卷:卷二《长安月下·关中六十夜》|主角:上官婉儿·Marco|地点:西安·碑林博物馆·夜|典:《旧唐书》“婉儿常代帝及后、长宁安乐二公主,数首并作,辞甚绮丽,时人咸讽诵之”、昆明池赐彩楼评诗事、《周易·系辞》“书不尽言,言不尽意”|主导感官:视(碑上的刀口、纸的白、朱笔的红)|过所进度:空白·一字未刻|金融-国学对应:Who Sets the Standard(评级权:真正的权力不在打分,在定分数怎么算)↔ 称量天下士

段一·现身

那支笔,是朱红的。

第六十八夜。

马可从大雁塔出来,本来打算回去睡觉。

结果他在寺门口,遇见了一个女人。

她穿着极其精致的唐式宫装,不是不夜城里那些演员穿的那种,那些是给游客看的,颜色艳,形制夸张。她这一身很素:藕色的襦,外面一件深青的半臂,裙是那种沉下去的红。发式很复杂,可没有一件多余的首饰。

她站得很直。手里拿着一支笔。

那支笔的杆是竹的,很旧。笔头是朱红的。

“你写的那三十七页,我看过了。”

马可怔住:“您是……”

“上官婉儿。”她说。

“我这一辈子的活,就是,替天下人,把文章分个高下。”

她看着马可:“你那三十七页,我给它评了个等。

马可站在原地,忽然觉得手心有点凉。

“……几等?”

上官婉儿把那支朱笔在指间转了半圈。

不入等。

马可站在寺门口,一时不知道该说什么。

他这一辈子被人否定过很多次。可从来没有人用这三个字。

“不入等”跟“不行”是不一样的。

“不行”是一个判断,你可以争,可以补,可以拿数据回来打。

“不入等”是,这东西压根没进到那个可以被评判的范围里。

他忽然想起在纽约那些年,他们内部有一个说法。收到外面投来的项目材料,如果一份材料连初筛都过不了,负责的人会在邮件里写两个字:pass。

不说理由。因为说理由,等于承认它值得一个理由。


段二·古文镜像

“书不尽言,言不尽意。”
English Mirror: Writing does not exhaust speech; speech does not exhaust meaning.

她带他去了碑林。

夜里的碑林是关着的。可门自己开了,或者说,马可后来也说不清是怎么进去的。

一进去,他就明白为什么是这里。

一排一排的房子,一排一排的碑。石头的。有的高过人头,有的只有半人高。上面全是字。

《开成石经》一百一十四块,立在那儿,像一片石头的森林。

《颜勤礼碑》。《多宝塔碑》。《玄秘塔碑》。《大唐三藏圣教序》,马可站在这一块前面停了很久,因为那是唐太宗为玄奘写的序,字是从王羲之的字里一个一个拼出来的。

上官婉儿在一块空地上站住了。

“年轻人,你知道这地方是干什么的?”

“……放碑的。”

“不对。”她说,“这地方是放标准的。”

马可怔住了。

“你看那些字。”她用朱笔往四周指了一圈,“颜真卿、褚遂良、欧阳询、柳公权。”

“一千多年来,中国人练字,练的是这几个人的字。”

“为什么是他们?”

马可摇头。

“因为有人把他们刻上了石头。”上官婉儿说。

她转过身,看着马可:“唐朝写字的人,不止这几个。写得好的,也不止这几个。”

“可能立在这儿的,只有这几个。”

她把朱笔举起来:“年轻人,我今夜要跟你讲的,就是这一件事。”

谁在决定,哪几个人能刻上石头。

马可在那一片石碑之间慢慢走。

他忽然注意到一件事:这些碑的样子差别很大。

有几块极精,石料细腻,字口深峻,一千多年了刀痕还是利的。旁边有几块就粗糙得多,石头本身有裂,字也刻得浅,有几处已经模糊了。

“女史,”他问,“为什么有的刻得好,有的刻得差?”

“钱。”上官婉儿说得很干脆。

“钱?”

“立一块碑要多少钱?石头要开、要运、要磨。写碑文要请人。刻要请良工。”她说,“颜真卿写的《颜勤礼碑》,那是他给自己曾祖父立的,他家是望族,请得起最好的石工。”

她抬起朱笔指了指远处一块残碑:“那一块是一个县尉给他母亲立的。他请不起,字是他自己写的,刻的是本地的石匠。”

马可站在那儿,忽然明白了。

能不能留下来,第一道门槛不是好不好。是,立得起立不起。


段三·事件主体

“我讲我自己的事。”上官婉儿说。

“我祖父叫上官仪。他是宰相,也是当时最好的诗人之一。”

“麟德元年,他给高宗写了一份废后的诏书草稿。”

马可屏住了呼吸。他知道这一段。要废的那个后,是武则天。

“那份草稿,武后知道了。”上官婉儿说,“同年,我祖父下狱,被杀。我父亲上官庭芝,一同被杀。”

“那您……”

“我那时候刚出生。”她说,“我和我母亲,被收进掖庭为奴。”

马可久久没有说话。

“掖庭是什么?”

“宫里做粗活的地方。”上官婉儿说,“我母亲郑氏在那儿教我读书。她说,你祖父的仇,你报不了。可你祖父的学问,你不能丢。”

“我在掖庭长到十四岁。”

“那十四年是什么样子的?”

上官婉儿想了一会儿。

“洗衣服。”她说,“掖庭里的活,大半是洗。宫里几千人的衣裳,都在那儿洗。”

“冬天的水凉。手上生了冻疮,裂开,血渗到衣服上,就得重洗。”

她说得很平常,没有一点自怜。

“那读书呢?”

“晚上。”她说,“我母亲把字写在地上,用水。写完晒干了,看不出来。”

马可怔住了:“用水写?”

“纸和墨,是掖庭里没有的东西。”上官婉儿说,“也是不许有的东西。”

她抬起手上那支朱笔,看了一眼:

“年轻人,我十四岁那年第一次拿到一支真笔。”

“那一天我拿着它,手抖了很久。”

“然后呢?”

“然后武后召见我。”


“为什么召见?”马可问。

“她听说掖庭里有个女孩子,会写文章。”上官婉儿说,“她当场出了个题,让我写。”

“您写了?”

“写了。”

“她怎么说?”

上官婉儿沉默了一会儿。

“她看完,说了四个字。”

“什么?”

‘此女有才。’

她把朱笔收进袖子里:“然后她把我放出了掖庭,让我掌管文书。”

马可站在那儿,只觉得后背发凉。

“杀您祖父的人,”

“是她。”上官婉儿说得很平静。

“提拔您的人,”

“也是她。”

石碑在夜里泛着微弱的光。

“年轻人,”她说,“你现在心里在想什么?”

马可老实说:“我在想……我不知道该怎么想。”

“大部分人到这儿都不知道该怎么想。”上官婉儿说,“所以他们就选一个说法:要么说我忍辱负重,要么说我认贼作父。”

“两个说法都很省事。”

“那真实的是什么?”

她看了他很久。

“真实的是,”她说,“我十四岁那年,只有一条路可以走。而那条路上,有一个人在等着我。

她顿了顿,又补了一句:

“还有一样,你们后人都不肯提。”

“什么?”

我那时候十四岁。”上官婉儿说,“一个十四岁的孩子,从掖庭里被叫出来,见到当朝最有权势的那个人,她要么写,要么回去洗衣服。”

“你们后世写我这一段,写的是一个成年人的抉择。”

她看着马可:“不是。那是一个孩子,抓住了她这一辈子唯一伸过来的一只手。”


“后来呢?”马可问。

“后来四十年。”上官婉儿说,“我从十四岁到四十六岁,一直在写文书。”

“写什么?”

“诏、敕、制、册。”她说,“皇帝的话,皇后的话,公主的话。”

她抬起眼:“都是我写的。

马可怔住了。

“《旧唐书》里怎么写这一段,你读过没有?”

“我读过一点。”

“它说:婉儿常代帝及后、长宁安乐二公主,数首并作,辞甚绮丽,时人咸讽诵之。”

她慢慢地说:“意思是,我一个人,同时替四个人写。写出来的东西各有各的口气,谁也认不出是一个人写的。”

“当时的人,人人都在背。”

“先生,”马可改口,“婉儿女史,这是不是一件很累的事?”

“很累。”她说,“可这不是最要紧的。”

“那什么最要紧?”

上官婉儿把那支朱笔又拿出来了。

神龙以后,我掌了另一样活。

“什么活?”

“评诗。”


“景龙年间,中宗常常大宴群臣,命百官作诗。”上官婉儿说。

“作完了呢?”

“作完了,我评。”

“怎么评?”

上官婉儿在一块石碑前站住,用手抚过上面的字。

“有一次在昆明池。中宗令群臣赋诗,我在彩楼上,把所有的诗都拿上来。”

“我看一首,评一句,不入选的,”

她做了一个动作:手一扬。

“扔下楼。”

马可怔住了。

“扔下去?”

“扔下去。”她说,“底下站着一百多个官。他们看着自己的诗纸,一张一张从楼上飘下来。”

“那些纸落在地上,谁都能看见。”

马可只觉得胸口发闷。

“最后剩下几张?”

“最后剩下两张。”上官婉儿说,“沈佺期的,宋之问的。”

“两张一样好?”

“不一样。”她说,“我评了宋之问的第一。”

“为什么?”

上官婉儿抬起眼。

“沈佺期最后两句是:‘微臣雕朽质,羞睹豫章材。’”

“宋之问最后两句是:‘不愁明月尽,自有夜珠来。’”

她看着马可:“沈的收得弱了。他在自谦,气就往下走。”

“宋的收得开。‘自有夜珠来’,月亮落了,还有夜明珠。”

“所以我评宋第一。”

“女史,”马可忍不住问,“这两句差得,有那么多吗?”

“不多。”上官婉儿说得很干脆,“差半分。”

马可怔住了。

“那您为什么要评出个第一?”

“因为中宗要一个第一。”她说,“他设了宴,要看百官献诗,最后要赏一个人。”

“他不要‘两首都好’。”

她把朱笔收起来:“年轻人,你们那一行有没有这样的事:两样东西差半分,可你必须选一个?”

马可想都没想:“天天都有。”

“那你怎么选?”

“我们会找一条能说得出口的理由。”马可说,“不一定是真正的理由。”

上官婉儿点了点头。

我那一天说的‘气’,就是一条说得出口的理由。


马可站在那儿,久久说不出话。

不是因为这个评判有什么问题。而是因为他忽然听懂了另一层东西。

“婉儿女史,”他慢慢地说,“您那一天,不只是在评两首诗。”

“说下去。”

“您在,教这一百多个人,往后该怎么写诗。”

上官婉儿的眼睛亮了一下。

“对。”

“他们回去以后,”马可说,“每一个人都会想:为什么宋之问的入了,我的被扔了?”

“然后他们会去研究宋之问那两句。”

“然后他们下一次写,就会照着那个方向写。”

他抬起头:“您改了整个朝廷写诗的方向。

上官婉儿把朱笔在手里转了一圈。

“年轻人,”她说,“你比大部分读那段故事的人,读得深。”

“大部分人读那一段,读的是,一个女人在楼上扔纸,多威风。”

“那不是威风。”她说,“那是,权。


“婉儿女史,”马可忽然问,“那您评得公吗?”

上官婉儿没有生气。

她甚至笑了一下。

“大体上公。”她说,“沈宋那一次,我评得对。宋之问那两句是好的,一千年来没人翻我这个案。”

“那不公的时候呢?”

“不公的时候也有。”她说得很直接,“有些人的诗我抬了,因为那是安乐公主要抬的人。”

“有些人的诗我压了,因为那是不能抬的人。”

马可沉默了。

“年轻人,你现在是不是觉得,那她那套评法,就不算数了?”

“……我不知道。”

“算数。”上官婉儿说,“这才是最狠的地方。

“怎么讲?”

她转过身,看着那一片石碑的森林。

“我评一百首,其中九十首是照才评的,十首是照人评的。”

“可外面的人不知道是哪十首。”

“所以他们只能,把那九十首和这十首一起学。”

马可后背一凉。

“他们学的东西里头,”他艰难地说,“混着假的。”

“对。”上官婉儿说,“而且他们分不出来。”

“年轻人,一套标准最厉害的时候,不是它全对。”

她一字一句地说:

是它对得足够多,让人不敢怀疑那些不对的。

马可站在石碑之间,忽然想起了另一件事。

“女史,”他说,“那些被扔下楼的诗,里头有好的吗?”

上官婉儿沉默了。

那是她今夜第一次真正的沉默。

“有。”她终于说。

“几首?”

“我记得两首。”她说,“一首是一个从岭南来的九品小官写的。我扔的时候手停了一下。”

“为什么扔?”

“因为他那首写得太生。”上官婉儿说,“不合当时的格。用的字硬,对仗也不齐。”

“可它好?”

“它好。”她说,“它里头有一样东西,那一百多首里只有它有。”

“什么东西?”

上官婉儿望着黑暗中的石碑。

“他真的看见过海。”

马可站在那儿,一句话也说不出来。

“那一天我要是评了他,”上官婉儿说,“那一百多个人回去以后,学的就是另一样东西了。”

“那您为什么没评?”

“因为我不敢。”她说得很干脆,“评一个九品小官第一,中宗会问我为什么。”

“而我答不上来,我没有一条说得出口的理由。”


马可在石碑之间走了很久。

他忽然想起自己那一行的一样东西。

评级。

他这十几年,无数次用过标普、穆迪、惠誉的评级。AAA、AA、A、BBB。他知道这些评级里有多少水,他知道二〇〇七年那些次级贷打包出来的东西被评了AAA,他知道评级机构是被发行方付钱的。

他知道这套标准有问题。

可他还是用它。

因为不用它,他就没有第二套可用的东西。

“婉儿女史,”他忽然说,“我们那一行也有一套这样的东西。”

“讲。”

马可把评级机构的事讲了一遍。讲发行方付费,讲二〇〇八年,讲那些被评了最高等级、后来一文不值的东西。

上官婉儿听完,问了一个问题:

“那一年之后,你们废了它吗?”

“没有。”马可说,“罚了钱,改了一些规则。可那三家还在。”

“为什么不废?”

马可想了很久。

“因为,”他慢慢地说,“没有它,我们就没办法比较了。”

“一个基金经理要在几千个债券里挑,他不可能每一个都自己去看。他需要一个,可以偷懒的东西。”

上官婉儿点了点头。

这就是我这一辈子那支笔,真正的分量。”她说。

“不是因为我评得准。是因为,天下那么多人写文章,皇帝一个人读不完。”

“他需要一个,可以偷懒的东西。”

她把朱笔举起来:“而那个东西,就是我。”

“女史,”马可忽然问了一个很唐突的问题,“那您怕不怕?”

“怕什么?”

“怕有一天,有人发现您评错了。”

上官婉儿笑了。

“年轻人,我死于景云元年。”她说,“临淄王起兵那一夜,我拿着中宗的遗制去见他,那份遗制上有立相王辅政的话,是我写的。”

“我以为那能救我。”

“救了吗?”

“没有。”她说,“他看了,说了一句‘此婢负国久矣’,让人把我斩了。”

马可闭上了眼睛。

“我四十六岁。”上官婉儿说。

她把朱笔在手里转了一圈:“所以你问我怕不怕有人发现我评错了,不怕。”

“我怕的不是这个。”

“那您怕什么?”

上官婉儿抬起眼,看着他。

我怕我这一辈子握着这支笔,最后什么也没定下来。

她望着那一片石碑:“我评了几千篇文章。可我自己的集子,二十卷,散了。今天存下来的,只有三十几首。”

“年轻人,这才是真正冷的地方:我一辈子都在定别人的等,可定我的等的那支笔,从来不在我手上。


天快亮的时候,马可问了那个他一直想问的问题。

“婉儿女史,那我那三十七页……为什么不入等?”

上官婉儿沉默了一会儿。

“我先说它好的地方。”她说。

“它有好的地方?”

“有。”她说,“数是真的。逻辑是通的。图是清楚的。要是拿来考科举,能中。”

“那为什么不入等?”

上官婉儿抬起那支朱笔,在空中虚划了一下。

“因为它答的不是一道题。”

马可怔住了。

“怎么讲?”

“一篇能入等的文章,”上官婉儿说,“不管长短,从头到尾只答一个问题。”

“你那三十七页,答了十一个问题。”

她放下笔:“陕西苹果产量多少、加价倍数多高、冷链缺口多大、组织化程度多低,这是十一件事。”

十一件事,等于没有一件事。

马可站在那儿,一句话也说不出来。

“年轻人,你在那间屋子里被十五分钟送出来,”上官婉儿说,“你以为是因为你讲错了受益人。”

“那是一层。”

“还有一层是,你压根没有一个题目。”


“那我该怎么办?”马可问。

上官婉儿看着他。

“你现在,把你要做的这件事,用一句话说给我听。”

马可开口了。

他讲了大概四十秒。讲耐心资本,讲果农被压价,讲三年不结果的条款,讲他在成都做了七个村。

上官婉儿抬手。

“停。”

马可停住。

“太长。”她说。

“可我还没讲完……”

“正是因为你还没讲完。”上官婉儿说,“一句话讲不完的东西,就是你自己还没想清楚。”

她把朱笔递了过来。

“拿着。”

马可接过那支笔。笔很轻,杆上有很多年握出来的凹痕。

“你回去,”上官婉儿说,“把你要做的事,写成一句话。”

“不许超过二十个字。”

“二十个字?”

“二十个字。”她说,“十七年后你的这条渠要是成了,后人只会记住二十个字。”

“你现在写不出来,往后别人会替你写。”

她转过身,往碑林的深处走。

“而别人替你写的那二十个字,”她的声音传过来,“十有八九,不是你要的那个意思。”


马可在本子上写:I met the woman who decided, for a generation, which poems counted.

Shangguan Wan'er. Her grandfather Shangguan Yi drafted an edict to depose Empress Wu; he and her father were executed for it; she was born that year and taken into the palace service quarters as a bondservant, where her mother taught her to read. At fourteen Wu Zetian heard there was a girl in the laundry quarters who could write, set her a topic, read the result, and said four words: this girl has talent. The woman who killed her grandfather is the woman who made her career. She stated both facts in the same tone. Her own comment: most people arriving at this point pick a story — either she endured nobly or she served her family's murderer — because both are convenient. The truth she offered instead: at fourteen there was exactly one road, and someone was standing on it.

Then forty years of drafting: edicts, rescripts, patents of nobility. She wrote for the Emperor, the Empress, and two princesses simultaneously, in four distinguishable voices, and the whole capital recited the results without knowing they came from one hand.

But the part that matters is the second job. At the Kunming Pool banquet the officials all composed poems and she sat in the pavilion and judged them, throwing the rejected sheets off the balcony while more than a hundred men watched their own paper drift down. Two were left. She ranked Song Zhiwen above Shen Quanqi, and her reason was technical and, a thousand years later, uncontested: Shen closed on self-deprecation and the energy sank; Song closed with "the moon may set, but the night-pearls come" and the energy opened.

Here is what I finally understood standing there. She was not judging two poems. She was instructing a hundred men in how to write for the next twenty years. Every one of them went home and studied the sheet that survived.

And then she said the thing I have not stopped turning over. I asked whether she judged fairly. Answer: mostly. Ninety in a hundred on merit; ten on whose client they were. And the reason that made the system stronger rather than weaker — nobody outside could tell which ten. So they had to learn all hundred, contaminated and clean together, indistinguishably. Her formulation: a standard is at its most powerful not when it is entirely correct, but when it is correct often enough that no one dares question the parts that aren't.

I told her about the rating agencies. Issuer-paid, AAA on subprime tranches, 2008, fines, rule changes, and all three still standing. She asked why we didn't abolish them. I said: because without it we can't compare. A manager choosing among thousands of bonds cannot read every one. He needs something he can be lazy with. She nodded and said that is exactly what her brush was — the Emperor could not read every writer in the empire either, and what he needed was something he could be lazy with, and that something was her.

Finally, the verdict on my thirty-seven pages. Not "unranked" because the numbers are wrong — she says the numbers are fine and the logic holds. Unranked because it answers eleven questions, and eleven questions is the same as none. She would not let me talk for more than forty seconds about what I am doing. Instruction: write it as one sentence, twenty characters maximum. Her closing line, walking away: in seventeen years, if this works, people will remember twenty characters. If you don't write them, someone else writes them for you — and eight or nine times out of ten, not in the sense you meant. — M.


段四·钩子结尾

天亮了。

马可一个人站在碑林里。

手里握着那支朱笔。

他抬头看四周。一排一排的碑,一片石头的森林。每一块碑上,都是被人选中、被人刻下、被人抄了一千多年的字。

他忽然想起一件事,他这几天在西安,见了六个人。

杜甫,一辈子没进去。 司马迁,写了正史里唯一一篇商人列传。 白居易,写了五十首给皇帝看的诗,可能一首都没送到。 王维,靠一首在牢里写的诗保住了命。 杜牧,写了一篇关于不存在的宫殿的赋。 玄奘,建了十个位置。

这六个人,全都在跟同一件事较劲:怎么让自己的东西,留下来。

而上官婉儿,是那个决定谁能留下来的人

马可捏着那支朱笔,忽然出了一身冷汗。

他这一路,一直以为自己在学“怎么把事做成”。

可他从来没有想过一个问题:

这件事做成了以后,是谁来决定它算不算成?


他掏出手机,翻开备忘录,新建了一条。

他打了几个字,删了。

又打,又删。

太阳升起来的时候,他终于打出了一句:

让果农有资格等一等。

十个字。

他盯着这十个字看了很久。

不完整。它没说清怎么等,没说钱从哪来,没说他自己挣什么。

可他忽然觉得,对了。

他把这十个字存下来,标题写:第一句

然后他又想起上官婉儿说的那两首被扔下楼的诗。

他忽然打开备忘录,新建了第二条。

标题:错题簿。

第一条:

第一错:三十七页材料,答了十一个问题。 谁发现的:上官婉儿。 为什么没有早点发现:因为我做那份材料的时候,目的不是想清楚,是想让人觉得我做足了功课。

他打完,停了很久。

然后他在第三行后面又加了一句:

补:这三十七页里没有一个人的名字。这两件事是同一件事,一个只想让人觉得自己做足了功课的人,不需要名字,只需要样本量。


他走出碑林的时候,门口站着杜甫。

“老弟。”

“先生。”马可举起那支朱笔,“上官女史给我的。”

杜甫看了一眼,点了点头。

“她给你评了几等?”

“不入等。”

杜甫笑了。那是他这几天笑得最真的一次。

“好。”他说。

“好?”

“她这一辈子,评过几千篇文章。”杜甫说,“她肯说‘不入等’的,是她读完了的。”

“读不完的,她连话都不说。”

马可怔住了。

“老弟,”杜甫说,“今夜是你在长安的第八夜。”

“是。”

“还有两夜,第一关就到了。”

马可心里一动:“先生,第一关到底是什么?”

杜甫没有直接答。

他往前走了几步,忽然回过头。

“明夜,我带你去见一个人。”

“谁?”

“一个因为写了一封信,被贬到八千里外的人。”

杜甫望着渐亮的天:“他这一辈子,只做过一件真正大的事,他把整个天下写文章的法子,改了。”

“可他改的法子,是从得罪所有人开始的。”

他顿了顿:“老弟,你手上现在有一句话了。”

“明夜你要学的,是这句话说出去的代价。”


欲知后事如何,请看下回,《昌黎一表贬潮阳》

附录一·Dongpo's Note

Foreign friend — Shangguan Wan'er writing, and I will use fewer words than you deserve, because economy is the whole of my subject.

I did not soften my own history for you and I will not soften it here. My grandfather drafted the edict to depose the Empress. He and my father were killed for it. I was born that year into bondage in the palace quarters, and at fourteen the same Empress heard there was a girl who could write, set me a topic, and on reading it said four words and lifted me out. Both of those women were her. Choose neither of the tidy stories: I had one road, and someone was standing on it.

What I want you to take is not the biography. It is the brush.

At the Kunming Pool I sat above a hundred officials and dropped the failed poems off the balcony where every man could watch his own paper fall. Two survived. I ranked Song above Shen because Shen closed on modesty and let the energy sink, and Song closed with night-pearls after the moon and let it open. That judgment has stood a thousand years. But the poems were never the point. A hundred men went home and studied the sheet that survived, and I had changed how the court would write for twenty years.

And you asked, correctly, whether I was fair. Ninety of a hundred on merit; ten because of whose people they were. Understand why that makes the standard stronger and not weaker: no one outside could tell which ten. So they learned all hundred together and could not separate the true from the arranged. A standard's greatest power is not being wholly right. It is being right often enough that nobody dares doubt the rest.

Your thirty-seven pages: the figures are sound, the reasoning holds, and it would pass an examination. It is unranked because it answers eleven questions, and eleven is the same as none. You then spoke for forty seconds when I asked what you are doing, which told me you have not decided.

So: one sentence, twenty characters. In seventeen years, if this works, that is all anyone will retain. Write them yourself. Otherwise someone writes them for you, and they will not mean what you meant.


附录二·Marco's Memo

Memo to file. Beilin (Stele Forest), Xi'an, overnight.

Subject: rating power. The scarce asset is not the judgment — it is control of how the judgment is computed.

1. The Stele Forest is not a museum of calligraphy. It is a repository of standards. Yan Zhenqing, Ouyang Xun, Chu Suiliang, Liu Gongquan — a thousand years of Chinese children have practised these hands. Many Tang calligraphers were excellent. These are the ones somebody cut into stone.

2. Kunming Pool, 709. She read, ranked, and physically dropped the rejects from the pavilion in front of the authors. Her ranking of Song Zhiwen over Shen Quanqi was made on the closing couplet and has never been overturned. The second-order effect is the real one: a hundred men then reverse-engineered the surviving sheet. She reset the composition norms of an entire court in one afternoon without writing a poem.

3. Her disclosure, unprompted: 90/100 judged on merit, 10/100 on patronage. And the mechanism by which this strengthens rather than discredits the system — the outside cannot identify which ten, so the corrupted and clean cases must be learned together as one body. "A standard's greatest power is not being wholly right; it is being right often enough that nobody dares doubt the rest." This is the most exact description of a rating agency I have ever encountered, and it predates one by 1,300 years.

4. Why we did not abolish the agencies after 2008, honestly stated for the first time in my own notes: a manager screening thousands of instruments cannot underwrite each one. He requires something it is safe to be lazy with. Her reply: that is precisely what she was to the throne. The demand for a standard is a demand for permitted laziness, and whoever supplies it holds the actual power.

5. Verdict on the thirty-seven pages. Numbers sound, logic intact, would pass an exam. Unranked because it answers eleven questions. Her formulation: eleven questions equals none. She cut me off at forty seconds when I tried to state my purpose aloud, on the grounds that a thing you cannot say in one sentence is a thing you have not yet decided.

6. Assignment: one sentence, twenty characters. Because in seventeen years the residue is twenty characters, and if I don't write them somebody else will, in a sense I did not intend.

First attempt, 06:50, ten characters: 让果农有资格等一等. Give the grower the standing to wait. Incomplete — says nothing about how, or the funding, or what I earn. Filed as "Sentence One" anyway.

7. Structural note, and I want it on the record. Six figures in eight nights, every one of them fighting the same fight: how to make the thing they built outlast them. She is the one who decided who got to. I have spent this entire trip asking how to make something work and never once asked who adjudicates whether it worked.


附录三·English Highlight

"Writing does not exhaust speech; speech does not exhaust meaning."
Shangguan Wan'er (664–710) was born the year her grandfather, the chief minister and poet Shangguan Yi, was executed for drafting an edict to depose Empress Wu. She and her mother were taken into the palace service quarters. At fourteen the Empress — the same woman — heard there was a girl there who could write, tested her, and installed her in charge of documents. She spent the next four decades drafting imperial edicts, and for a period wrote simultaneously in the voices of the Emperor, the Empress, and two princesses, in styles nobody could trace to a single hand.
Her second office is the one worth studying. At the Kunming Pool banquet of 709, the assembled officials composed poems and she judged them from a pavilion, dropping the rejected sheets off the balcony in view of their authors. Two remained. She ranked Song Zhiwen first over Shen Quanqi on the strength of the closing couplet: Shen's ended in self-deprecation and let the poem sink; Song's — the moon may set, but night-pearls will come — opened it out. The judgment has stood.
But she was not ranking two poems. A hundred officials went home and studied the surviving sheet, and the court's compositional norms shifted for a generation. That is the difference between having an opinion and owning a standard.
Asked whether she judged fairly, she gave a figure: ninety of a hundred on merit, ten on patronage — and then the point that outlasts her, which is that nobody outside could tell which ten, so all hundred had to be learned together. A standard is most powerful not when it is entirely right, but when it is right often enough that no one dares doubt the rest.
Any modern reader who has used a credit rating knows this feeling. The demand for a standard is a demand for permitted laziness. Whoever meets that demand holds the power, and accuracy is only the price of admission.

附录四·Dongpo's Cheat Sheet

今夜之事 出处 现代对应 一句话
碑林放的是标准 西安碑林 谁被刻上石头 写得好的很多,能立在这儿的只有几个
杀祖父者提拔她 麟德元年上官仪案 无从选择的处境 十四岁只有一条路,路上有人等着
一人四声口气 《旧唐书》婉儿传 代笔与影响力 全城在背,没人知道是一只手
昆明池扔诗 景龙间彩楼评诗 评级即定向 她不是在评两首诗,是在教一百人怎么写
九十真十假 婉儿自陈 标准的可信度机制 对得足够多,人就不敢怀疑不对的
可以偷懒的东西 评级需求的本质 委外判断 需要标准,就是需要一个偷懒的凭据
十一个问题等于没问题 三十七页评语 论点收敛 一句话讲不完,就是你还没想清楚
二十个字 婉儿留题 定义权 你不写,别人替你写,而且写歪