第十五篇：隋——唐的秦
Essay 15: Sui — Tang's Qin
公元581年，杨坚接受北周静帝禅让，建立隋朝。589年灭陈，统一天下。618年隋炀帝在江都被部下杀害，隋亡。
从统一到灭亡，二十九年。从建国到灭亡，三十七年。
这个寿命和秦几乎一样——秦从统一（前221年）到灭亡（前207年）是十五年，从建国（如果从秦王政即位算）到灭亡大约三十七年。巧合到令人不安。
更令人不安的是结构性的相似：两个朝代都结束了长期分裂、实现了大一统，都在统一后推行了极其精密的制度改革，都在第二代皇帝手里因为过度扩张而崩溃，都被紧随其后的长寿王朝（汉、唐）几乎原封不动地继承了全套制度。
秦是汉的制度供应商。隋是唐的制度供应商。供应商短命，使用者长寿。
这个模式值得深挖。
一、杨坚——精密的构型工程师
杨坚是关陇集团的核心成员。他的父亲杨忠是西魏/北周的十二大将军之一，封随国公。杨坚的妻子独孤伽罗是独孤信的女儿——独孤信是关陇集团最重要的创始成员之一，他的三个女儿分别嫁给了北周明帝（长女）、杨坚（七女）和李渊之父李昞（四女）。
一个人的岳父的三个女儿分别是两个朝代皇帝的母亲和一个朝代皇帝的妻子。关陇集团的联姻网络之密，由此可见一斑。
杨坚篡北周的过程和司马篡魏在形式上类似——外戚摄政，然后禅让。但杨坚建国之后做的事情和司马家完全不同。司马家拥抱豪族，消灭制衡。杨坚的方向相反：他在建立一个高度集权的、制度化的、不依赖于任何特定社会集团的治理体系。
杨坚的制度设计覆盖了政治、行政、法律、经济、军事的全部层面。这套设计的精密程度在中国政治史上只有商鞅变法可以比肩。
三省六部制。
中央行政机构分为三省——中书省（起草政令）、门下省（审核政令）、尚书省（执行政令）。尚书省下设六部——吏、户、礼、兵、刑、工，分管人事、财政、礼仪、军事、司法、工程。
这个设计的精巧之处在于权力的分割。在3.0版原版中（汉朝），丞相一人独揽行政大权，权力太集中，容易形成权臣（霍光、王莽）。三省制把丞相的权力拆成了三份：决策权（中书）、审核权（门下）、执行权（尚书）。三省互相牵制，没有任何一个机构可以独揽大权。
门下省的审核权尤其值得注意。中书省起草的政令如果门下省认为不妥，可以封驳——退回去重写。这是一种制度化的纠错机制。它不依赖于某个谏臣的个人勇气（你进谏可能被杀），而是内嵌在行政流程中的（门下省的职责就是审核，你不审核才是失职）。
从凿构的角度看，三省六部制是对3.0版单点结构的第一次系统性修补。3.0版的致命弱点是整个系统依赖于皇帝一个人的判断力——东汉的外戚宦官交替就是这个弱点的产物。三省制在皇帝和政策执行之间插入了一个多节点的决策链条。皇帝仍然是最终决策者，但他的决策需要经过一个制度化的加工流程，而不是从他的脑袋里直接蹦到执行层面。
科举制的确立。
杨坚废除了九品中正制，开始用考试来选拔官员。杨广进一步发展，设置了进士科——以考试成绩为标准选拔人才，不看门第、不看推荐、只看答卷。
这是对门阀构的终极否定。九品中正制的本质是让门阀来评价门阀——中正官本身就是大族出身，他们评价的标准自然偏向门第。科举制的本质是让考试来评价所有人——不管你姓什么、你爹是谁，你在考场上写出来的东西决定你的前途。
科举制对中国政治史的影响怎么强调都不为过。它一举解决了（或者说大幅缓解了）3.0版以来的几个核心余项：
门阀垄断——被打破了。任何有才学的人都有机会进入官僚系统，不再需要大族的推荐。寒门出贵子从此成为可能。
人才选拔的标准化——实现了。不再依赖某个人的主观判断（察举制的问题），而是依赖一个相对客观的、可重复的评价程序。
社会流动——被激活了。底层的聪明人有了上升通道，这个通道的存在本身就大幅降低了底层的反抗动力。你为什么要造反？你可以考科举。考上了你就是人上人。这条路虽然窄，但它存在就够了——它给了每一个人一个"不造反也可以改变命运"的希望。
科举制是一种极其高明的余项管理工具。它不压制余项（秦的方式），不放任余项（黄老的方式），不用意识形态重新编码余项（独尊儒术的方式），而是为余项提供一条不威胁构的释放通道。你对现状不满？不用造反，去考试。你有才能没有出路？不用投靠门阀，去考试。考试吸收了大量本来可能变成反抗能量的社会张力。
《开皇律》。
杨坚主持编纂了新的法典《开皇律》，大幅简化了北朝以来繁复的法律条文，减轻了刑罚力度。只保留五种刑罚（笞、杖、徒、流、死），废除了许多酷刑。
从构的角度看，《开皇律》的核心设计思路是：法律要简明到每一个人都能理解，刑罚要适度到不会制造过多的怨恨。这是对秦律的反面——秦律的特征是条文繁密、刑罚严酷。简明和适度不是因为杨坚心善，而是因为他（或者他的法律顾问）理解了一个道理：法律的目的是维持秩序，不是制造恐惧。恐惧可以在短期内压制反抗，但恐惧积累到一定程度会变成愤怒。简明和适度的法律把恐惧控制在一个不会翻转为愤怒的水平上。
均田制和租庸调制的完善。
隋继承了北魏的均田制，并进一步完善了配套的赋税制度——租庸调制。租是谷物税，庸是劳役（可以用布帛代替），调是布帛税。标准明确、计算简单、负担适中。
这套经济制度的优点是透明——每个农民知道自己该交多少，不会被层层加码。透明降低了腐败空间，也降低了民怨——你交的税虽然不少，但至少你知道为什么交这么多，不会觉得自己被黑了。
把杨坚的全部制度设计放在一起看，一个极其精密的3.5版操作系统浮现出来：
政治层面：三省六部（分权制衡的行政体系）。 人事层面：科举制（开放的人才选拔通道）。 法律层面：开皇律（简明适度的法律体系）。 经济层面：均田制+租庸调（透明的土地和赋税制度）。 军事层面：府兵制（兵农合一的动员体系）。
每一个层面都比3.0版有显著的改进。三省六部比丞相制更分权。科举比察举更开放。开皇律比秦律更人道。均田制比自由土地市场更能抑制兼并。府兵制比征兵制更高效。
杨坚用自己的一生完成了这个升级。他在位二十四年（581—604年），勤政节俭，亲自处理大量政务，国库充盈到了"天下储积得供五六十年"的程度。
然后他的儿子来了。
二、杨广——构的推进速度超过了社会的吸收速度
隋炀帝杨广是中国历史上被妖魔化最严重的帝王之一。传统叙事把他描绘为纯粹的暴君——好大喜功、穷奢极欲、荒淫无道。这个画像的制造者是唐朝——推翻他的人。考虑到末代昏君叙事的模板效应（第二篇讨论过），这个画像的可信度需要打很大折扣。
如果剥开道德叙事看结构，杨广的问题不是无能或荒淫——他实际上是一个有雄才大略的政治家——他的问题是秦始皇式的：构的推进速度超过了社会吸收余项的速度。
大运河。
杨广下令开凿大运河，连通南北水系。这是一项功在千秋的基础设施工程——大运河此后一千多年持续发挥作用，是中国南北经济整合的大动脉。唐朝能够以关中为基地统治全国，很大程度上依赖大运河把南方的粮食运到北方。
但修建大运河的人力成本是惊人的。征调民夫数以百万计。工期被极度压缩——杨广要在几年之内完成一项本来应该用几十年来慢慢做的工程。压缩工期意味着加大劳动强度，加大劳动强度意味着大量民夫的死亡和家庭的破碎。
三征高句丽。
612、613、614年，杨广三次大规模远征高句丽（今朝鲜半岛北部和辽东）。第一次出动了一百一十三万大军（号称，实际可能少一些但数量级应该没问题），后勤运输的民夫更是倍于军队。三次远征全部失败。第一次在辽东被击退，渡过鸭绿江的前军在平壤城下遭到毁灭性打击。后两次因为国内叛乱不得不中途撤军。
三征高句丽的军事损失和后勤消耗是灾难性的。百万级别的军队和民夫被投入一个遥远的、补给困难的、地形对攻方极不利的战场。即使打赢了，获得的也只是一块偏远的、经济价值有限的土地。投入和回报完全不成比例。
巡游江南。
杨广多次乘龙舟巡游江南。龙舟队伍绵延数百里，沿途征用民力物力，消耗巨大。传统叙事把这些巡游描绘为单纯的享乐行为。实际上巡游有政治功能——宣示对南方的统治、考察地方情况、安抚南方豪族。但不管动机如何，巡游的成本确实由沿途百姓承担。
把大运河、三征高句丽和巡游江南放在一起，一个熟悉的模式出现了：和秦始皇一模一样的模式。
大运河=修长城+修驰道（基础设施的超大规模投入）。 三征高句丽=北击匈奴+南征百越（对外的军事扩张）。 巡游江南=始皇帝巡游天下（权力的物质化展示）。
连失败的方式都相似。秦始皇的大工程和对外战争掏空了社会的承受力，陈胜吴广揭竿而起。杨广的大工程和对外战争同样掏空了社会的承受力，杨玄感叛乱（613年）是第一个信号，此后全国各地叛乱蜂起。
但杨广和秦始皇之间有一个关键的区别，这个区别决定了隋和秦的不同历史遗产。
秦始皇的制度设计和他的过度扩张是同一个行为的两面——书同文车同轨度同制和修长城征百越是同时推进的，社会分不清哪些是好的制度建设哪些是坏的暴力征用。所以秦亡之后，社会对秦的制度有强烈的排斥情绪（虽然实际上大部分制度被汉继承了，但这个过程花了七十年的缓冲期）。
杨广的过度扩张和杨坚的制度设计是可以分开的。杨坚用二十四年建立的制度体系（三省六部、科举、开皇律、均田制）在杨广的暴政中没有被摧毁——被摧毁的是社会的承受力，不是制度的框架。制度框架完好无损地保留着。
这意味着隋亡之后，继承者不需要像汉初那样花七十年来慢慢摸索——他们可以直接继承隋的全套制度，只需要调整推进的速度（从杨广的疯狂加速调回正常速度）。
唐朝就是这样做的。
三、隋的灭亡——速度杀人
杨广的每一个项目单独来看都有道理。大运河是必要的基础设施。征高句丽有地缘政治的考量（高句丽是北方边境的潜在威胁）。巡游有政治功能。制度改革（科举的发展、法律的完善）是正确的方向。
但所有这些项目同时推进，就超出了社会的承受能力。
这是一个速度问题，不是方向问题。方向是对的，速度太快了。
社会吸收变革的能力是有限的。每一项变革都会产生余项——受损者的不满、新旧制度的磨合摩擦、资源重新分配导致的冲突。如果变革的速度慢到社会可以在消化一项的余项之后再接受下一项，系统可以保持稳定。如果变革的速度快到多项的余项同时涌入，系统就会过载。
杨坚理解这一点。他在位二十四年，制度改革的推进是渐进的——先建三省六部，稳定之后再推科举，科举稳定之后再完善法律。每一步都留了消化时间。
杨广不理解这一点。或者他理解但不在乎。他要在自己的有生之年完成所有的大事业。他的时间观是个人的（我活着的时候要看到成果），不是制度的（制度可以等，成果可以留给下一代）。
这种"急于在自己的任期内完成一切"的心态，和秦始皇如出一辙。始皇帝也是要在自己活着的时候看到一切完成——修长城、修驰道、修阿房宫、征百越、书同文、车同轨，全部同时推进。两个人面对的诱惑是一样的：我有这个能力，我有这个资源（杨坚给他攒的家底），我为什么不做？
答案是：你有能力做，但社会没有能力同时承受。你的能力和社会的承受力之间有一个差距。这个差距就是余项积累的空间。你做得越多越快，余项积累得越多越快。积累到临界点，系统崩溃。
隋炀帝统治的十四年（604—618年），社会从"天下储积得供五六十年"的富庶状态急剧滑入全面崩溃。这个速度之快，和秦从统一到灭亡的十五年几乎完全对等。
不是因为杨广比杨坚差太多——杨广在很多方面甚至比杨坚更有想象力和魄力。是因为构的推进速度和社会的吸收速度之间的关系，不取决于推进者的能力，取决于社会的弹性。社会的弹性是一个客观常数，不会因为推进者更有才华就增大。
你可以是一个天才。但你推一辆车的时候，车轴的承受力不会因为你是天才就变大。你使的力气超过了车轴的极限，车轴断了。不是车轴的错，是你不了解车轴。
四、隋唐之间——为什么唐能直接继承隋
隋亡唐兴之间的过渡期极短。从618年李渊在太原起兵到624年唐基本统一全国，只有六年。相比之下，秦亡到汉统一用了七年，但汉初的制度摸索又花了七十年（黄老无为+文景之治）。唐不需要这七十年，因为隋的制度框架完好无损。
李渊即位后做了什么？几乎照搬了隋的全部制度。三省六部——沿用。科举制——沿用并发展。均田制+租庸调——沿用。府兵制——沿用。《开皇律》的精神和框架——沿用，稍作修订后变成了《武德律》，后来又变成更完善的《贞观律》和《永徽律》。
唐的制度创新当然有（后面会谈），但它的基础是隋。没有隋的制度供应，唐不可能在这么短的时间内建立一个高效运转的帝国。
这个"供应商短命、使用者长寿"的模式为什么反复出现？
因为制度创新和制度运营是两种不同的能力，需要两种不同的政治气质。
制度创新需要强势、激进、不怕得罪人。你要打破旧利益格局，建立新规则，把反对者压下去。这种气质在创新阶段是必需的，但在创新完成之后的运营阶段是致命的——因为运营需要的是妥协、耐心、倾听、缓和矛盾。
秦始皇和杨广都是极致的创新者——他们有能力强行推动系统性的制度变革。但他们不知道什么时候该从创新模式切换到运营模式。他们一直踩油门，不知道有刹车这个东西。
刘邦和李渊/李世民则是优秀的运营者——他们不发明新制度（或者只做微调），他们运营前人发明的制度。运营需要的不是想象力和魄力，而是耐心和分寸感。知道什么时候加压什么时候减压，知道什么时候推进什么时候等待，知道社会的承受力在哪里然后不越过那条线。
一个朝代需要两个阶段：创新阶段和运营阶段。最好的情况是同一个朝代在内部完成这个切换（杨坚创新→某个后继者平稳运营）。但如果创新者在切换之前就把系统搞崩了（杨广），那就只好换一个朝代来接管运营。
秦创新，汉运营。隋创新，唐运营。供应商付出了全部代价，使用者收获了全部红利。
五、杨坚vs杨广——同一个系统的两种用法
杨坚和杨广父子之间的反差，是构的代际衰减的又一个案例，但有一个特殊之处。
通常的代际衰减是能力衰减——第一代是天才，第二代是庸才。太康不如禹，二世不如始皇，这都是能力的下降。
杨广不是庸才。他的个人能力可能不低于杨坚——他有战略视野（大运河的构想是跨世纪的），有军事能力（灭陈之战他是实际的前线总指挥，二十岁），有文学才华（他的诗写得不差），有政治手腕（他能在杨坚面前长期伪装自己，夺嫡成功）。
杨广的衰减不在能力，在判断力——具体地说，在对"什么时候该停"的判断力上。
杨坚知道什么时候该停。他攒了一辈子的家底不是因为他没有花钱的欲望，而是因为他判断"现在不是花钱的时候"。他在位二十四年，有足够的能力和资源去搞大工程、打大仗，但他几乎没有做。他在等。等社会完全适应新的制度框架，等国力积累到足以承受大规模投入。
杨广不知道什么时候该停。或者更准确地说，他不愿意等。他有杨坚攒的家底，他有自己的雄心壮志，他看不到"等待"的价值。等什么？等老？等死？我现在就有能力做，我为什么要等？
这不是能力的衰减，是耐心的衰减。耐心是一种不太被重视的政治品质，但它可能是所有政治品质中最重要的一种。因为政治的本质不是做正确的事，而是在正确的时间做正确的事。正确的事在错误的时间做就变成了错误的事。大运河在一百年的时间里慢慢修是伟业，在五年里赶着修就是暴政。区别只在于时间。
代际衰减在这里有了一个新的维度：不只是"第二代更差"，也可能是"第二代一样好但更急"。急的原因是代际衰减的另一面——第二代没有经历过创业的艰辛（杨坚经历了北周末年的政治险恶和创业初期的物资匮乏），所以他不知道现在的家底是怎么攒出来的，不知道攒的过程有多难、多慢。他只看到了结果（满满的国库），看不到过程（二十四年的克制）。
看不到过程的人，不会珍惜结果。不珍惜结果的人，会把二十四年攒的家底在十四年里花光。
六、隋的遗产——3.5版的正式发布
隋朝虽然只存在了三十七年，但它的制度遗产是中国政治史上最丰厚的之一。
如果把3.0版（汉武帝/董仲舒建立的儒法合流体系）比作操作系统的初始版本，那么隋就是3.5版的开发者。3.5版在3.0版的基础上做了以下升级：
行政体系从丞相独揽升级为三省分权。这减少了权臣篡权的风险，增加了决策的制度化程度。
人才选拔从察举/九品中正升级为科举。这打破了门阀垄断，建立了相对公平的社会流动通道。
法律从繁密严酷升级为简明适度。这降低了社会的怨恨积累速度。
土地制度从完全自由市场升级为均田制（有限的国家调控）。这延缓了土地兼并的速度。
军事制度从纯征兵升级为府兵（兵农合一）。这降低了维持常备军的财政负担。
这些升级不是隋一家的功劳——均田制来自北魏，府兵制来自西魏/北周，科举的雏形来自南朝——但隋把它们整合成了一个完整的、互相配套的系统。整合本身就是巨大的贡献。零件是别人造的，但把零件组装成一台能运转的机器是隋的工作。
唐几乎原封不动地接收了这台机器。然后在这台机器上运行了将近三百年。
这是隋对凿构周期律的最大贡献：它证明了制度的寿命可以远超制度创建者的政权寿命。制度不是属于某个朝代的，制度属于文明。朝代可以灭亡，制度活下来。杨坚和杨广都死了，三省六部活了一千多年。
从余项守恒的角度看，隋的故事有一条最深的教训：余项的积累速度不取决于你做的事情是不是对的，而取决于你做的速度是不是社会能消化的。
正确的事情如果推得太快，产生的余项和错误的事情一样多。大运河是正确的，三征高句丽的战略考量是有道理的，科举制是先进的，均田制是合理的——但所有这些正确的事情同时以最快速度推进，产生的余项足以摧毁一个帝国。
正确不是免死金牌。正确只意味着方向没问题。方向没问题但速度太快，你一样会翻车。社会的承受力不关心你的方向是不是对的。它只关心你施加的压力有没有超过它的极限。超过了就断裂。不管你多正确。
下一篇：唐前期——太宗到玄宗。贞观是构的黄金态，3.5版的巅峰运行。但府兵制到募兵制的转换埋下了安史之乱的种子。武则天不是破坏构的人，她是用构的逻辑攻击构的性别假设，暴露了构的一个隐含前提——3.5版默认操作者是男性。这个假设从来没有被检验过，直到她检验了它。
In 581 CE, Yang Jian accepted abdication from Northern Zhou's last emperor and established the Sui dynasty. In 589, Sui destroyed Chen and unified the realm. In 618, Emperor Yang of Sui was killed by his own subordinates in Jiangdu. Sui ended.
From unification to collapse: twenty-nine years. From founding to collapse: thirty-seven years.
This lifespan was almost identical to Qin's — Qin lasted fifteen years from unification (221 BCE) to collapse (207 BCE), and about thirty-seven years from establishment to end. The coincidence is unsettling.
More unsettling is the structural similarity: both dynasties ended long divisions and achieved grand unification; both implemented extremely precise institutional reforms after unification; both collapsed in the second generation through overextension; and both were followed by long-lived successor dynasties (Han, Tang) that inherited their complete institutional systems almost intact.
Qin was Han's institutional supplier. Sui was Tang's institutional supplier. The supplier lived briefly; the user lived long.
I. Yang Jian: A Precise Construct Engineer
Yang Jian was a core member of the Guanlong Bloc. His father Yang Zhong was one of Western Wei/Northern Zhou's twelve great generals. His wife, Dugu Qieluo, was the daughter of Dugu Xin — one of the Guanlong Bloc's most important founding members, whose three daughters married Emperor Ming of Northern Zhou (eldest), Yang Jian (seventh), and Li Bing, father of Tang founder Li Yuan (fourth).
Yang Jian's institutional design covered politics, administration, law, economics, and military affairs at every level. Its precision in Chinese political history was comparable only to Shang Yang's reforms.
The Three Departments and Six Ministries (三省六部制). The central administration was divided into three departments: the Secretariat (drafting edicts), the Chancellery (reviewing edicts), and the Department of State Affairs (executing edicts). Under the Department of State Affairs were six ministries covering personnel, finance, ritual, military, justice, and public works.
The design's elegance lay in power division. In version 3.0's original form (Han dynasty), a single chancellor wielded enormous administrative power, easily producing overmighty ministers (Huo Guang, Wang Mang). The three-department system split the chancellor's power into three: decision-making authority (Secretariat), review authority (Chancellery), execution authority (Department of State Affairs). The three departments checked each other; no single institution could monopolize power.
The Chancellery's review authority deserves particular attention. If the Secretariat drafted an edict the Chancellery considered inappropriate, it could refuse and return it for revision. This was an institutionalized error-correction mechanism — not depending on any individual official's personal courage (you might be executed for remonstrating), but embedded in the administrative process itself (the Chancellery's duty was to review; failing to do so was dereliction). From the Chisel-Construct perspective, this was the first systematic patching of version 3.0's fatal single-point structure.
Establishment of the civil examination system (科举制). Yang Jian abolished the Nine-Rank System and began selecting officials through examinations. Yang Guang further developed this, establishing the jinshi examination — selecting talent purely by examination performance, regardless of family background or letters of recommendation.
This was the ultimate negation of the gentry construct. The Nine-Rank System's substance was gentry evaluating gentry; the examination system's substance was examinations evaluating everyone. It simultaneously resolved several of version 3.0's core remainders:
Gentry monopoly was broken — anyone with talent had a path into the bureaucratic system without needing great family recommendations.
Talent selection was standardized — no longer dependent on someone's subjective judgment but on a relatively objective, repeatable evaluation process.
Social mobility was activated — bright people from humble origins now had a path upward. The path was narrow, but its existence dramatically reduced the incentive for revolt. Why rebel when you can take the examinations? Pass them and you become one of the elevated. The examination system was an extremely sophisticated remainder-management tool — providing a release channel for social tension that did not threaten the construct.
The Kaihuang Code (开皇律). Yang Jian oversaw compilation of a new legal code, dramatically simplifying the verbose laws of the Northern Dynasties and reducing punishments. Only five categories of punishment (flogging, beating, imprisonment, exile, death), many cruel punishments abolished.
The core design philosophy: law should be simple enough for everyone to understand, punishment moderate enough not to generate excessive resentment. This was the opposite of Qin law — which was characterized by dense provisions and severe punishment. Simple and moderate law kept fear at a level that would not flip into rage.
Refinement of the equal-field system and the tax-corvée-cloth system (租庸调制). Sui inherited Northern Wei's equal-field system and further developed its complementary tax system. Grain tax (租), corvée labor (庸, which could be substituted with cloth), and cloth tax (调). Standards clear, calculation simple, burden moderate.
This economic system's advantage was transparency — every farmer knew what they owed and would not be arbitrarily overcharged. Transparency reduced both corruption and popular grievance.
Taken together, Yang Jian's institutional design was a precisely engineered version 3.5 operating system:
Political: Three Departments and Six Ministries (power-dividing, mutually checking administrative structure). Personnel: Civil examination system (open talent selection channel). Legal: Kaihuang Code (simple and moderate legal framework). Economic: Equal-field system plus tax-corvée-cloth (transparent land and tax system). Military: Garrison militia system (combined farming-soldiering mobilization structure).
Each level was a significant improvement over version 3.0. Yang Jian spent his life completing this upgrade. In his twenty-four years on the throne, he was diligent and frugal, managing vast amounts of state business personally. The treasury filled to a level where "the empire's stored goods could supply fifty to sixty years of expenses." Then his son arrived.
II. Yang Guang: Speed Exceeding Society's Absorption Capacity
Emperor Yang of Sui (Yang Guang) is one of the most demonized emperors in Chinese history. Traditional narrative depicts him as a pure tyrant — vainglorious, extravagant, dissolute. This portrait was manufactured by Tang — the people who overthrew him. Given the template effect of the "last wicked ruler" narrative (discussed in essay 2), this portrait requires large discounts.
Strip away the moral narrative and look at structure: Yang Guang's problem was not incompetence or depravity — he was actually a statesman of grand ambition — his problem was isomorphic with Qin Shihuang's: the speed of construct implementation exceeded society's capacity to absorb the remainder.
The Grand Canal. Yang Guang ordered the excavation of the Grand Canal, connecting south and north water systems. This was an infrastructure project with merit for a thousand years — the Grand Canal continued to function for over a millennium afterward, the main artery of China's north-south economic integration. Tang's ability to rule the realm from Guanzhong depended heavily on the Grand Canal transporting southern grain northward.
But the human cost of building it was staggering. Millions of workers conscripted. The timeline was compressed to the extreme — Yang Guang wanted to complete in a few years a project that should have been done gradually over decades. Compressed timelines meant intensified labor; intensified labor meant mass worker deaths and broken families.
Three campaigns against Goguryeo. In 612, 613, and 614, Yang Guang launched three massive campaigns against Goguryeo (northern Korean peninsula and Liaodong). The first mobilized one million one hundred thirty thousand troops (official figure; reality may have been somewhat less but the order of magnitude is credible), with supply transport workers outnumbering the soldiers. All three campaigns failed. The third was aborted mid-campaign due to domestic rebellions.
Touring Jiangnan. Multiple elaborate tours by dragon-boat flotillas stretching hundreds of li, conscripting civilian labor and resources en route.
The pattern that emerges is familiar: identical to Qin Shihuang.
Grand Canal = building the Great Wall plus the Express Roads (massive infrastructure investment). Three Goguryeo campaigns = campaigns against Xiongnu plus the southern Yue people (external military expansion). Tours = First Emperor's tours of the realm (physical expressions of power).
The failures were structurally similar too. Qin's mega-projects and external wars exhausted society's capacity; Chen Sheng and Wu Guang rose. Sui's mega-projects and external wars similarly exhausted society's capacity; Yang Xuangang's rebellion (613) was the first signal, followed by uprisings everywhere.
But Yang Guang and Qin Shihuang had one critical difference, which determined Sui and Qin's different historical legacies.
The First Emperor's institutional design and his overextension were two faces of the same action — standardizing script, measurement, and axle widths while building the Great Wall and campaigning in the south were all pushed simultaneously; society couldn't distinguish good institutional construction from violent conscription. So after Qin fell, society had strong rejection of Qin's institutions (though most were inherited by Han — but this process required seventy years of cushioning).
Yang Guang's overextension and Yang Jian's institutional design were separable. The institutional framework Yang Jian built over twenty-four years (Three Departments, civil examinations, Kaihuang Code, equal-field system) was not destroyed during Yang Guang's tyranny — what was destroyed was society's tolerance, not the institutional framework. The institutional framework remained intact.
This meant that after Sui fell, the inheritors did not need, as early Han did, seventy years of groping toward solutions — they could directly inherit Sui's complete institutional system, only needing to adjust the speed of implementation (from Yang Guang's manic acceleration back to normal pace).
Tang did exactly this.
III. The Supplier-User Pattern
Why does the "short-lived supplier, long-lived user" pattern recur?
Because institutional innovation and institutional operation are two different capabilities requiring two different political temperaments.
Institutional innovation requires forcefulness, radicalism, and willingness to offend. You must break old interest patterns, establish new rules, suppress opponents. This temperament is essential in the innovation phase, but lethal in the subsequent operation phase — because operation requires compromise, patience, listening, defusing contradictions.
The First Emperor and Yang Guang were extreme innovators — capable of forcefully driving systemic institutional change. But they did not know when to switch from innovation mode to operation mode. They kept accelerating; they did not know the brakes existed.
Liu Bang and Li Yuan/Li Shimin were excellent operators — they did not invent new institutions (or only made fine adjustments); they operated institutions invented by their predecessors. Operation requires not imagination and boldness but patience and a sense of proportion. Knowing when to add pressure and when to reduce it. Knowing when to advance and when to wait. Knowing where society's tolerance limit lies and not crossing it.
A dynasty needs two phases: innovation phase and operation phase. The best case is the same dynasty completing this switch internally (Yang Jian innovates → a capable successor smoothly operates). But if the innovator crashes the system before the switch (Yang Guang), a different dynasty inherits the operation.
Qin innovated; Han operated. Sui innovated; Tang operated. The supplier paid all the costs; the user reaped all the dividends.
IV. Speed as the Variable
Yang Guang's every project, taken alone, had justification. The Grand Canal was necessary infrastructure. Campaigning against Goguryeo had geopolitical reasoning. The tours had political functions. The institutional reforms (civil examination development, legal refinement) were correct in direction.
But all these projects pushed simultaneously exceeded society's tolerance.
This was a speed problem, not a direction problem. The direction was right; the speed was too fast.
Society's capacity to absorb change is limited. Every change produces remainder — the discontent of those harmed, the friction between old and new institutions, the conflicts from resource redistribution. If change proceeds slowly enough for society to digest one change's remainder before the next arrives, the system can remain stable. If change proceeds so fast that multiple changes' remainders flood in simultaneously, the system overloads.
Yang Jian understood this. In his twenty-four years, institutional reforms were incremental — first establish the Three Departments, then after stabilization push the examination system, then after that stabilizes refine the law. Each step left digestion time.
Yang Guang did not understand this — or understood but didn't care. He wanted to accomplish everything great in his own lifetime. His temporal framework was personal (I want to see results while I live), not institutional (institutions can wait; results can be left to the next generation).
Sui's treasury went from "can supply the empire for fifty to sixty years" to complete collapse in fourteen years — almost perfectly mirroring Qin's fifteen years from unification to collapse.
Society's tolerance doesn't care whether what you're doing is correct. It only cares whether the pressure you're applying exceeds its limit. Exceed it and things break — no matter how right you are.
Next: Essay 16 — Tang's early period: Taizong to Xuanzong. The Zhenguan era as version 3.5's peak operation. The shift from garrison militia to mercenary armies silently planting the seed of the An-Shi Rebellion.