卷:卷二《长安月下·关中六十夜》|主角:刘晏·Marco|地点:西安·广运潭遗址(今浐灞)·冬夜|典:《旧唐书·刘晏传》“诸道各置知院官,每旬月具州县雨雪丰歉之状白使司”“晏常以厚直募善走者,置递相望,四方物价之贵贱,虽极远不四五日知”“因民所急而税之,则国用足”“知所以取人不怨”|主导感官:听(水声、驿马的铃、算珠、雪打在冰上)|过所进度:三字·「客」「侣」「本」|金融-国学对应:Information Is the Cheapest Infrastructure(最便宜的基建是信息;四五天知道天下物价,比多修一座仓管用)↔ 四方物价之贵贱,虽极远不四五日知

段一·现身

那条河,冬天冻住了。

第九十二夜。

一月九日。

离二月十号,还有三十二天。

马可又去了西安。

这一次他有正经事,他去见了一个人。

那个人是省农科院的一个研究员,姓吴,四十来岁。

是小张介绍的。

小张在电话里说得很含糊:「吴老师做冷链的,您跟他聊聊,也许有用。」


他们在一家咖啡馆见的面。

吴老师带了一个笔记本电脑。

他一开口,马可就知道这个人是真懂的。

“马先生,我先给您看几个数。”

他打开一个表。

陕西一年产苹果一千三百万吨左右。

贮藏能力,气调库加冷库,大概六百万吨。

马可怔了一下。

将近一半?

“对。”吴老师说,“从库容上看,我们不缺库。

马可停住了。

“吴老师,那洛川那个库,”

“空了三年?”吴老师笑了一下。

“您知道?”

这样的库,全省有的是。


马可在那间咖啡馆里坐了两个小时。

他后来在本子上记了三条:

一、全省库容不缺。缺的是「库怎么用」。

二、很多库建起来以后,实际使用率不到四成。

三、吴老师说了一句话:

马可把那句话原样抄了下来。

「建库是一次性投入,有指标,好考核。」

「用库是常年的事,没指标,不好考核。」

「所以我们这儿,库年年建,用得不多。」


马可从咖啡馆出来,天已经黑了。

他没有回去。

他在西安东边一直走。

走到了浐河边上。

冬天,河面结着薄冰。

那一带现在叫浐灞,有一片湿地公园,冬天没什么人。

可一千二百多年前,这儿叫广运潭。

天宝二年,韦坚在这儿开了一个大潭,各地的船从这儿进长安。

马可站在河边。

河面上的冰,反着远处高楼的光。


段二·古文镜像

“晏常以厚直募善走者,置递相望,四方物价之贵贱,虽极远不四五日知。”
English Mirror: Liu Yan hired fast runners at high wages and set relay posts within sight of one another, so that the prices of goods in every quarter — however distant — were known within four or five days.

“你在看什么?”

马可回头。

一个人站在河堤上。

六十来岁。穿着唐代官员的常服,深绯色,可衣服很旧,袖口磨得发亮,跟他那个品级不太相称。

他手里拿着一个算袋。

那种装算筹的布袋,鼓鼓的。

“先生您是,”

“刘晏。”那人说,“字士安。”

马可怔住了。

唐代宗时的理财家。

安史之乱以后,重整漕运、盐政、常平的那一个。

“刘公,”

“别叫公。”刘晏说,“我七岁就做了官,一辈子被人叫惯了,听着累。


“七岁?”

“开元年间,玄宗封泰山。”刘晏说,“我那时候七岁,献了一篇《东封书》。”

“皇帝怎么说?”

“他让我做了个太子正字。”

马可算了一下。

七岁的九品官。

“先生,那是,”

“那是个玩意儿。”刘晏说得很干脆,“皇帝身边有个七岁的神童,好看。

他把算袋在手里掂了掂。

我真正干活,是四十岁以后。

“那中间三十几年呢?”

刘晏想了想。

在县里。

“做什么?”

“做过夏县令,做过温县令。”他说,“管过一个县的钱粮。

马可点了点头。

“先生,那后来您管天下的钱,”

跟管一个县,是一回事。”刘晏说得很干脆。

马可怔住了。

“怎么会是一回事?一个县几万人,天下几千万人。”

刘晏看了他一眼。

人多了,数大了。可要问的问题,一个字没变。

“什么问题?”

他伸出三根手指:

东西在哪儿。什么时候会缺。缺的时候从哪儿调。

他把手放下:

我在夏县问这三个问题,在长安还是问这三个。


段三·事件主体

“先生,”马可说,“我今天听了一天数。”

“讲。”

马可把吴老师那三条说了。

全省库容不缺。使用率不到四成。建库有指标,用库没指标。

刘晏听完,一点也不意外。

这个我熟。

“先生也遇到过?”

刘晏在河堤上找了块石头坐下。

“年轻人,我接手漕运那年,是宝应二年。”

“那是什么情形?”

刘晏想了很久。

关中饿。

“多饿?”

“长安米价,一斗一千钱。”刘晏说,“平常年景,一斗三十钱。

马可算了一下。

三十三倍。


“为什么这么贵?”

“江淮的米运不上来。”刘晏说。

“路断了?”

“路没断。”刘晏说,“路一直在。

“那为什么运不上来?”

刘晏在地上画了一条线。

从扬州到长安,走漕渠,两千多里。

“中间要过江,过淮,过汴,过黄河,最后进渭。”

他在那条线上画了几个点。

每一段的水,脾气不一样。

“怎么讲?”

“江里的船,到了汴河搁浅。汴河的船,进了黄河翻。黄河的船,到了渭水又太大。”

马可听懂了。

“所以每一段要换船。”

“对。”刘晏说,“可从前的规矩是:一条船,从扬州装到长安。


马可怔住了。

“一条船走全程?”

“一条船走全程。”

“那,”

那就是照着最难走的那一段造船。”刘晏说。

“结果呢?”

“结果那船在江里太小,在黄河里太脆。”

他停了一下。

一年翻掉的粮,有四成。

马可站在那儿,一句话也说不出来。

“先生,那您怎么改的?”

刘晏用手在那条线上一段一段地划。

分段。


“怎么分?”

“江船不入汴,汴船不入河,河船不入渭。”刘晏说。

“每一段用每一段的船?”

“对。”

“那货怎么办?”

在交界处设仓。”刘晏说,“江船到了,卸进仓。汴船来了,从仓里装。”

马可拿出本子,飞快地记。

这是转运。

这是他这一行说的:分段运输加中转仓储。

“先生,这样不是更麻烦吗?多了装卸。”

刘晏看了他一眼。

多了装卸,少了翻船。

他说得很实在:

我改完以后,一年损耗,从四成,降到不到一成。


马可在那河堤上坐了下来。

“先生,那这一段我懂了。”他说,“可我想问另一件事。”

“讲。”

您那些仓,都用起来了吗?

刘晏的表情变了一点。

那是马可今夜第一次看见他停顿。

你问到点子上了。


“怎么讲?”

“年轻人,”刘晏说,“仓好建。

“为什么好建?”

因为建仓要钱,要地,要工。这三样都是可以报上去的。

“报上去以后呢?”

“报上去以后,上头拨钱。仓建起来,我上报:某处建仓若干。”

他把算袋放在膝上。

这是一件干干净净的功劳。

马可点了点头。

他今天上午刚听吴老师说过一模一样的话。

“那用仓呢?”

刘晏看着那条冻住的河。

用仓是天天的事。

“怎么讲?”

“今天进多少,明天出多少,什么时候该买,什么时候该放。”

“这些事,”他说,“没有一件是可以报上去的。


马可在河堤上坐着,忽然浑身一震。

“先生,那您怎么办的?”

刘晏把那个算袋打开了。

里头不是算筹。

是一叠很薄的纸片。

马可凑近看,那些纸片上写着字,很小,密密麻麻。

“这是什么?”

刘晏抽出一张。

这是宝应三年三月,宋州的一张。

他念了起来:

宋州,三月,雨足。麦长势中。米价斗四十三。

同月,颍州,旱。米价斗六十一。

马可听得手心冒汗。

“先生,这是,”

知院状。”刘晏说。


“什么是知院状?”

“我在各道设了知院官。”刘晏说,“每一个人管一片。”

“他们干什么?”

每旬每月,把本州县的雨雪、丰歉、物价,写成一状,报上来。

“报到哪儿?”

“报到我这儿。”

马可算了一下。

先生,那是多少个州?

“一百多个。”

“每旬一报?”

“每旬一报。”

马可站起来了。

先生,那是一年,一万多份?

刘晏点了点头。

差不多。


马可在那河堤上走了两步。

“先生,这些报上来,您怎么看得过来?”

“看不过来。”刘晏说得很干脆。

马可怔住了。

“那报它干什么?”

刘晏把那叠纸片举起来。

“年轻人,我不是每一份都看。”

“那您看什么?”

我看变的那些。


“怎么讲?”

“宋州上个月米价四十三,这个月四十一,这个不用看。”刘晏说。

“那看什么?”

上个月四十三,这个月六十一,这个要看。

马可站在那儿,忽然全明白了。

这不是全量阅读。

这是异常检测。

“先生,”他说,“那您看见变的了,怎么办?”

刘晏很干脆:

买或者卖。

“就这么简单?”

“就这么简单。”

他停了一下。

难的不是这一步。


“那难在哪儿?”

刘晏站起来,走到河边。

河面的薄冰在灯光下泛着一点白。

难在,我得比商人先知道。

马可怔住了。

“为什么要比商人先?”

“年轻人,”刘晏说,“一个地方旱了,谁先知道?

马可想了想。

“……本地人。”

“对。然后呢?”

然后是本地的粮商。

“对。”刘晏说,“粮商知道了,他就囤。

“囤了以后价更高。”

“对。”

他转过身:

所以我要是等到米价涨上去才知道,那我就晚了。


“那怎么办?”

刘晏说了一句让马可后来在本子上写了整整一页的话:

我雇跑得快的人。

马可怔了一下。

“跑得快的人?”

厚直募善走者。”刘晏说,“我出高价,雇能跑的人。”

“雇来干什么?”

置递相望。

“什么意思?”

刘晏在地上画了一串点。

每三十里设一个递铺。

“一个站到下一个站,站在这儿能看见那儿。”

“然后呢?”

一站一站往下传。

他抬起头:

四方物价之贵贱,虽极远不四五日知。


马可站在那河堤上,忽然出了一身冷汗。

天下最远的地方的物价,四五天之内传到长安。

在八世纪。

“先生,”他艰难地问,“这要花多少钱?”

刘晏想了想。

不多。

“不多?”

“比一座仓便宜得多。”刘晏说。

马可怔住了。

“比一座仓还便宜?”

“年轻人,你算算。”刘晏说,“一座大仓,要地,要木石,要工,要几年。”

一个递铺,两间屋,两个人,几匹马。

他伸出手,比了一个数:

我在几条主要的道上设递,一年的钱,不到一座大仓。


马可站在冬天的河边,脑子里在飞快地转。

他忽然想起老周那个微信群。

四百八十七个人。凌晨三点到五点。

「白水今日七毛九,货一般。」

「洛川塬上还有货,八毛五能收。」

免费。而且比任何一个终端都快。

“先生,”他忽然说,“我们那儿现在有这个东西。”

“讲。”

马可把那个群讲了。

刘晏听完,问了一个问题。

那个群里,有种苹果的人吗?

马可停住了。


他站在那儿,一句话也说不出来。

那个群里四百八十七个人。

全是收苹果的。

没有一个是种苹果的。

“先生,”

“年轻人,”刘晏说,“你现在懂那一块九毛三是怎么来的了吗?

马可站在河堤上,忽然明白了一件他这半年一直在绕的事。

不是中间商吃掉了那一块九毛三。

老周挣五分。

可老周知道白水七毛九、静宁下雪、广州走货慢。

而强东林什么也不知道。

他只知道,明天有一个人会来他院子里,报一个数。


“先生,”马可说,“那这个差别值多少钱?”

刘晏看了他一眼。

你自己算。

马可蹲下来,在河堤的雪地上算。

强东林那三万斤,卖八毛七。

要是他知道当天白水是七毛九、洛川塬上普遍八毛五、而这个客商在别的村已经收够了、还差一车,

他能不能多要两分?

三万斤 × 两分 = 六百块。

六百块。

他抬起头。

“先生,一户六百块。”

“那一个村呢?”

马可算:四百户 × 六百 = 二十四万。

“那一个县呢?”

马可停住了。

洛川有六万多户果农。

六万 × 六百 = 三千六百万。


马可猛地站起来。

三千六百万。

跟省上那笔项目款,是同一个数量级。

“先生,”他的声音有点抖。

刘晏点了点头。

“年轻人,你现在知道那三十个库,最要紧的不是库了吧?”

是那个群。

“对。”刘晏说,“是让种的那些人,也进那个群。


马可在河堤上站了很久。

然后他忽然想到一个问题。

“先生,可这个事,”

“讲。”

这个事没法立项。

刘晏笑了。

“为什么?”

“因为它不花钱。”马可说,“一个不花钱的事,报上去,谁给你批?

刘晏在那儿站了很久。

然后他说了一句极其冷的话:

年轻人,我那些递铺,从来没有单独立过项。

马可怔住了。

“那您那钱从哪儿出的?”

刘晏看着他:

从盐利里出的。

“盐利是干什么的?”

盐利是给朝廷用兵、赈灾、发俸禄的。

他停了一下。

我挪了一点,拿去养那些跑腿的。


马可站在那儿,一时不知道该说什么。

“先生,那要是有人查,”

“查过。”刘晏说得很平常。

“怎么样?”

我说不清。

马可闭上了眼睛。

“因为那些钱,没有一张凭据能说明它换来了什么。”刘晏说。

“可它换来了整个天下的物价。”

那个东西,没法入账。

他把那叠纸片收回算袋里。

“年轻人,我最后是被赐死的。”

“罪名是什么?”

刘晏想了想。

罪名很多。”他说,“杨炎说我参与了废太子的事。

“那是真的吗?”

刘晏沉默了很久。

不是。

“那为什么定了?”

他望着那条冻住的河。

因为我管了十几年的钱,账上有很多说不清的地方。

一个说不清的人,你要定他什么罪,都定得上。


段四·钩子结尾

天很冷。

刘晏要走了。

“年轻人,我给你三句话。”

“先生请讲。”

第一,建仓好报,用仓难报。所以天下到处是空仓。

马可点头。

第二,最便宜的东西是信。

“信?”

信息。”刘晏说,“一个递铺两间屋两个人,比一座仓便宜一百倍。”

“可它救的粮,比那座仓多。”

马可在本子上飞快地记。

第三句最要紧。

“先生请讲。”

刘晏转过身,看着他。

我这一辈子最大的成就,不是漕运,不是盐法。

马可怔住了。

“那是什么?”

刘晏说了八个字:

知所以取人不怨。


“什么意思?”

懂得从哪儿取,人就不怨。

“怎么讲?”

刘晏在河堤上走了两步。

“年轻人,朝廷要钱。钱从哪儿来?从百姓身上来。”

“这个躲不掉。”

可从哪儿取,怎么取,取多少,这里头差得很远。

“比方说?”

从前收盐税,是在盐场收,也在市上收,还在过关的时候收。

“一路收?”

“一路收。”刘晏说,“百姓买一斤盐,不知道自己交了几道税。

“那您怎么改的?”

只在盐场收一次。

马可停下笔。

“就一次?”

“就一次。”刘晏说,“盐出了场,商人爱怎么运怎么运,爱卖多少卖多少,我一概不管。


“先生,那朝廷不是少收了?”

刘晏摇头。

多了。

“为什么?”

“因为从前一路收,路上有多少道关,就有多少人伸手。”刘晏说,“真正到朝廷手里的,不到三成。

“改了以后呢?”

改了以后,一次收清,全进国库。

他抬起眼:

天宝末年,盐利一年四十万缗。到我手上第十年,六百万缗。

马可算了一下。

十五倍。

“先生,可百姓的负担,”

百姓交的,比从前少。”刘晏说。

“怎么可能两头都好?”

刘晏看着他,说了今夜最后一句:

因为从前那些钱,大半没进国库,也没留在百姓手里。

它散在路上了。


马可站在那条冻住的河边,很久没有动。

他忽然明白了那一块九毛三。

那一块九毛三,大半也不在任何一个人的口袋里。

它散在路上了。

散在运费里,纸箱里,摊位费里,还有那百分之十二烂掉的苹果里。

没有一个人拿走它。

它就是,没了。

他忽然想起自己在纽约算过的一个东西。

他们那一行管它叫“摩擦成本”。

买卖价差、佣金、冲击成本、税。

每一次交易都有,每一次都很小,谁也不在乎。

可他们组做过一个测算:一个换手率很高的组合,一年下来,摩擦成本能吃掉三到四个百分点。

而那三到四个点,不进任何一个人的口袋。

它就是,散在路上了。

他站在冬天的河边,忽然明白了一件事:

从八世纪的漕船到二十一世纪的交易台,中间隔了一千二百年。

可最大的那一笔支出,一直是同一个东西:

不是谁赚走的。是运输本身。


刘晏走了。

马可一个人站在浐河边上。

他掏出手机,给强磊发了一条微信。

强磊,我问你一件事。

你说。

你在那个收苹果的群里吗?

那边隔了一会儿。

我不在。

那个群是收货的人的群。

我们种的人进不去。

马可站在冬天的河边,慢慢地打:

咱们村四百来户,有多少人有微信?

那边回得很快。

基本都有。年轻人肯定有。老人也有一半。

马可看着那条冰河,打了下一句:

要是咱们自己建一个群呢?

那边过了很久。

建群干啥?

马可打字:

每天报一个数。

报啥数?

当天有客商来的人,报一句:今天几号,谁来收,出多少价,收了没有。


那边隔了很长时间。

久到马可以为他睡了。

然后消息来了。

马哥。

在。

这个不行。

马可怔住了。

为啥?

那边打了很久。

因为谁也不愿意说自己卖了多少钱。

马可站在那儿,忽然想起第七十一夜。

强东林那张化肥袋子上的账,他媳妇都不知道。

李叔那个十一年的本子,也是一个人记的。

这个塬上,没有人愿意告诉别人自己卖了多少钱。

他站在冬天的河边,慢慢地打了一行字:

那要是不用报名字呢?

那边过了一会儿。

咋个不报名字?

马可想了很久。

然后他打了一段:

你每天在群里发一句:今日××村,客商报价××,成交/未成交。

不说是谁家的。

你一个人发。

那边隔了很久。

那我从哪儿知道?

马可看着那条冻住的河,一个字一个字地打:

你去问。

每天走一趟,问三家。

不用问他卖了多少,问他今天有没有人来,报的什么价。


那边沉默了很长时间。

然后强磊回了一条:

马哥。

在。

这个事,得干几年?

马可站在那儿,忽然想起贾岛那句话。

两句三年得。

他打了三个字:

不知道。

那边又沉默了。

过了大概两分钟,来了一条:

行。

我明天开始。

然后又是一条:

那我今天先记一个:

今天没人来收。塬上没货了。

马可站在浐河边上,看着这条消息,忽然笑了。

这是那个群的第一条。

内容是:什么也没发生。

他把手机收起来,又拿出来,把那条消息截了个图。

存进相册。

他这一年存过三张图。

第一张是成都魁星楼街那条空街,什么也没拍到。

第二张是那块刚挂上去的铁牌,光线很暗,有反光。

第三张是今天这一条:今天没人来收。塬上没货了。

三张都不好看。

三张都是开头。


欲知后事如何,请看下回,《陆贽两税论利病》

附录一·Dongpo's Note

Foreign friend — Liu Yan. I was made an official at seven for presenting an essay when Xuanzong sacrificed on Taishan, which was a decoration, and I did no real work until I was past forty. Then I was given the grain transport in a year when rice in Chang'an stood at a thousand cash the dou against a normal thirty.

The old rule was that one boat carried the grain from Yangzhou to the capital, two thousand li through river, canal, Yellow River, and Wei — four waters with four tempers. So every hull was built for the worst reach of the journey, and was wrong for all the others, and four parts in ten went to the bottom every year.

I broke it into stages. River boats do not enter the Bian; Bian boats do not enter the He; He boats do not enter the Wei. Granaries at each junction, unload, reload. More handling, fewer wrecks — losses fell below one part in ten.

But you asked the question that matters: were the granaries used. Here is the shape of it, and it has not changed in twelve hundred years.

A granary is easy to report. It costs land, timber, and labour, all of which can be written down, submitted, and approved, and when it stands you memorialise: a granary has been built at such a place. That is a clean, finished piece of merit.

Using a granary cannot be reported. How much in today, how much out tomorrow, when to buy, when to release — not one of those is a thing you can put in a memorial. So the empire fills with granaries nobody opens.

What I did instead was cheaper than any of them. I posted knowledge officers in every circuit to send a statement each ten days: rain and snow, harvest, prices. Over a hundred prefectures, more than ten thousand statements a year, and no, I did not read them. I read the ones that had moved. Forty-three last month and forty-one this month is nothing. Forty-three to sixty-one is everything.

And because the local grain merchant learns of a drought before I do and begins to hoard, I could not afford to learn it from the price. So I hired fast runners at high wages and set relay posts within sight of each other, thirty li apart, and the price of goods anywhere in the empire reached me in four or five days. Two rooms, two men, a few horses — a hundredth of the cost of one granary, and worth more than several.

None of it was ever a project. There was no line for it. I took the money from the salt revenue, which was for armies and famine relief, and when they audited me I could not account for it, because nothing I bought with it could be shown. I was ultimately given death, on charges I did not commit, and the true reason is simpler: a man whose accounts contain things he cannot explain can be convicted of anything.

One last thing, and it is what I would keep if I could keep only one. Salt was formerly taxed at the works, in the markets, and at every barrier, and a peasant buying a jin of salt could not have told you how many times he had paid. I collected once, at the works, and let the merchants do as they liked thereafter. Revenue went from four hundred thousand strings a year to six million, and the peasant paid less than before.

Both, at once. Because most of what he used to pay had never reached the treasury and had never stayed with him.

It was lost along the road.

Your one yuan ninety-three is the same. Nobody took it.


附录二·Marco's Memo

Memo to file. Xi'an — a café, then the Chan riverbank at Guangyuntan. Night ninety-two. 9 January. 32 days to 10 February.

Subject: the cheapest infrastructure is information, and it cannot be funded.

1. Provincial capacity data, from a Shaanxi Academy of Agricultural Sciences researcher introduced by Xiao Zhang. Output ~13m tonnes/yr; controlled-atmosphere plus cold storage ~6m tonnes. Capacity is roughly half of output. The province does not lack stores. Realistic utilisation on many of them: under 40%.

His formulation, unprompted and identical to Liu Yan's twelve hundred years earlier: building a store is a one-off with an indicator and is easy to assess; operating a store is perennial, has no indicator, and cannot be assessed. Hence stores get built every year and are not used.

This invalidates the premise I have been working from since July. I have been treating the empty store as a local pathology. It is the standard output of the funding mechanism.

2. Liu Yan on stage-transport: one hull for a four-water route means every hull is built for the worst reach and wrong for the rest — 40% annual loss. Segmented with transfer granaries at each junction: under 10%. More handling, fewer wrecks. Standard logistics, eighth century.

3. The part I came away with. Knowledge officers in every circuit, ten-day returns on rain, harvest, and prices, 100+ prefectures, >10,000 statements a year. He did not read them. He read the deltas. 43→41 is noise; 43→61 is an event. Anomaly detection, not reporting.

4. Why speed mattered: the local grain merchant learns of a drought before the capital and begins hoarding, so learning it from the price is already too late. Solution: relay posts thirty li apart, within line of sight, fast runners at premium wages. Prices from anywhere in the empire in four or five days.

Cost: two rooms, two men, a few horses per post — his estimate is roughly one-hundredth of a large granary, and more effective.

5. The finding that reorganises my whole thesis, and I should have seen it in the paomo shop.

Zhou's WeChat group has 487 members and is faster and better than any terminal I paid for in New York. Liu Yan asked one question: are there any growers in it?

There are not. It is entirely buyers.

So the ¥1.93 spread is not middleman rent — Zhou earns five fen. The asymmetry is informational. Zhou knows Baishui printed 0.79, that it snowed in Jingning, that Guangzhou is moving slowly. Qiang Donglin knows that a man will arrive in his yard tomorrow and say a number.

Rough sizing of what that asymmetry costs: if a grower knew same-day regional prices and could hold two fen, that is ~¥600 per household on 30,000 jin. Luochuan has ~60,000 grower households. ≈¥36m — the same order of magnitude as the entire provincial programme, obtainable at approximately zero capital cost.

6. And it cannot be funded, for exactly the reason the store sits empty. It buys nothing. Liu Yan never had a line item for the relay posts either — he took it out of the salt revenue, could not account for it under audit, and was eventually executed on charges he says were false. His own diagnosis: a man with unexplainable items in his accounts can be convicted of anything.

Note carefully: he does not say he was innocent of irregularity. He says the irregularity was real and the thing it bought could not be entered.

7. Salt reform, kept for the principle: formerly taxed at works, market, and every barrier, so the buyer could not count how many times he had paid. He collected once, at the works, and released the trade. Revenue 400,000 → 6,000,000 strings; burden on the buyer fell.

Both, because most of what was formerly paid reached neither the treasury nor stayed with the payer. It was lost along the road.

That is the ¥1.93. Nobody took it.

8. Action initiated tonight, and it is free.

Proposed a grower-side price group to Qiang Lei. He rejected the first version immediately and correctly: nobody here will disclose what they sold for. Confirmed by everything since night seventy-one — the fertiliser sack his wife had never seen, Li's notebook kept alone for eleven years.

Revised: no names, no household figures. One person — him — walks the village daily, asks three households whether a buyer came and what was quoted, and posts a single anonymised line.

He asked how many years this has to run. I said I don't know. He said he'd start tomorrow.

First entry, tonight: no buyers came, no fruit left on the plateau.

The first record in the dataset is an absence, which is correct. A price series with no zero days is a series someone has curated.


附录三·English Highlight

"He hired fast runners at high wages and set relay posts within sight of one another, so that the price of goods in every quarter — however distant — was known within four or five days."
Liu Yan took charge of the Tang grain transport when rice in Chang'an had reached a thousand cash per dou against a normal thirty. His reform of the shipping is the famous part: a single hull had been sailing the whole two thousand li from Yangzhou through four different waters, which meant every boat was built for the worst reach of the route and wrong for all the others, and four parts in ten sank each year. He broke the journey into stages with transfer granaries at each junction — more handling, fewer wrecks, losses under ten percent.
The unfamous part is better. He installed knowledge officers across every circuit filing ten-day returns on weather, harvest, and local prices — over ten thousand statements a year, which he did not read. He read the ones that had moved. Forty-three to forty-one is nothing; forty-three to sixty-one is an event.
And because the local grain merchant learns of a drought before the capital does and starts hoarding, learning it from the price was already too late. Hence the relay posts: thirty li apart, each within sight of the next, running at premium wages. Two rooms, two men, a few horses — by his estimate a hundredth the cost of one granary, and worth more than several.
None of it was ever a funded project, because it purchased nothing. He paid for it out of the salt revenue, could not account for it when audited, and was eventually given death on charges he denies. His own explanation is not a protest of innocence: a man whose accounts contain items he cannot explain can be convicted of anything.
And the sentence he would keep above all his reforms — know where to take from, and the people will not resent it. Salt had been taxed at the works, in the markets, and at every barrier, so no buyer could count how often he had paid. Liu Yan taxed once, at the source, and released the trade. Revenue rose fifteenfold and the buyer paid less, because most of what he used to pay had reached neither the treasury nor stayed in his hands.
It was lost along the road.

附录四·Dongpo's Cheat Sheet

今夜之事 出处 现代对应 一句话
全省库容够一半 农科院数据 不缺库 缺的是“库怎么用”
建库有指标,用库没有 吴老师原话 资金机制的产物 空库不是地方病,是拨款方式的标准产出
一条船走全程 旧漕运 照最难那段造船 一年翻掉四成
江船不入汴 刘晏分段转运 多装卸,少翻船 损耗四成降到一成以下
一万多份知院状 每旬每月上报 他不全看 他只看变的那些
三十里一递铺 厚直募善走者 最便宜的基建 两间屋两个人,抵一百分之一座仓
群里有没有种苹果的 刘晏一问 信息不对称 四百八十七人,全是收货的
一户六百,一县三千六百万 马可雪地里算 零资本的量级 跟那笔项目款一个量级
从盐利里挪的 递铺的钱 无法立项之物 它买不到任何看得见的东西
说不清的人 刘晏之死 审计的政治 账上有说不清的地方,定什么罪都定得上
只在盐场收一次 榷盐法 一次课征 四十万缗到六百万缗,而百姓交得更少
它散在路上了 今夜最后一句 那一块九毛三 没有一个人拿走它