唯一的区别是出于 whatever 原因，黑人的再犯率更高，这体现在右边的方块1和3更宽。假设正方形的边长为1，则描述1里没有再犯但当初被模型预测会再犯的比例就是方块2的高度（假阳性率），再犯但当初模型预测不会再犯的比例是方块3的高度（假阴性率）。描述2里被预测会再犯的人真的再犯的比例则是方块1的面积除以方块1加2的面积（精准度/precision/positive predictive value）。现在你认为描述1里的模型是有偏见的，也就是说你认为一个公平的模型应该在不同群体上有着同样的假阳性率和假阴性率，也就是两个正方形的方块1和2应该有着相同的高度，也就是我图里画的那样。但你同时认为描述2是公平的，也就是你希望精准度在两个群体上也是相等的，能做到吗？很难，除非你的模型是100%准确之类的极端情况，否则在两个基础犯罪率不一样的群体上，假阳性率、假阳性率和精准度是不可能同时做到公平的。

在机器学习公平性研究里，假阳性率和假阴性率的公平被称作几率公平（equalized odds），假如你只关心假阳性率，也就是只在乎模型对那些无辜的人的预测，这种特例叫做机会公平（equalized opportunity），精准度的公平则叫做预测公平（predictive parity）。开头的数字并不是我编的，它们来自法院真实使用的软件 COMPAS。第一条描述的是 ProPublica 对 COMPAS 的质疑，第二条则来自 COMPAS 的自我辩护。我想说的不是“COMPAS 是对的，预测公平才是真的公平，法院没有歧视”，而是假如你没有去论证这种情境下哪个公平标准更可欲，你就不应该直接从数字上的不相等直接认定模型存在歧视。

随着 ML Fairness 这几年的快速发展，还有其他一些公平性指标也被提了出来。让我们再看一个常见的例子：银行发放贷款。（以下图片来自这篇文章，深色点代表会还贷的申请者，浅色点代表会违约的人）

机会公平：对不会违约的申请者来说，批准率是一样的。

几率公平：对不会违约的申请者来说，批准率是一样的，而且对会违约的申请者来说，批准率也是一样的。

群体公平（demographic parity）：对所有申请者来说，批准率是一样的。
预测公平：对所有申请被批准的人来说（虚线以上的点），不同群体的违约率是一样的（虚线以上深色点和浅色点的比例一样）。那篇文章没有这个情况的图，我也懒得做，自己脑补吧。
初心公平（unawareness）：我既不歧视 A，也不歧视 B，我甚至都不在乎他们是不是人，我只在乎钱。我的目标函数里就没有种族，怎么样能赚最多的钱我就怎么做。

如果还弄不明白，玩一下 Google Research 做的这个可视化应该也就清楚了（那里的命名有点不符合主流，它的 max profit 对应这里的 unawareness，它的 group unaware 对应这里的 demographic parity）。

我们可以从这例子中学到什么？至少有这么几点吧。

1.可以证明，几率公平、群体公平和预测公平是两两互斥的，除了模型100%准确这种不现实的情况，实现其中一个则另外两个必定无法实现。

2.想想这几个目标函数，按公平性排个序；然后再看看它们产生的结果，你再排个序；然后你对比下你的两次排序，它们一样吗？如果不一样，你脑子里在想些什么？

3.即使两个群体的平均信用一样，合格/不合格申请者的平均信用一样，申请者数量一样，被批准的人里两个群体的数量依然未必一样，除非你采用的是群体公平策略。换句话说，除非你认为群体公平是唯一正确的标准，否则两个群体平均信用一样+他们贷款批准率不一样未必意味着歧视。同样的道理适用于“男女生在中学的时候平均理科分数差不多，而 MIT 理科教授却男女比例失衡，所以这一定是性别歧视”。分布很重要。

4.无论你采用哪种标准并努力去实现，都依然有可能被认为是歧视。在黑人区部署和白人区一样的警力，那没有考虑到历史造成的不平等；部署更多的警力，那是选择性执法；部署更少的警力，那是只知道保护白人社区。这种动辄得咎不动也得咎的情况和反垄断法非常相似，回想一下科斯说过的话：“我被反垄断法烦透了。假如价格涨了，法官就说是‘垄断定价’；价格跌了，就说是‘掠夺定价’；价格不变，就说是‘勾结定价’。”

最后我们考虑一个思想实验，假设现在有一个 AI，它负责在路上拦车查毒品，它一心只想尽可能地多查到毒品（初心公平），已知甲组携带毒品的概率比乙组高，请问它该怎么做？正确的做法是每次看到甲组的车都优先拦下来，因为这样最有可能发现毒品。随着甲组被针对性执法，只要他们不傻，他们就会少带毒品，但只要他们携带毒品的概率依然大于乙组，AI 就会继续针对他们，直到两组携带率一样。所以假如最终两组的携带率一样，那就说明这是个一心为公毫无私念的好 AI，否则携带率低——注意是低——的那组被歧视了，AI 应该把更多的精力花在携带率高的那组上。

与之前的例子不同，这里初心公平自动实现了预测公平（不妨称之为“后效公平”），原因也很简单，我们考虑了激励。可以理解，这种考虑激励的分析在新兴的 ML Fairness 社区还没有多少人谈论（我没准是第一个把它们联系到一块的？），但似乎没有什么明显的理由认为它不值得考虑。Again，我不是说后效公平才是真正的公平，而是你不能因为两个群体携带毒品的概率一样但被拦下来的概率不一样就觉得这说明了系统性歧视，除非你花时间去论证了后效公平为什么不成立。不然它除了说明你思考的不够全面，什么也说明不了。

有些人敬爱的林三土老师

海德沙龙翻译组

Fri, 12 Jan 2018 00:00:00 +0000

利用假期帮海德沙龙翻译组搭了个网站 https://translations.headsalon.org，不为什么，就是比较闲。

评论系统几经权衡还是决定和本博客一样用 Disqus，墙内无法访问，暂时只能靠大家自力更生，抱歉。

有了谢益辉大大的 blogdown，建站其实已经很简单了，所以大部分时间都花在写爬虫和正则表达式上了~~（以及实在写不出来的时候手动一个个修改）~~，数据清洗确实是数据分析中最累人的那一步。。

当然这是值得的，作为中文世界品味最高的文章翻译组，他们理应得到更多的阅读。

大家新年快乐，多吸海德。

我的师承

Sat, 05 Aug 2017 00:00:00 +0000

这是一个无趣的世界，但有趣在混沌中存在。——王小波

王小波写过一篇同名文章，谈论了那些在文学上影响过他的人。我虽然还没到回首人生的时候（还是已经到了？…），但为了给偶像打广告，也还是决定说上那么几句。

我的知识可以大略分成两部分，一部分是数理化，兴趣源于初二时偶然读到的相对论科普，应该是湖南科技出版社第一推动丛书中的某一本。对一个上初中的小p孩来说，这简直是打开了新世界的大门。那几年我把能找到的第一推动系列都读了一遍。大部分当然看不懂，但能勉强理解的一小部分已经足以让我跟小伙伴们吹嘘半天了。我还记得在读《皇帝新脑》时读到康托的对角线删除法，那可能是我第一次遇上一个我能看懂但完全不明白他怎么想到的证明，仿佛天上掉下来的一般，深深地震撼了我。那阵子我逮着机会就会问那些对数学也比较感兴趣的朋友：“你觉得自然数多还是有理数多？”

“有理数多。”

“不对，一样多。那有理数多还是实数多？”

“一样多。”

“不对，实数多。”

“……”

总之，那确实是一段快乐的时光……之后和别人大概也没什么两样，就是按部就班地从微积分一路学下去，毕竟数学技巧是这些学科的基本门槛。高考报志愿我也就直接填了数理化，生物我都看不上，那时候的我非常同意霍金的一句话，他上大学时父亲希望他去学医，但他也不喜欢生物学，觉得那个学科不够抽象，“过于叙述性并且不够基础。”这种看法现在看来当然很幼稚，但其实也保护了我，那时候的我还没有摸到进化论这条线索，所以按卢瑟福的分法，生物学对我来说就像集邮一样混乱难解，假如冒然闯进去，我也许就一头栽进经典里再也出不来了。就像杜布赞斯基所说的，"Nothing in biology makes sense except in the light of evolution"，我最终摸到这条线索已经是大学快毕业的时候了，而这都要归功于一个人，他就是我的偶像——辉格。

因为高中时已经自学完了大部分本科化学，我的大学基本都宅在宿舍睡觉上网，那时候我其实处于一种对”理解“极度渴求的状态——到底该怎么看这世界才能有点深入的理解？有一两年我几乎每天都会把网上能找到的时评全读一遍（联合早报、凤凰、星岛环球网……），一开始还有点收获，但渐渐地觉得这些时评家也就是些车轱辘话反复说，对这世界的理解似乎还是差了那么一层。但混的久了信息毕竟广了一些，知道了牛博网，却一直没上去，直到某天又闲得无聊，于是在浏览器里敲下了“牛博”……又一次，新世界的大门在我面前徐徐打开，那里的每个博客我都喜欢的不行，连那些摄影博客我都看得津津有味（我一直都是个对艺术基本无感的人）。我兴奋地一个个博客读过去，完全停不下来，然后我看到一个博客，名字叫，唔，辉格。当时的情景大概是这样的：辉格？这词听着怪怪的，好像有什么深意，搜了一下是英国历史上的一个党，看了半天维基也还是不明觉厉，牛博上其他奇奇怪怪的名字也很多，但我至少都能get到点，而这个博客却从名字开始就给我一种怪怪的感觉。点进去看发现和其他牛博博客不一样，我居然基本都看不懂。我看不懂的东西很多，高深的数学我知道自己看不懂，但全是大白话好像在说着什么我却看不懂还是第一次，而且这个在牛博首页上，应该是很牛逼的啊，但为啥我都看不懂呢？不甘心的我不断地挖掘，然后就看到了辉格关于乳房起源的文章，那时的我依然还没摸到进化论，但却对自己的科学素养颇有自信，看了之后完全不得要领，觉得这不就是个民科吗，老罗这么牛逼怎么会请个这样的人来？但那时我还处于被牛博震慑住的状态，所以满腹狐疑地暂时保留了态度。

之后的时光就在每天刷着牛博中度过，大概半年到一年后吧，我的兴趣渐渐从时政类博客（又读腻了）转到了经济学类（现在的铅笔经济学社），这是我第一次接触奥派，自由市场的思想给了我巨大的快感，很多事情终于都清晰了起来，然后可能又是半年到一年，经济学类我也读腻了，国奥们翻来覆去就是那些东西，然后某天百无聊赖的我又在牛博首页上点来点去，我又点进了辉格的博客……诶，这些经济学的文章我居然可以看懂了，我去，怎么这么深刻，这不就是我一直在寻找的比铅笔社更进一层的那种文章吗？再看一篇，怎么还是那么有道理，再下一篇……很快我就明白辉格为什么会在牛博首页了，并且成了他的粉丝，但他的许多文章我还是只能囫囵吞枣，在之后反复阅读理解他博客的过程中，不知不觉，他把我带入了经济学、心理学、进化论、法理学、人类学、科学哲学……的领域，而这些就是我知识体系中的第二部分。与数理化不同，它们的门槛不是数学，而是品味。这里许多学科的共识程度不高，学派林立，假如你走错路那可能就再也出不来了（想象一下如果一个研究哲学的人一上来选择了新马克思主义。。）因此在学习的早期能有高手指点方向就非常重要。在那之前我都没能摸到方向，幸好我品味尚可，所以对它们就没有产生多大的兴趣，而这一次，我终于摸到了线索，开始明白诸如生物学这些学科也可以那么的有道理。之后就是数年的阅读了，有了线索之后阅读思考都轻松愉快了许多，我也一步步变成了今天这样……

但是辉格的名气显然不大，对于第一次知道他的人，很自然地会问，他真的有那么厉害吗？你在网上搜的话也会发现评价两级化，有人不屑一顾，觉得他民科（跟当年的我一样），这自然会加深你的疑虑。对此我打算举几个例子，国内的知识分子大概可以分为两类，人文社科类和科学青年。在人文公知里，老罗应该算很有名望了（至少在他做手机前……），他显然并不欣赏辉格，最后还把辉格从牛博赶走了（于是辉格创立了独立博客海德沙龙），在说明文章里他说辉格本来就是朋友推荐才请来的，他算不上喜欢（大意如此，牛博已逝，这篇文章我没能找到备份）。但他非常推崇黄章晋，在某次演讲中甚至说过黄章晋是当代的胡适，你们之所以不知道他只是因为他还没死（具体是哪个演讲我也忘了，欢迎提供链接）。可是黄章晋却是辉格的多年粉丝，这几年辉格能多少有了那么点名气与教官和他大象公会不遗余力的推广是分不开的。科学青年方面看不起辉格的就更多了，一个个都觉得他民科，但他们基本上都很钦佩果壳主编 Ent，可假如你长期关注 Ent 的话，你就会知道他对辉格是非常尊重的（这毫不丢人，我也是这样的）。这种偶像的偶像我却觉得傻逼的现象并不常见，也说明了许多东西，有些人确实是你越牛逼你才会觉得他越牛逼的。

在我们这些铁杆粉丝看来，辉格是中文世界罕见的高峰，他是足以和世界一流思想家坐下来谈论问题的人，而且虽然并非本意，他精准的文字也为现代汉语这门已经有点脱节于现代文明的语言带来了新的可能。辉格之前写过两本书，今年刚又出了一本，这次在老粉的安排下，还出了个爱的供养版，让我们这些多年受益于辉总的人有了个打钱的借口和机会。我写这篇文章主要就是自带干粮来推荐大家购买的，不过我得提醒一句，阅读这样一个人是有风险的，这些年我越来越难以与没有熟读海德三百篇的人交流了，毕竟，当你看着一个人自信满满地犯着你五六年前就想明白的错误时，你还能有多少交流的欲望？为此某些可能深有同感的粉丝做了个海德刷机图，帮助人们降低入门难度。不过图片点击起来不是很方便，所以我这里重制了一下。

最后，看在我都做到这样的份上，下次见面要是还是没啥话说就别再怪我了吧。。

我们没有敌人？

Thu, 20 Jul 2017 00:00:00 +0000

上周有个人走了，大家应该都知道了。

他的事迹无疑是令人敬佩的，他的事业也无疑是失败了。

独自原谅换来的是瘐死狱中，而外面的世界早已爬满了小粉红。

大家都赞美甘地。爱因斯坦说：“后世的子孙也许很难相信，历史上竟有过这样一个人。”

即使出了专业领域，爱神说话也还是挺准确的，除了要把“一个”改成“一群”之外。

曹晚植，朝鲜甘地，因为和苏联不对付，老大哥把他抓了起来，没几年就死因不明了，历史记住他了吗？

“你有狼牙棒，我有天灵盖”的腐儒历史上多的是，又成过什么事呢？

甘地常有，而英国人不常有。

我们是有敌人的。

饼图的妙用

Wed, 19 Jul 2017 00:00:00 +0000

在数据可视化领域，饼图被普遍认为是个非常糟糕的可视化方法，只有在一些特定情况下你才应该使用它。

但是谁说糟糕的东西就一定没用了呢？

上图是美国农业部1984年推出的饮食建议 food wheel。应该不用我说了，它简直就是数据可视化灾难，你从图上能看出每天该多吃谷物还是肉吗？

当然我们农业部是对美国人民负责滴，我们都是请最专业的科学家来为我们写膳食指南。同行评议！学科共识！你们的身体健康就放心地交给政府吧！于是在1992年，农业部根据最新的研究进展，把饼图换成了金字塔。

这就好看多了嘛，至少能看出每天吃最多的得是谷物。这种改进大家肯定都应该赞成的对不对？当然不对啊，肉类行业的人就不乐意了嘛，原来我可以躲在饼图里浑水摸鱼，大家看起来都差不多，现在你们把谷物放在最底层，~~会不会给人一种钦定的感觉~~所有人不都一眼就看出来应该多吃饭少吃肉了吗？于是他们进行了大规模的游说活动，当然没有收到任何效果，谁让他们的对手是科学呢？当时的场景大概是这样的（纯属虚构）：

什么？你想把金字塔改回饼图？不可能，饼图的可视化效果太糟糕了，我们要为老百姓着想，让他们能更方便地理解最新的学术成果，我们要尊重科学……哦你不改图了？你想把 “decrease consumption of meat and increase consumption of poultry and fish”改成“decrease consumption of animal fat, and choose meats, poultry, and fish which will reduce saturated fat intake”？嗯这样一改感觉更客观了呢，而且原来的说法语气确实有点生硬，万一对肉类行业造成冲击，容易扰乱经济秩序，我作为参议员，得负起对人民的责任……

然后到了2005年，food pyramid 又被改成了 MyPyramid。

咦……嗯……这个可视化效果嘛……反正农业部是不会骗我们的，他们一定是根据最新的学科共识才决定改成这样的！

然后到了2011年，

唔…………

《美国居民膳食指南》是营养学科普作家最爱引用的资料，看着这本业内权威的书呈现核心内容的方式从饼图到金字塔到后现代金字塔再到后现代饼(?)图，我不禁陷入了婶婶的思考，“幸好我没去学营养学啊，不然以我的脑袋肯定是跟不上进步这么迅速的学科共识的。”

然后我忍不住又继续想，还有什么学科也是这样拿着政府文件当个宝的吗？好像没……哦哦，IPCC 嘛。这样一来很多事情好像都说得通了呢。

好吧，我严肃一点。大部分科学家都认为政府应该资助科学研究，我是不同意的，我就不提政府对科学的腐蚀作用了，你自己再看一眼上面的那些图和链接就好了。让我们听听科学家的理由吧，最常见的就两个：第一，科学太美好，是人类智力的最高成就。然后呢？没了……大概渐渐意识到”因为我很有好奇心所以纳税人得养我“确实太不要脸，现在大部分人都会补上第二个理由，科学发现的过程中会产生许多有用的副产品，阿波罗计划不就给我们带来了许多新材料吗？这点车轱辘话在明星物理学家 Brian Cox 的 TED 演讲里算是被发挥到了极致。

让我们来理下他的套路，先给你看一大堆太空照片，让你领略科学之美（确实很美，这个没毛病）；然后说阿波罗计划是有经济效益的，不亏钱（这是大科学的代表作，不提不好意思）；至此两点理由都说完了，剩下的就是情怀了。首先作为萨根生态位的继承者，你最后肯定得念一段优美的暗淡蓝点（确实很美，这个也没毛病），然后再以“如果在20世纪20年代做出这样的结论，人们就不会发现青霉素。如果是在19世纪90年代，人们就不会发明晶体管。”这样的话作为结束，唤起观众的危机感，最后大家起立鼓掌，又一个标准的TED演讲。

为了不显得太吹毛求疵，我们就先假装不知道青霉素的发现部分是由洛克菲勒赞助的吧，也假装晶体管的发明真的是在19世纪，但是 Brian 啊，你一面引用着暗淡蓝点，一面却似乎忘了——或者假装忘了，你的偶像卡尔·萨根在同名著作中是这样说的：

费曼那就更直接了，

如果你好奇为什么这么有名的科学家会在大庭广众犯些这么低级的错误，那说明你想错了问题。政府资金分配是个公共政策问题，不是科学问题，所以科学家表现差是正常的，这事还是得听经济学家的。

经济学101对此的标准答案是科学有正的外部性，市场会供给不足，所以得靠政府介入。这论证一看就比 Brian 不知道高到哪里去了，不过可惜，还是有问题的。这个论证的问题在于把不完美的市场拿去和假定完美的政府比，比不过不是很正常吗？在几十年前你还可以为这种观点辩护：“我们知道政府也不可能完美，但经济学家还不知道怎么用经济学理论研究政府行为，所以我们可以假装没看到，毕竟我们经济学家只在路灯下找钥匙。”但随着公共选择学派的兴起，这理由已经不好使了，于是现在我们不得不比较不完美的市场和不完美的政府，这就只能靠定量分析了。为了避免陷入大量的技术细节把这篇文章又写成一万字，我还是不展开了，但我觉得说双方都没有压倒性的证据应该还是没问题的。

然后我想说就算政府真的完美的能弥补一点市场的不足吧，那效果有多大呢？根据维基，2013年美国全年投入研发的资金是4561亿，其中3225亿来自私人部门的，占71%。就算政府资金全部撤出后私人只能补上三分一吧，那总额也还能有现在的80%。少那20%很严重吗？当然，因为政府经费主要集中在学院，学术圈受的冲击会大于20%，不过现在一年全世界诞生上百万篇的新论文，就算我们以后每年只能产出现在60%的论文，好像也不是很严重吧？如果你觉得很严重，那我也没啥办法（因为这就涉及定量了），只好笑笑然后给你讲个段子，

Nature 杂志的体积增长速度已经超过了光速，但是这并不违反相对论，因为它不携带任何信息。

最后，作为一个深爱着科学的人，我不喜欢这些政府资助的鼓吹者还有情感上的理由。如果你真的觉得一个东西美好，你怎么会觉得除了政府没人会愿意赞助她？如果你真的觉得一个东西纯洁，你怎么会希望官僚的脏手去碰她？如果你是个自爱的人，你怎么会把自己搞得跟希腊公务员一个德行？

科学家上街抗议分到的税金不够多

心理学的危机

Sun, 16 Jul 2017 00:00:00 +0000

NBA 总决赛刚结束不久，今年最幸福的肯定就是勇士球迷了。宇宙勇的球迷一定很关心一个东西，因为它决定着你看到的是海啸组合还是铁花兄弟——它就是手感。

但是。。手感这东西真的存在吗？

你肯定会说这问题怎么这么蠢，打过球的谁没有篮筐大的像海洋的时候？别急。。让我先介绍下这问题的背景。

缘起

n年以前，斯金纳在一系列经典的鸽子实验中展示了如何通过强化学习让鸽子学会各种习惯，比如下面的视频展示了斯金纳是怎么在一分钟内把鸽子调教到会原地转圈的。

但斯金纳的手段远不止此，通过改变不同的强化方式，他发现了一个有趣的现象。假如你在鸽子做出某个动作后让食物掉出来，那鸽子很快就会学会这个动作，就像上面学会原地转圈那样；如果鸽子不管做什么你都不放食物出来，那鸽子很快就啥都不做了。这些都很合理，但当你放出食物的时机与鸽子的动作毫无关系时（比如固定每隔 20 秒掉食物），那鸽子会展现出各种奇怪的强化行为——有的会不断地逆时针转圈、有的会向左边扭一下头再去看有没有吃的、有的会使劲地晃动脑袋……这些行为也很好解释，当鸽子做出某个动作时食物碰巧掉出来了，鸽子就会认为它一定做对了什么，从而重复这个动作，而过一会食物果然又掉出来了……最终鸽子就这样学会了各种奇奇怪怪的动作，斯金纳把这称为“鸽子的迷信”¹。

你说鸽子笨嘛，脑袋那么小，难怪容易迷信。那请你想想赌场里赌徒们千奇百怪的禁忌、想想星座、想想本命年的红内裤……我们这个物种的许多成员表现又好到哪里去呢？事实上我们和鸽子一样，非常擅长从随机中看出“规律” (Apophenia)——如果不是更擅长的话^2-5。iTunes早期的随机播放是完全随机的，结果收到用户投诉，说咋这么不随机？乔布斯不得不采用了更复杂、完全不随机的算法，才让用户满意。用乔布斯自己的话说:“We’ve actually added Smart Shuffle to make it less random. But it seems more random.”

这种根深蒂固的本能自然有着进化上的道理，在一个没规律的地方看出规律你可能也就多晃几下脑袋，但假如食物真的是因为晃脑袋掉出来的你却没发现，那你就得饿死了。既然错过规律比找错规律代价要大得多，那我们肯定宁愿多找出些“规律”。这种认知偏见自然没有逃过心理学家的眼睛，于是有一天几个心理学家琢磨着：“人们认为的手热会不会也是纯粹的随机现象呢？”

手热迷思

1985 年 Gilovich、Vallone 和 Tversky 对此进行了研究，结果很惊人，他们发现手热效应是不存在的⁶。这篇论文与心理学界对人类非理性行为的认知完全相符，所以被迅速接受并成为经典。因为在圈内广为人知，它常常直接被称作 GVT 论文——和爱因斯坦的 EPR 论文一样，以三位作者名字的首字母命名。之后不少人重复了这个研究，大部分都得出了类似的结论——手热只是幻觉。这话题果壳网其实介绍过，但并不让人满意，因为文章最后一笔带过的 Joshua Miller 的最新研究恰恰是整个话题峰回路转的最高潮，所以我重新写一下。

GVT 论文里进行了这么几项研究：

调查球迷的态度，大部分人认为手热是存在的（当然）；
分析了 1980-81 赛季76人队的比赛数据，分析方法是这样的：统计出刚刚连续命中一个（或数个）篮和刚刚连续投失一个（或数个）篮情况下的命中率，对比后发现刚刚投中过球并没有让你变得更准，事实上命中率反而更低了；
把每个76人球员的赛季投篮数据分成四个四个的一段，统计热得发烫（四中三或全中）、普通（四中二）和冰凉（四中一或全失）的出现频率，假如手感确实会忽冷忽热，那它们的分布与用平均投篮命中率算出来的随机分布应该有区别，但并没有；
采访了76人队队员，绝大多数都相信手热现象，并认为自己有过手热的经历（当然）；
以上的分析有个显而易见的问题，就是干扰项太多：手热的时候对方可能会对你防守更严，你可能会因为自信而浪投……所以 GVT 又分析了罚球数据，第一罚命中会让你第二罚更准吗？他们分析了 1980-82 两个赛季凯尔特人的罚球数据，没有发现显著性；
然后又找来了康奈尔大学一些校队院队球员进行受控实验，让他们在无对抗情况下各投篮 100 次，然后进行了和 2、3 中一样的分析，依然无果；
最后他们研究了对手热的感知。在校队队员每次投篮前，他们让投篮者和队友预测球会不会进，结果自我预测和投篮命中的相关性只有 0.02，队友预测和命中的相关性是 0.04。

不难看出整个研究做的挺细致，质量上乘，结论也是理聪最爱的那种，所以很快就流传开来。古尔德写过，卡尼曼也写过，科学家们乐此不疲地教育着头脑简单的运动员和大众：“都是幻觉，都是幻觉！”然后几十年过去了……

“手热迷思”迷思

第一个转折来自于对 GVT 论文效力的研究：假如我用电脑模拟手热效应存在的投篮数据，然后用 GVT 的方法去分析，你猜怎么着？大多数情况下是检测不出手热的⁷。也就是说不管手热存不存在，他们的方法都检测不出来——统计效力 (power) 不足。这一下就推翻了 GVT 的大部分论证。不过这只说明了他们的方法不对，结论也许碰巧还是对的呢？

2015 年 Joshua Miller 为 GVT 送上了最后一击，他们发现 GVT （和许多后续研究）都犯下了一个极其微妙的数学错误，在纠正后，这些数据不仅没有否定手热现象，反而强烈地支持它⁸。除了重新分析数据，他们还指出了 GVT 在受控实验部分设计上的一些不足，并进行了改进的实验，发现了显著的效应⁹。

那这个微妙的错误是什么呢？我们先来做几道数学题。

你拿一个硬币，扔啊扔，然后连扔出了三个正面，请问你下一次扔还是正面的概率是多少？

你说这太简单了，独立事件嘛，赌徒谬误嘛，哪怕我连扔出一百次正面下一次得到正面的概率都还是0.5。

嗯没错，那现在你再拿一个硬币，扔一百次，每当连扔出三个正面时你就把下一次投掷的结果记下来，最后你的结果里正反面各占多少？

这个问题的答案——惊人地——不是 0.5。为了证明它真的不是 0.5，我们看一下最简单的情况，总共扔三次，然后统计连续两次正面出现的概率。

序列	HX的次数	HH的次数	HH的概率
HHH	2	2	1
HHT	2	1	0.5
HTH	1	0	0
HTT	1	0	0
THH	1	1	1
THT	1	0	0
TTH	0	0	-
TTT	0	0	-
均值	1	0.5	5/12

你看，预期概率确实不是 0.5，而且可以证明它一定小于 0.5，也就是说用这种方法计算的话，连续多次正面后确实有更大的概率出现反面。我相信这简单的数数你肯定看懂了，但你的直觉未必能跟上，所以我再多说几句，看看能不能帮你从直觉上也理解它。

首先，直觉是没错的，扔出多少次正面都不会影响后续的结果，这体现在第三列的和一定是第二列和的一半，但第四列的和却一定小于第二列的一半。区别在哪呢？在于 $E(X)/E(Y) \neq E(X/Y) $。$E(HH)/E(HX)$ 一定等于 0.5，但 $E(HH/HX) $ 就不一定了。这样我们就理解了为什么这个概率不是 0.5，那它为什么还一定会小于 0.5 呢？我们假想这样一张表，它的前两列与上表完全一样，但第三列的数都正好是第二列的一半，从而第四列就都是 0.5。假想表与上表的第三列虽然不一样，但和是一样的，所以我们可以在不同行间挪动第三列的数值，这行加一点，那行就减掉一点，始终保持第三列的和不变，这样最终我们可以得到上表。这种操作会给第四列的均值带来什么影响呢？会让它变小。理由是这样的：记第二列第 $i$ 行的数为 $n_i$，第三列第 $i$ 行的数为 $k_i$，那第四列的相应数值就是 $k_i / n_i $，当我们从第 $i$ 行挪一单位第三列的值给第 $j$ 行时，第四列的和的变化是 $1/n_j -1/n_i $。所以假如 $n$ 大的那些行 $k$ 倾向于大于 $n/2$，那第四列的均值就会小于假想表的均值——0.5。观察上表你会发现这个倾向是存在的，所以均值确实小于 0.5。那这个倾向是怎么来的呢？来源于重复计算。也就是 HHT 会被我们当做 HH 和 HT 算两遍，这在序列长度为 4 的情况下会看的更清楚，如果我们把 HHHT 算作 HH、HH、HT，那H多的序列 HH 会异常的多，算出来的概率就会小于 0.5，如果我们不重复计算，只把它拆成 HH、HT，那就不会有任何偏差，预期概率= 0.5，这点你可以自行验证。

下一次投掷是正面的概率对不同的投掷次数n、连续出现正面的次数k和单次投掷出现正面的概率 p 作图。“扔一百次，观察连扔出三个正面后下一次的结果”对应着 n=100，k=3，p=0.5，所以下一次是正面的概率约为 0.46。图片取自论文⁸。

假如你还没想明白，这篇论文¹⁰提供了这个问题由易到难的各种变形，其中还包括一道 Google 以前的面试题：

有个国家特别地重男轻女，每家都一直生一直生直到生出儿子为止（只要生出儿子他们就不再生了），请问这个国家人口中女性的比例预期是多少？

我这里就不再展开了，希望你已经能理解答案并不是官方给的 0.5。

好了，说了那么多脑筋急转弯，这些跟手热不热有什么关系呢？请你再看一遍 GVT 里的第 6 项研究：他们让球员各投 100 次，然后算出连续命中三次（或两次、一次）之后下一次的命中率。他们的做法和从一个给定次数的样本中统计连续三次正面后下一次硬币结果的做法是一模一样的，所以这个估计是有偏的。而且偏差很大，在他们的数据里达到了 8%。他们拿着这个有偏的估计发现手热效应不存在，这恰恰证明了手热是存在的，当你把 8% 加回去，原来不显著的现在全都显著了。而 8% 是什么概念呢？库里这赛季的三分和总命中率分别是 41.1% 和 58%，扣掉 8% 在后卫里对应的大概是德文哈里斯和德隆威廉姆斯。所以手热效应不仅存在，而且还很重要。

至此 GVT 中还比较有意义的研究就剩第 7 个：人们感受不到谁手正热，哪怕自己都感受不到。假如真感受不到那哪怕手热了意义也不太大，但人们真的感受不到吗？再看一遍他们的结论，他们发现的是人们的预测和实际投篮结果的相关性只有零点零几，这难道还不说明感受不到？这种论证忽视了一个问题，那就是即使你能准确的感受到队友手热，也不可能准确预测投篮结果，因为哪怕队友的手热得发烫，命中率从平时的 45% 变成了 55%，那也还是有将近一半的概率投丢。我们不妨做个计算，假设队友平时的命中率是 45%，手热时命中率是 55%，手热出现的几率是 15%，你能完美地知道队友处于手热状态，所以在他手热时你全预测投进，手不热的时候全预测投丢，这样相关性会是多少？

以 X 表示投篮，Y 表示预测，1 表示进，0 表示没进，显然 X、Y 都服从伯努利分布。 $$\mu_X = p_X = 0.85*0.45 + 0.15*0.55 = 0.465 $$ $$\mu_Y = p_Y = 0.15$$ $$\sigma_X = \sqrt{p_X(1-p_X)}$$ $$\sigma_Y = \sqrt{p_Y(1-p_Y)}$$ $$cov(X, Y) = E[(X-\mu_X)(Y-\mu_Y)] = E[XY] - \mu_X \mu_Y = 0.55*0.15 - 0.465*.15 = 0.01275$$ $$corr(X, Y) = \frac{cov(X, Y)}{\sigma_X \sigma_Y} = 0.072$$ 所以即使你能完美预测，预测和投篮的相关性也就是零点零几，因此 GVT 的原始数据并不能说明人们无法感受到手热。相关性这么低的原因在于二项分布是个很难准确估计的分布，命中率变化 10% 在观测上要很久才能看出来。Miller 改进并重做了投篮预测实验，发现人们是能够准确感受到手热状态的。

科学危机

GVT 只不过是这几年科学可重复性危机的冰山一角，在心理学领域大量重复失败的研究里¹¹， GVT 的论文质量算是比较好的了。我再随便举两个出了问题的研究：

哈佛美女教授 Amy Cuddy 在 TED 上的著名演讲，认为强势的身体动作 (power pose) 可以助你成功。这个演讲在 TED 上观看次数超过四千两百万，是历史第二多，当时在社交网站上也掀起了一波 "Fake it until you make it" 的鸡汤潮。结果不仅没重复出来，原论文的另一个作者还直接在网上写了一封公开信，声称观点已变，现在认为 power pose 效应不存在了，建议其他研究者别再浪费时间。
社会启动效应 (social priming)。这个领域发端于 John Bargh 1996 年的研究¹²，他们让学生从五个词里挑四个出来造句，一组学生拿到的是普通词，另一组拿到的是与老年相关的词汇，比如健忘、秃顶、满脸皱纹、佛罗里达，实验完后他们告诉学生穿过大厅去另一个地方做下一组实验，但他们关心的其实既不是造句也不是下一个实验，而是学生穿过走廊的速度，他们发现那些被老年相关词汇启动 (prime) 了的学生比另一组慢了将近一秒。这个惊人的结果启发了大量后续研究，十几年来整个领域欣欣向荣，Bargh 的原论文已被引用了四千多次。人们发现啥都能拿来 prime，比如考卷字迹模糊能让你考得更好，让你描述一个严厉教授也能让你在常识测验中表现更好……卡尼曼在《思考：快与慢》里举了整整一章的例子，我们来稍微感受一下，请注意卡尼曼最后的结论，“你别无选择”、“你必须承认”……好吧，现在 Bargh 的最初实验和其他许多实验都没能重复出来，整个领域风雨飘摇。Bargh 前几年没憋住还发了通火，又引来更多非议。这里有一篇手把手教你做 social priming 研究的文章，辛辣搞笑，推荐阅读。当然我不是说这领域一定就全错了，毕竟我也不是专业人士，我只是觉得怎么看我应该都还是有点选择的。

所以心理学到底是怎么落到这步田地的呢？

统计困局

There are three kinds of lies: lies, damned lies, and statistics. - Benjamin Disraeli

The Earth Is Round (p < .05) ¹³

一方面其实是因为心理学比许多同样危机重重的学科更开放、关注度更高，另一方面它也确实有着自身的问题。心理学中广泛使用的p值——以及更一般的，整个假设检验范式，在这次危机中起着核心作用。越来越多人知道单纯地依赖这套范式不好，但大多数人还没意识到它到底有多糟，对此，备受尊敬的前美国心理学会会长保罗·米尔 (Paul Meehl) 是这么说的¹⁴，

I believe that the almost universal reliance on merely refuting the null hypothesis as the standard method for corroborating substantive theories in the soft areas is a terrible mistake, is basically unsound, poor scientific strategy, and one of the worst things that ever happened in the history of psychology.

p值带来了太多的误解 ^15-17，连美国统计学会都少见的出来说了几句¹⁸，如何改进也是众说纷纭^19-22，这里是几点主要的。

Confusion of the inverse

这个问题已经是老生常谈了。p值的含义是在零假设成立的情况下观察到手上数据（或比手上数据更极端的数据）的概率，也就是说它指向的是数据，但大部分情况下我们根本不关心这，我们关心的是凭着手上的数据，假设有多大的可能成立。我们关心的是假设，不是数据。xkcd 的这幅漫画很好地体现了这种区别。

这里我们根本不关心骰子出现这种结果的概率有多大，我们关心的是太阳爆炸的概率有多大。虽然这二者是成正比的，但比例系数（先验概率）可大可小（这里太阳爆炸的先验概率就很小），所以从前者常常推不出太多关于后者的信息，而我们又极度地渴望后者，结果就是p值总是被误用和误读，当它难得被正确使用时，它却是在回答错误的问题 ²³。

The garden of forking paths²⁴

假设检验的另一个严重问题是多重比较，说到这肯定有人想到了 xkcd 的另一幅经典漫画。

这当然是幅很好的漫画，但也有些误导性。它容易让人以为只有这样明显地 p-hacking 才算多重比较，而大部分科学家肯定是不会这样做的，所以他们就不觉得自己会有多重比较的问题。不是这样的，哪怕你自始至终只检验了一个假设，也依然可能会涉及多重比较，让我们来看个例子。

进化心理学之父/母图比 (John Tooby)、柯斯米德斯 (Leda Cosmides) 夫妇提出了一个设想：人类的进化史充满了战斗，而上肢力量对战斗显然很重要，上肢强壮的人更有可能积极参与掠夺和守护资源，即对资源有更强的占有欲，现代社会不打架了，但这种部落时代的心理依然保留了下来，可能会以其他形式体现出来，比如他们可能会更反对政府的再分配政策。然后他们就去收集数据，测量了几百个人的二头肌，并询问他们对再分配政策的态度，果然发现胳膊越粗的人越反对再分配，p=0.007。

多好的研究，有什么问题吗？没有任何问题——除了结果是我编的。他们并没有发现手臂越粗的人越反对再分配，他们发现的是男性手臂越粗就越反对，但女性身上没有这种效应。

很合理啊，战斗、掠夺、守卫资源本来就是男性的事，完美符合进化史，有什么问题吗？没有任何问题——除了结果是我编的。他们也没有发现男性手臂越粗越反对再分配，他们发现的是社会地位高的男性手臂越粗越反对，社会地位低的手臂越粗反而越支持²⁵。

也很合理啊，地位低当然喜欢再分配嘛，有什么问题吗……

有啊，假如他们发现胳膊越粗的人越反对再分配他们会发表文章并视为对自己理论的支持吗？我相信会的；假如他们发现男性手臂越粗越反对再分配他们会发表文章并视为对自己理论的支持吗？我相信会的；假如他们这些都没有发现，但是发现年纪大的男性手臂越粗越反对再分配但年纪轻的男性反之他们会发表文章并视为对自己理论的支持吗？我相信会的，因为年长的人往往积累了更多财富需要捍卫嘛；假如他们发现年纪轻的男性手臂越粗越反对再分配但年长的男性反之他们会发表文章并视为对自己理论的支持吗？我相信会的，因为年轻正是自我打拼赚资源的时候，再分配不符合这种雄心壮志……

我随口编的这些“解释”也许并不都合理，我也不认为 Tooby 夫妇会有任何学术不端（做为进化心理学的忠实拥趸，我很尊敬他们），我想说的正是即使你十分正直诚实，你完全在数据的指引下寻找结论，你自始至终只做了一次假设检验，你依然可能涉及多重比较——潜在的多重比较²⁶，而这也正是假设检验的危险之处。当然这种情况还叫多重比较已经不合适了，有人把它叫做多重建模 (multiple modeling)²⁷，不过我觉得更好的叫法是研究者自由度 (researcher degrees of freedom)。这种自由度可以在参与者浑然不觉的情况下轻易地让p值产生巨大的变化²⁸，因此也非常地难以避免。一个简单的解决办法就是把研究显式地分为探索阶段和验证阶段，探索阶段有了任何新发现，就固定数据分析方法，进行预注册实验，看看在验证阶段是否能重复出来。这适用于成本低廉的实验，但有些实验重复起来太不现实，那就只好采用其它方法，贝叶斯多层模型是一个不错的选择，因为 partial pooling，它能在很大程度上自动校正多重比较问题²⁹。

一项研究发现单身女性在排卵期内会更有可能投票给奥巴马，相反，处于排卵期的非单身女性更愿意投给罗姆尼，变化值超过十个百分点³⁰，这背后当然也有一套进化理论啦，但让我们来看下数据处理上可能的自由度吧：大姨妈后的第几天到第几天算排卵期？那些大姨妈不规律的样本要不要排除掉？如果调查对象对上次大姨妈的时间不够确定，要排除掉吗？多不确定才排除？怎么样算非单身？处于暧昧期的算吗？一垒呢？二垒呢？三垒呢？四垒呢？五垒呢？……Andrew Gelman 认为整个数据至少有168种完全合理的处理方式，然后他把168个置信区间全算了一遍，得到上图，蓝线对应的是论文³⁰实际采用的分析方法。

The law of small numbers

在 GVT 的讨论里我们已经提到了效力不足的问题，教科书上对效力的说法是效力不足你就不能发现该发现的东西，现实比这糟的多，你不是没有发现该发现的，而是会发现一堆乱七八糟不该发现的东西。

首先，在效力不足的情况下，得到的p值会很不稳定 (p dancing¹⁹)，不显著还好，要是碰巧显著了，恭喜你，你又为学术共同体贡献了一个虚假发现。

效应值=1，样本标准差=2，对应的 Cohen's d=0.5，在心理学中被认为是中等效应。图上是模拟实验一万次得到的p值分布，最左边的柱对应着显著 (p≤0.05)，不难看出p值的不稳定。

其次，如果你在效力不足时得到了显著性，你很有可能会严重高估效应值，甚至连符号都可能是错的，这被称作抖 S(ign) 和 M(agnitude) 型错误³¹。~~S怎么也能算错误~~

真实效应值=1，可以看出即使效力=0.8估计值也普遍偏高。

那现实中科学研究的效力如何呢？这几年有人评估了神经科学领域的研究，认为中位效力只有 21%³²；经济学领域则不超过 18%³³；Cohen 在 62 年评估了心理学界，认为如果研究的是小效应，当时的效力大概只有 18%，中等效应的话有 48%³⁴；二十几年后人们用 Cohen 的方法又研究了一遍，发现心理学研究效力不仅没有改善，还恶化了³⁵。这些研究有的有争议 ³⁶，但即使最乐观的研究者大概也不会认为现在的平均效力超过0.5。其实早在 1971 年，后来的诺奖得主卡尼曼和他的好基友特沃斯基就注意到他们的同行常常严重高估实验的效力，仿照大数定律，他们把这种认为小样本也能很好地代表整体的倾向命名为小数定律³⁷。这种错误倾向有多难以克服呢？你只要再读一遍论文作者的名字就知道了。看到了吗？特沃斯基，他就是 GVT 里的那个T啊，连他都掉到自己发现的坑里去了。而卡尼曼也没好到哪去，几个月前刚刚有人指出《思考：快与慢》中关于 social priming 的那一章是效力严重不足的……卡尼曼爽快地承认了错误。

为了避免让你产生这届心理学家不行的误解，我们来做个测试。

Satoshi Kanazawa 是伦敦政经的进化心理学家，他热衷于扩展 Trivers–Willard 假说，所以论文题目看起来常常是这样的，

他还合著过一本书，书名也很简单粗暴，叫 Why Beautiful People Have More Daughters（中译本叫《生猛的进化心理学》）。我对政治不正确的研究没有任何意见——如果不是喜闻乐见的话，但是你的方法得靠谱啊。Kanazawa 的研究充满了各种统计错误^31,38-40，我们以上图最后一篇论文为例，他分析了Add Health 数据集，这是最大、最综合的关于青少年的长期追踪数据集，从人们还是青少年时开始追踪，四次回访，记录了这些人的许多信息，其中两个与这里的话题有关，一是这些人的孩子的性别，二是调查人员会对被采访者的外貌打分，从最丑到最美共有五档。Kanazawa 选取了第三次回访中有孩子的人（2972个）进行了分析，他发现最美的一档生女孩的可能性比其他人高 8 个百分点(56% vs 48%)，p < 0.05，所以，beautiful parents have more daughters。首先这里存在明显的多重比较问题，第一档和其他档比显著了，和最丑的一档比怎么反而不显著了呢？为什么不把最美的两档和剩下的三档比，假如这么比显著的话 Kanazawa 会不提吗？……这里更好的办法显然是对外貌和出生性别建模，弄个回归方程啥的，而不是进行假设检验。不过即使那样也是毫无意义的，为什么呢？因为，你觉得 2972 个样本在这里效力够吗？

假如放平时我相信大部分人都觉得够了，快三千个样本，在心理学里是很大型的研究了，今天你还可以看到很多研究就几十个样本呢。但现在我这么一问你应该也知道答案是不够，那我再问，你觉得多少样本才够呢？

我们还是来算算吧。生男生女是个伯努利事件，假设我们现在有数量相等的两档人，每一群人数是 n，从过往的研究中我们知道人类在极度营养不良下生女孩的概率大约会增加 3%⁴⁰，所以长相对性别比的影响极不可能是 8%，撑死 2% 吧。一群人出生性别比的样本标准差是 $\sqrt{p(1-p)/n} \approx 0.5/\sqrt{n} $，两个群体的差值的标准差就是 $0.5 * \sqrt{2/n} $。2%的效应想要显著，意味着 2% 得大于两倍标准差， $$0.02 > 2 * 0.5 * \sqrt{2/n} \Rightarrow n > 5000 $$ 这是一个群体的数量，两个群体就至少得要一万人。这已经是很乐观的估计了，假如效应值是 1%（实际情况很可能更小），那样本数就得要四万了。需要这么大的样本的原因和分析投篮数据时是一样的——二项分布难以估计准确。所以这项研究的问题与假设检验无关，改用贝叶斯也救不回来，这个研究想法出生的那刻就已经被宣判死刑了。

现在你知道通过直觉判断效力有多容易出错了吗？这也是效力计算为何如此重要的原因。对于那些复杂的实验，可能没办法从数学上进行计算，那可以先在已知数据上进行验证，例如用电脑生成的手热数据去验证 GVT 的统计方法。我这里还有两个特别有趣的例子，一个是关于功能性磁共振成像 (fMRI) 的。你应该读过一些科学家一边让被试做某类事情一边用 fMRI 扫被试大脑，然后得出XX脑区负责XX行为的研究。09 年的时候有人想这玩意到底可不可靠啊，于是他们就用 fMRI 去扫描了一条三文鱼的大脑，结果发现了信号，好像没什么不对？可是，他们扫的是一条死三文鱼啊⁴¹（这项研究众望所归地拿到了搞笑诺贝尔奖）。同年还有另外两篇论文指出当时 fMRI 实践中大量存在的几个统计问题^42,43，自那以后大部分研究者都提高了警惕，整个领域发展的更健康了（至少比心理学好）。而就在今年，有人把神经科学中常用的一些研究方法应用到红白机处理器上去，想看看这些方法到底能不能帮助我们理解一个信息处理系统，结果不能算特别乐观⁴⁴。这篇文章太新了，结论可以再讨论，不过这个研究想法是非常好的。

和物理学的比较

在说了那么多假设检验的坏话后，有两个问题是无法回避的：其他学科，包括物理学，也有使用假设检验，为什么没这么多问题？假设检验在它的提出者手里为什么表现的挺好，不然后来也不会被这么广泛地使用？这是两个好问题，通过对比成功范例可以让我们更好地理解心理学里的假设检验到底哪出了问题，以及为什么这种问题更多的是因为使用上、而不是理论上的错误^23,45。

粒子物理学里也使用假设检验，虽然他们往往不说p值，而说几个标准差，但是一个意思，那为什么希格斯粒子的发现就显得那么可靠？首先，他们使用的p值很小。发现新粒子的黄金标准是5个标准差，对应的p值大概是三百万分之一。这还是开新闻发布会时候的数字，等写成论文发表的时候 ATLAS 的数据已经达到5.9个标准差，对应的p值是 $1.7*10^{-9}$⁴⁶。相应地，最近有人就提议把社会科学的p值改为 0.005 作为一种临时的应对措施^20-22。

其次，物理学的原假设是很强的，希格斯粒子不存在是个完全有可能的假设，需要很强的数据才能否定它，所以当它真的被否定时，对互补假设是强大的支持。心理学则完全不是这样。心理学里最常见的零假设是 X 和 Y 没有关系（相关 = 0），或者稍微强一点的也就是 X 对 Y 的效应是正的（或者负的），否定这样一个假设你得到的结论也不过是 X 和 Y 有关系，至于多大的关系、重要吗，则完全无法回答。而考虑到心理学中各因素的普遍联系，X 和 Y 没有关系几乎是一个必然错误的假设，所以推翻它毫无意义。经期跟投票倾向有关吗？当然有关啊，而且不仅经期有关，你的身高体重鞋码三围无名指长度姓名首字母织毛衣的技巧……都会有关好吗？只不过有些相关太小了需要极大的样本才能检测出来，Paul Meehl 曾经分析过一个有 45 个变量、包含 57000 个明尼苏达高中生资料的数据集，计算了所有变量的两两关系，结果其中 92% 是显著的，71% 的p值小于 $10^{-6}$⁴⁷，所以有人说假设检验本质上是对于样本数的检验，就是这个意思。这种弱检验导致了心理学和物理学的一个重要区别：随着测量手段的进步，物理学里否定一个假设越来越难了，从而得到的结论也越来越可靠；而心理学里否定零假设却越来越容易了，你放眼望去哪都是效应，整个学科却没什么进展⁴⁸。这个方法论上的区别最早还是由 Meehl 注意到的，所以也被称作 Meehl's paradox ^14,49,50。

最后，物理学从理论假设到实验数据的链条紧实，实验数据可以很有效地作用在理论上。什么意思呢？当一个理论对现实做出预测时，它往往需要涉及一些辅助假设，举例来说，从长得好看的人多生女孩有进化优势（理论假设）推出在 Add Health 数据集中最美的一档会比其他人多生女孩（观测）时，我们暗中引入了许多辅助假设，比如好看的女性往往也嫁给社会地位高的男性，假如好看女性的基因会让她多生女孩，那成功男性的基因会不会导致多生男孩呢？Add Health 的数据有代表性吗？对外貌的评分准确吗？Kanazawa 使用的是第三次回访的数据，这数据有什么系统性偏差吗？……只有当这许多假设都成立的时候我们才能从理论推出预期的观测结果，反过来说，如果我们没观察到预期结果，也不意味着理论一定错了，因为也可能是辅助假设错了（这点后文还会进一步讨论）。所以如果整个逻辑链条过于松垮，实验对理论的验证效果就很弱。物理学中的推理当然也需要辅助假设，例如从希格斯粒子的存在到监视器上的信号，中间你至少需要一套关于加速器正确运作的假设。但物理学中的辅助假设常常可以被单独验证，比如 LHC 开机后，做的第一件事不是寻找希格斯粒子，而是先跑一组已知结果的实验，看看能不能重复出来（对 LHC 发现希格斯粒子的整个过程，《粒子狂热》是一部不错的纪录片）。这也顺便回答了本节开头的第二个问题：为什么假设检验在它的提出者手里应用的不错？首先，不管是费雪、内曼还是皮尔逊，他们要是看到今天心理学家对假设检验的使用都绝对会气得晕过去¹⁷，他们在提出这套方法时都特意强调了它的局限和问题；其次，费雪当时研究的是农业产量问题，理论（施肥会带来产量增长）到观测（那块施了肥的地玉米确实长得比较多）几乎零距离，只要效力足够，假设检验用起来是没什么问题的⁵¹。

抽屉效应/发表偏见

理论上，抽屉效应指的是因为期刊喜欢发表新发现，所以一个没得到显著性的研究会被锁在抽屉里，没有机会发表，从而你在期刊上看到的都是正面结果。抽屉效应确实是发表偏见的结果，但我不同意很多人因此把二者混为一谈，进一步的区分是有意义的。现实中，心理学的发表偏见不仅仅是不发表没有得到显著性的论文，如果只是这样，那倒也还好，毕竟有着研究者自由度，躺在抽屉里的文章其实不会太多。糟糕的是它们还不喜欢发表重复实验。心理学家极少重复前人的实验⁵²，直接重复就更少了，他们偏好间接重复 (conceptual replication)，也就是把原始实验改来改去，理论上相符就好。这一恶习甚至连费曼都注意到了。

错误链式反应

当我们把上面的种种因素拼在一起时，一个危机蔓延的动力学模型就呼之欲出了：

一个统计学不过关的研究者一天有了个新奇的想法，于是设计了一个实验（效力不足）;
然后轻易地得到了显著性（研究者自由度）;
他得到的效应值非常惊人，迅速吸引了大量眼球（M型错误）;
其他研究者跟进，不直接重复，而是略微修改原来的实验（发表偏见）;
大部分研究者也得到了显著性（还是研究者自由度）;
许多结果也很惊人（还是M型错误）;
少数不支持原假设的实验从没有得到发表，研究者甚至都没有把它们投出去，所以大多数人都不知道还有负面结果（抽屉效应）;
重复 4-7;
整个领域欣欣向荣，一切看起来那么的真实美好;
一天有人说：“要不我们来重复一下最初的实验吧……”

这似乎就是发生在 social priming 领域的事（最早的那篇 Bargh 的论文样本数是 30……），不然你很难解释卡尼曼怎么会那么信誓旦旦。这个领域中不乏正直的人，仅仅用发表压力下的学术不端解释不了这样大规模的错误，只能说大家是真的相信。

这个模型并不是我的首创，again，Meehl 1967 年就提出了类似的看法⁴⁹。这真的让人觉得很奇怪，Meehl 在心理学界备受尊重，所有人都知道他，他 60 年代写的一系列批评假设检验的文章也很多人读，但就是没人听，几十年过去了整个学科还是老样子。到了 90 年代，Meehl 自己都受不了了，又写了好几篇文章，其中有一篇无奈地说道⁵³：

Ten years later, I wrote at greater length along similar lines (Meehl, 1978); but, despite my having received more than 1,000 reprint requests for that article in the first year after its appearance, I cannot discern that it had more impact on research habits in soft psychology than did Morrison and Henkel. Our graduate students typically plan and write their doctoral dissertations in blissful ignorance that “the significance test controversy” even exists, or could have a bearing on their research problems. This article (see also Meehl, 1990c) is my final attempt to call attention to a methodological problem of our field that I insist is not minor but of grave import.

然后现在危机降临，我们又把他翻了出来，希望这次心理学家能走点心吧……

应对

While it is easy to lie with statistics, it is even easier to lie without them. - Frederick Mosteller

正确的应对

现在来说下作为外人怎么消化这次危机吧。因为心理学的软科学性质，心理学家中最优秀的那一撮对方法论其实思考颇多（比如 Meehl），所以他们对危机的到来早有准备，我们来看看大卫·巴斯在《进化心理学》里为大家打的预防针。

大卫·巴斯把整个第二章都贡献给了方法论，这在外国教材中很常见，而在中国人写的书里你基本看不到。就如辉格指出过的，中国人最需要普及的不是科学知识，而是科学哲学。

熟悉科学哲学的朋友一眼就能看出这套“硬核-保护带-具体假设”的模型正是拉卡托斯的理论，不过可能为了减轻学生负担，巴斯把拉卡托斯的名字藏到了最后引用的文献里（那篇论文的题目叫 "Are evolutionary explanations unfalsifiable? Evolutionary psychology and the Lakatosian philosophy of science."）。

通过拉卡托斯的框架可以很好地消化这次危机。经过这次危机，我的认知结构里关于心理学的许多具体假设完全改变了，一部分中级理论节点进行了适当更新，至于硬核……谁会怀疑进化论呢？但显然不是每个人都像我这样。

错误的应对

最常见的反应是虚无主义，当面对一个巨大的冲击时，全部否定也确实是最省事的办法，例如这是三联周刊科学记者土摩托。

再下一步就是相对主义。“科学并不总是对的”，每隔一段时间就会有那么一个人（一般是妹子）跟我说这句话，试图让我明白她相信的那套星座、中医……和科学应该有着同样的地位。如果你喜欢观察小粉红的话，你对这种“XX其实也会XX，所以XX和XX本质上是一样的”的论点一定不会陌生，不过小粉红太脏了，我还是举个高贵的女权主义的例子吧。Eve Ensler——前几年很火的《阴道独白》的作者，认为对女性的压迫在世界各地都是完全一样的 (exactly the same)，无论是在西方国家，还是在巴基斯坦、沙特阿拉伯和伊朗。

正在被石刑处决的阿富汗少女。至于女性割礼的图，我就不放了吧。

最后就是解构主义了，不过像《跨越界线：通往量子引力的转换诠释学》这种东西实在不是我写得来的，还是把它留给各位后现代大神（经）吧。

你应该已经看出一个层次化的认知模型有多重要，否则稍不注意就不知道滑到哪里去了。但这自然带来了一个疑问，我们之前刚刚批评过心理学中从理论到观测的链条太过松垮，拉卡托斯的这个模型难道不也是这样吗？我的回答是这样的：辅助假设是不可避免的，你应该让它尽量的紧实（就像物理学中那样），但如果你拉得太过头，赋予了它多于事实能支撑的强度，那整个结构又会牵一发而动全身，也是不行的。至于怎么样的强度算恰当，这就取决于你的学术品味了。品味这东西一两句也说不清，我只能尽力而为，一方面我可以给你推荐一些比较好的书，品味都是多吃好东西培养出来的；另一方面我可以提供一些拇指法则，尽量帮你避开一些坑：

假如你对一项研究没把握，请假设它是错的^54,55；
或者至少假设它的结论是严重高估的³¹；
如果一篇论文没有任何描述性统计的图，而全是假设检验的表，这是一个危险信号；如果只报告p值，完全没提效应值和置信区间，这是一个危险信号；如果使用了一堆假设检验却完全没有效力分析，这是一个危险信号；如果一篇文章有两个危险信号，你最好还是把它扔到垃圾桶去；
除非你特别特别感兴趣，社会启动效应和具身认知的东西先不要看了；
如果你对魔鬼经济学或格拉德威尔深信不疑，那你对心理学的信任太高了 ⁵⁶；
……

结语

当代科学是在未知领域中的航行，走过的每一步都留下怯懦的教训，很多旅客宁可留在家里。——卡尔·萨根

写了这么多，可能有人会觉得”认识这世界好难啊“，但我想说的其实是它也好有趣啊。我知道大多数人可能无法体会这种乐趣，想想你有两个差不多的选择和只有一个选择时的心理状态，前者常常反而让你更焦虑。在不确定性中权衡、评估、决策是件费脑的事，我们渴望确定性。

对此进化给出的解法是观测-评估-权衡-决策-保持-迭代，举例来说，当一个人搬了新家，为了解决吃饭问题，他一开始会先多尝试几家餐馆，在评估比较后选出一些适合的（适合自己吃的、适合和朋友夜宵的、适合约妹子的……），之后这个决策过程就极大地自动化了，直到餐馆质量发生了足够大的变化，那就再启动新一轮的评估。当然各个环节并不是泾渭分明的，整个过程更类似于多臂老虎机问题的 epsilon-decreasing 解法，一开始探索性质多一些，然后慢慢减少。这样的做法牺牲了一定的准确性，但节省了大量的认知资源，是合理的。问题出在有些人的迭代频率如此之低，低到一辈子就评估那么一次。

有的人 25 岁就死了，只是到 75 岁才埋葬。——富兰克林

在生理上，这对应着大脑神经会在青春期经历一次大规模的修剪，之后就相对稳定下来。这种模式在停滞的传统社会是适应的，通过青春期躁动在部落/村庄中找到尽可能好的生态位，之后就安稳过日子吧，就算你想学也没啥东西能让你学了，但在现在这个时代，这种做法显然已经不合适了。可惜，很多人的大脑还没能跟上这个变化。

这些年我看着不少朋友变得日渐愚蠢偏执，曾经的理性和趣味在他们身上慢慢褪去，说实话有时候是很伤感的。当然我能够体谅，但依然觉得遗憾，所以写下这篇文章，希望你们能再出门看看。

参考文献

就我所见，对此次危机的统计方面讨论得最透彻的是 Andrew Gelman，他的博客和文章是个不错的起点。假如你对更一般的方法论和科学哲学感兴趣，那 Paul Meehl 是绕不过去的。

1.Skinner, B. F. "Superstition" in the pigeon. Journal of Experimental Psychology: General 121, 273 (1992).
2.Herbranson, W. T. & Schroeder, J. Are birds smarter than mathematicians? Pigeons (Columba livia) perform optimally on a version of the Monty Hall Dilemma. Journal of Comparative Psychology 124, 1 (2010).
3.Herbranson, W. T. Pigeons, humans, and the Monty Hall dilemma. Current Directions in Psychological Science 21, 297–301 (2012).
4.Zentall, T. R., Case, J. P. & Collins, T. L. The Monty Hall dilemma with pigeons: No, you choose for me. Learning & behavior 43, 209–216 (2015).
5.Stagner, J. P. & Zentall, T. R. Further investigation of the Monty Hall Dilemma in pigeons and rats. Behavioural processes 112, 14–21 (2015).
6.Gilovich, T., Vallone, R. & Tversky, A. The hot hand in basketball: On the misperception of random sequences. Cognitive psychology 17, 295–314 (1985).
7.Korb, K. B. & Stillwell, M. The story of the hot hand: Powerful myth or powerless critique. in international conference on cognitive science (2003).
8.Miller, J. B. & Sanjurjo, A. Surprised by the gambler’s and hot hand fallacies? A truth in the law of small numbers. (2016).
9.Miller, J. B. & Sanjurjo, A. A cold shower for the hot hand fallacy. (2014).
10.Miller, J. B. & Sanjurjo, A. A Bridge from Monty Hall to the (Anti-) Hot Hand: Restricted Choice, Selection Bias, and Empirical Practice. (2015).
11.Collaboration, O. S. & others. Estimating the reproducibility of psychological science. Science 349, aac4716 (2015).
12.Bargh, J. A., Chen, M. & Burrows, L. Automaticity of social behavior: Direct effects of trait construct and stereotype activation on action. Journal of personality and social psychology 71, 230 (1996).
13.Cohen, J. The Earth Is Round (p < .05). American Psychologist 997–1003 (1993).
14.Meehl, P. E. Theoretical risks and tabular asterisks: Sir Karl, Sir Ronald, and the slow progress of soft psychology. Journal of consulting and clinical Psychology 46, 806 (1978).
15.Greenland, S. et al. Statistical tests, P values, confidence intervals, and power: a guide to misinterpretations. European journal of epidemiology 31, 337–350 (2016).
16.Haller, H. & Krauss, S. Misinterpretations of significance: A problem students share with their teachers. Methods of Psychological Research 7, 1–20 (2002).
17.Gigerenzer, G. Mindless statistics. The Journal of Socio-Economics 33, 587–606 (2004).
18.Wasserstein, R. L. & Lazar, N. A. The ASA’s statement on p-values: context, process, and purpose. (2016).
19.Cumming, G. The new statistics: Why and how. Psychological science 25, 7–29 (2014).
20.Benjamin, D. J. et al. Redefine statistical significance. Nature Human Behaviour (2017).
21.Amrhein, V. & Greenland, S. Remove, rather than redefine, statistical significance. Nature Human Behaviour 1 (2017).
22.Lakens, D. et al. Justify Your Alpha: A Response to “Redefine Statistical Significance.” (2017).
23.Gelman, A. Commentary: P values and statistical practice. Epidemiology 24, 69–72 (2013).
24.Gelman, A. & Loken, E. The garden of forking paths: Why multiple comparisons can be a problem, even when there is no “fishing expedition” or “p-hacking” and the research hypothesis was posited ahead of time. Department of Statistics, Columbia University (2013).
25.Petersen, M. B., Sznycer, D., Sell, A., Cosmides, L. & Tooby, J. The ancestral logic of politics: Upper-body strength regulates men’s assertion of self-interest over economic redistribution. Psychological Science 24, 1098–1103 (2013).
26.Gelman, A. The problems with p-values are not just with p-values. The American Statistician, supplemental material to the ASA statement on p-values and statistical significance 10, 1154108 (2016).
27.Young, S. S. & Karr, A. Deming, data and observational studies. Significance 8, 116–120 (2011).
28.Simmons, J. P., Nelson, L. D. & Simonsohn, U. False-positive psychology: Undisclosed flexibility in data collection and analysis allows presenting anything as significant. Psychological science 22, 1359–1366 (2011).
29.Gelman, A., Hill, J. & Yajima, M. Why we (usually) don’t have to worry about multiple comparisons. Journal of Research on Educational Effectiveness 5, 189–211 (2012).
30.Durante, K. M., Rae, A. & Griskevicius, V. The fluctuating female vote: Politics, religion, and the ovulatory cycle. Psychological Science 24, 1007–1016 (2013).
31.Gelman, A. & Carlin, J. Beyond power calculations: Assessing Type S (sign) and Type M (magnitude) errors. Perspectives on Psychological Science 9, 641–651 (2014).
32.Button, K. S. et al. Power failure: why small sample size undermines the reliability of neuroscience. Nature Reviews Neuroscience 14, 365–376 (2013).
33.Ioannidis, J. P. A., Stanley, T. D. & Doucouliagos, H. The power of bias in economics research. The Economic Journal, forthcoming, SWP, Economics Series 1, (2016).
34.Cohen, J. The statistical power of abnormal-social psychological research: a review. The Journal of Abnormal and Social Psychology 65, 145 (1962).
35.Sedlmeier, P. & Gigerenzer, G. Do studies of statistical power have an effect on the power of studies? Psychological bulletin 105, 309 (1989).
36.Nord, C. L., Valton, V., Wood, J. & Roiser, J. P. Power-up: A Reanalysis of’Power Failure’in Neuroscience Using Mixture Modeling. Journal of Neuroscience 37, 8051–8061 (2017).
37.Tversky, A. & Kahneman, D. Belief in the law of small numbers. Psychological bulletin 76, 105 (1971).
38.Denny, K. Big and tall parents do not have more sons. (2007).
39.Gelman, A. Letter to the editors regarding some papers of Dr. Satoshi Kanazawa. Journal of Theoretical Biology 245, 597–599 (2007).
40.Gelman, A. & Weakliem, D. Of beauty, sex and power: Too little attention has been paid to the statistical challenges in estimating small effects. American Scientist 97, 310–316 (2009).
41.Bennett, C. M., Miller, M. B. & Wolford, G. L. Neural correlates of interspecies perspective taking in the post-mortem Atlantic Salmon: An argument for multiple comparisons correction. Neuroimage 47, S125 (2009).
42.Vul, E., Harris, C., Winkielman, P. & Pashler, H. Puzzlingly high correlations in fMRI studies of emotion, personality, and social cognition. Perspectives on psychological science 4, 274–290 (2009).
43.Kriegeskorte, N., Simmons, W. K., Bellgowan, P. S. F. & Baker, C. I. Circular analysis in systems neuroscience: the dangers of double dipping. Nature neuroscience 12, 535–540 (2009).
44.Jonas, E. & Kording, K. P. Could a neuroscientist understand a microprocessor? PLoS computational biology 13, e1005268 (2017).
45.Gelman, A. The problem with p-values is how they’re used. Ecology. Online: http://www.stat.columbia.edu/gelman/research/unpublished/murtaugh2.pdf (2013).
46.Aad, G. et al. Observation of a new particle in the search for the Standard Model Higgs boson with the ATLAS detector at the LHC. Physics Letters B 716, 1–29 (2012).
47.Meehl, P. E. Why summaries of research on psychological theories are often uninterpretable. Psychological reports 66, 195–244 (1990).
48.Newell, A. You can’t play 20 questions with nature and win: Projective comments on the papers of this symposium. (1973).
49.Meehl, P. E. Theory-testing in psychology and physics: A methodological paradox. Philosophy of science 34, 103–115 (1967).
50.Meehl, P. E. The problem is epistemology, not statistics: Replace significance tests by confidence intervals and quantify accuracy of risky numerical predictions. (1997).
51.Meehl, P. E. What Social Scientists Don’t Understand.
52.Makel, M. C., Plucker, J. A. & Hegarty, B. Replications in psychology research: How often do they really occur? Perspectives on Psychological Science 7, 537–542 (2012).
53.Meehl, P. E. Appraising and amending theories: The strategy of Lakatosian defense and two principles that warrant it. Psychological Inquiry 1, 108–141 (1990).
54.Ioannidis, J. P. A. Why most published research findings are false. PLoS medicine 2, e124 (2005).
55.Siegfried, T. Odds are, it’s wrong: Science fails to face the shortcomings of statistics. Science news 177, 26–29 (2010).
56.Gelman, A. & Fung, K. Freakonomics: What Went Wrong? American Scientist 100, 6 (2012).

一个直播

Tue, 22 Nov 2016 00:00:00 +0000

最近在国内玩耍，前阵子顺便帮万门大学做了一个直播，讲的是如何自学编程。

内容主要包括三部分：学习的科学 (Learning science)、数据科学行业介绍和找工作的一些问题。讲的并不深入，主要是告诉你这世界上还有这样一些东西，你觉得有帮助的话就自己去学吧。这也是我的一贯风格，用黑话来说，我喜欢（似乎也比较擅长）帮别人把 unknown unknown 变成 known unknown，至于最终怎么到达 known known，我一般开点资料就完事了，因为手把手教实在太麻烦，不适合我这种懒人。。现在视频已经剪辑制作完了，万门也把关于 learning science 的那部分作为试看免费放在了网上。我个人认为这部分是最通用的，如果你想转行当个程序员，那当然也可以交钱看完整的录像，不然看看这部分就可以了。

第一次出镜，~~比陶哲轩还快三倍的语速暴露了我那颗想当网红的内心，~~希望能对大家有所帮助。

2017/10/14更新：最近跟统计之都的朋友们做了个直播，也聊了相关的话题。

环美旅行

Sun, 03 Jul 2016 00:00:00 +0000

打算这个月中到9月初环美一趟，之后可能再回国玩一阵。行程还有一些未定的部分，愿意接待/面基的朋友欢迎联系我 tcya24@gmail.com。

万事屋

Dennett走了，Dennett来了

为什么金斯伯格反对罗诉韦德案

Speaking in a Judicial Voice 以司法之声发言

Introduction 导言

I. Collegiality in Appellate Decisionmaking 上诉判决的合议性

II Measured Motions in Third Branch Decisionmaking 第三权判决中的审慎提议

Conclusion 结论

公平的多样性

海德沙龙翻译组

我的师承

我们没有敌人？

饼图的妙用

心理学的危机

缘起

手热迷思

“手热迷思”迷思

科学危机

统计困局

The Earth Is Round (p < .05) 13

Confusion of the inverse

The garden of forking paths24

The law of small numbers

和物理学的比较

抽屉效应/发表偏见

错误链式反应

应对

正确的应对

错误的应对

结语

参考文献

一个直播

环美旅行

The Earth Is Round (p < .05) ¹³

The garden of forking paths²⁴