一个想法要进期刊，必须先能被审；要能被审，它就得有边界、有方法、有对照、有可核的结论——也就是说，它必须先凝固。这套要求没有恶意，甚至是文明的：没有它，学术就是一堆各说各话，谁嗓门大谁赢。但它有一笔从不入账的代价：凡是还在生成中、还说不清自己是什么的东西，一律过不了这道门——而那恰好是所有新东西出生时的样子。所以审稿人只是执行者，真正的筛子是可审性本身，它在稿件写下第一行之前就已经在工作了。这个频道不反对期刊，它想弄清楚的是这笔代价具体记在哪儿、由谁付、以及在必须交这笔钱的前提下，还有哪些东西能被夹带进去。
To enter a journal, an idea must first be reviewable; to be reviewable it must have boundaries, a method, controls, a checkable conclusion — that is, it must first harden. The demand carries no malice and is arguably civilised: without it, scholarship is a crowd talking past itself and the loudest wins. But it carries a cost that is never entered in the books: anything still in the making, anything that cannot yet say what it is, fails at the door — and that is what everything new looks like when it is born. The reviewer is only the executor. The real sieve is reviewability itself, and it starts working before the first line is written. This channel is not against journals. It wants to know where the cost is booked, who pays it, and what can still be smuggled through by those who must.
四问追的是同一台装置的四个部件。它们没有一个是坏的——每一个单看都合情合理，合起来才是那张网。
Four questions tracking four parts of one apparatus. Not one of them is broken; each is entirely reasonable on its own. Only together do they make the net.
- 摘要的暴力The Violence of the Abstract「用两百字说清你的贡献」这个要求，本身就在挑选贡献的类型：凡是两百字说不清的贡献，等于不存在。这条规则从不拒绝任何人，它只是让某一类工作在填表阶段自动消失——连被拒的记录都不会留下。"State your contribution in 200 words" does not merely describe a contribution; it selects for a type. Any contribution that cannot be stated in 200 words does not exist. The rule rejects no one. It simply makes a class of work vanish at the form-filling stage, leaving not even a record of rejection.
- 三个审稿人互相矛盾，编辑说「综合修改」Three Reviewers Contradict; the Editor Says "Revise Accordingly"这句话在要求什么？它要求的不是更对，是更少可攻击面。你照着改，改出来的东西同时满足了三个互相矛盾的立场——那通常只有一种可能：它不再对任何一个立场构成挑战。修改后的版本更容易通过，也更没有必要存在。What is being asked? Not for the paper to be more right, but for it to present less surface to attack. Comply, and you produce something that satisfies three mutually contradictory positions at once — which usually means one thing: it no longer challenges any of them. The revised version passes more easily, and needs to exist less.
- 影响因子的回写The Impact Factor Writes Back你以为自己在选投哪里，其实是在被选写什么。期刊的位置反过来决定了什么算问题——一个题目「有没有出口」在动笔之前就参与了选题，而这道回写是无声的：没有人下过命令，每个人都只是理性。You believe you are choosing where to submit; you are being chosen for what to write. A journal's standing decides, in reverse, what counts as a problem — whether a topic "has an outlet" enters the choice of topic before the first line is written. The feedback is silent: no one issued an order, everyone was merely being rational.
- 拒绝的社会生产The Social Production of Rejection评审装置的社会产物不是准确，是让拒绝变得无可争辩。这不是阴谋——一个必须每年淘汰九成来稿的系统，最紧要的事从来不是选对，是选得让人没话说。测度就是它为此分泌出来的那层壳。The social product of a review apparatus is not accuracy but rejection made unarguable. This is not a conspiracy: for a system that must turn away nine in ten submissions each year, the pressing task was never to choose rightly but to choose in a way that leaves nothing to say. Measurement is the shell it secretes for that purpose.
第四篇（寄食性裂变）是中华智问三车道碰撞引擎的产出，它与本专栏的律篇押在同一注的两面——那不是安排出来的，是同日产出撞上的。前三篇按 CSSCI 形制写：中英双摘要、中图分类号、与本领域具体论争对话、可证伪推论、参考文献不编造一条。三篇撞的是三块不同的地基，而它们撞的方式是同一个：一件每一步都做对的事，合起来产生了它的反面——而每一步之所以都对，正是因为出事的条件到最近才凑齐。
Written to CSSCI form: bilingual abstracts, CLC number, argument with the field's actual disputes, falsifiable corollaries, not one fabricated citation. Three different foundation stones, struck the same way: a thing correct at every step, producing its own opposite in sum — every step correct precisely because the conditions for failure only just arrived.
寄食性裂变：AI时代教育系统困境的制度病理学诊断Parasitic Fission: An Institutional Pathology of the Education System in the Age of AI
教育系统越成功地维持它对"未来确定感"的垄断，就越加速自己核心使命的瓦解——生存策略本身就是死因。三阶段链：功能寄生（把阶层流动、代际公平、劳动力筛选内化成自己的器官，稀释掉认知使命）、免疫寄生（把能力凭证、在线学位这类异质替代方案"编译"成加强自身合法性的养分）、宿主瓦解与能量裂变。它押在律篇附录第 7 号的反面：它预测确定感需求不可逆地迁移到企业认证与社区学习网络，第 7 号预测那些新装置在人工智能条件下根本建不起来。结算日 2031-12，两边都同意结算办法，而谁也没有为了赢它多写一个字——这是这个引擎最便宜的一次质量检验。本文按典范论文频道自己的三条判准只过一条，因此作为期刊论文而非典范发表，理由与那道坎的位置写在文首。
The more successfully the education system holds its monopoly on certainty about the future, the faster it dissolves its own core mission: the survival strategy is the cause of death. It sits on the opposite side of bet 7 in the law paper's register, settling 2031-12, and neither side wrote a word extra to win it. It clears one of the Paradigm channel's own three criteria, so it is published here rather than there.
读全文 →Read → 教育学 · G423 · 约 1.26 万字Education · ~12.6k可规格化性：能力本位课程的隐蔽筛选器及其自我取消Specifiability: The Hidden Filter of Competency-Based Curriculum
课程目标要写清楚——这句话六十年没被撼动过。艾斯纳、斯滕豪斯、比斯塔都反对过，理由是测量窄化了教育的价值；而这是个价值主张，所以每次都被"那就把重要的也写成目标"吸收掉了。本文换一条路：能写成可测量目标，意味着说得出"什么算做到了"；而这正是把它交给机器所需要的全部条件。评分标准与规格说明是同一份文本的两次使用。于是改革越彻底，课程指定的越是那些最不需要长在人身上的能力。高阶思维不是避难所——它的评分标准写得最细，因而规格最完整。
Objectives should be stated clearly — unshaken for sixty years. Eisner, Stenhouse and Biesta all objected, on the ground that measurement narrows educational value; a value claim, and so absorbed each time by "then write the important things as objectives too". This paper takes another route: to write a measurable objective is to state what counts as attainment — which is all a machine needs. A rubric and a specification are one text used twice.
读全文 →Read → 健康行为学 · R193;B849 · 约 1.19 万字Health Behaviour · ~11.9k过度理由的规模化：即时反馈对行为自持性的结构性侵蚀Overjustification at Scale
数字健康的试验终点设在干预期末；而削弱效应恰恰发生在奖励撤除之后——两个文献群的测量窗口不重叠，所以四十年互相看不见。德西那份 128 项实验的元分析说：有形的、预期的、任务依随的奖励削弱内在动机——这三个条件正是游戏化的定义性特征，一个不差。机制更狠：填补自然后果的延迟是游戏化全部的正当性理由，而填补延迟的方式，是抢在延迟后果到来之前占据归因的位置。正当性理由与失效机制是同一件事。
Digital health trials end when the intervention does; the undermining effect appears only after rewards are withdrawn. The windows do not overlap, so forty years of mutual invisibility. Deci's 128-experiment meta-analysis: tangible, expected, task-contingent rewards undermine intrinsic motivation — the three defining features of gamification, exactly.
读全文 →Read → 企业管理 · F272.3 · 约 1.24 万字Management · ~12.4k负和的强制投入：AI采纳中的租金耗散与问题空间的所有权A Negative-Sum Compulsory Investment
卡尔说 IT 已成中性成本，像电力。本文说不：电力改变生产函数，它改变产出本身——产出直接就是客户看见的交付物，因此它在普及的同时抬高了合格线。采纳者被夹在寡头供应商与完全竞争的下游之间，是链条上唯一既付钱又不能涨价的环节：租金经订阅费流向上游、经降价流向客户，采纳者所得为零。剩下什么？一切可买的输入都不构成优势，唯一不可买的是问题——而问题受阿罗悖论约束，在原理上不可交易。
Carr said IT became a neutral cost, like electricity. Not so: electricity changed the production function; this changes the output itself — it is the deliverable the customer sees, so diffusion raises the passing line. The adopter is the one link that pays and cannot reprice.
读全文 →Read → 与学位论文频道共用的一条纪律A discipline shared with Degree Theses为什么这三篇正文里没有一个外来术语Why Not One Foreign Term Appears
要投出去的东西，一律用目标学科的母语说话：可规格化性接泰勒与马杰，过度理由接莱珀与德西，租金耗散接巴尼与卡尔。一篇论文若必须先教会审稿人一套新话才能被读懂，它就还没写完。三篇也一律自报家门是理论论证：没有原始数据就不假装有，参考文献一条不编，把可证伪推论钉死在文中等着被杀——其中健康篇那条最要命的检验，所需数据完整存在于设备厂商手里，而它长期没被公开做出来。
Anything meant to be submitted speaks its field's own language: specifiability to Tyler and Mager, overjustification to Lepper and Deci, rent dissipation to Barney and Carr. A paper that must first teach its reviewer a private vocabulary is not finished.
对照学位论文频道 →Compare →三篇都不是关于期刊的，都是关于同一台装置的。头两篇是一条律和它的推论：律篇问的是这台装置到底在你身上造了什么，推论篇是它在教育评价上的兑现（律篇发表的同时订正了推论篇三处，包括推论篇最不可替代的那一笔）。第三篇拆它怎么把管不了的东西抵押出去。第四问（拒绝的社会生产）在律篇里被推到了本频道自己身上：版面稀缺，因此同行评审交付的那个"写得更严谨的我"由今年这批来稿的分布定义，因此他不可结清。
None is about journals; all three are about the same apparatus. The first two are a law and its corollary; the third takes apart how the apparatus mortgages what it cannot govern. The fourth question is turned on this channel: page space is scarce, so the "more rigorous me" that peer review hands you is defined by this year's submission pool — and therefore cannot be settled.
反事实自我的社会制造：一步守恒、不可结清性与稀缺分配装置的产品The Social Manufacture of the Counterfactual Self: One-Step Conservation and Non-Settleability
备考投入越过了"再加码自己成绩就会下降"这条线仍在加码，而所有竞争理论都预测它会停。因为守恒的是距离，不是水平：装置交给你的不是一个位置，是一个必须持有、且必须永远差你一步的自己——距离为零它不存在（你做尽了还是没过，那这事就跟你无关了），距离为无穷它拿不住（"我天生不行"，同样跟你无关）。由此：装置只能选没有顶的属性（刷题没有顶，掌握求根公式有顶）；外壳规格是三条不是两条，第三条是不可结清；而分界不在名额、不在语法、也不在稀缺——在你手上那个人能不能被兑现。飞行执照的那个人可以（学会就是学会），排名的那个人不行（他站在别人的位置上，位置在动）。全文第二节零引用，文献到第七节才出场，且不是为了分岔，是被指派工位：默顿的失范五型是这条生产线的一个成功态加四个失败态；法兰克福吵了半个世纪的"归责需不需要他择可能"，旁边那间屋子没人进过——承受归责需不需要它。
Effort escalates past the point where more of it lowers your own score, and every theory of competition predicts it stops. It does not, because what is conserved is the distance, not the level: the apparatus hands you a self you must hold and must never reach. Hence no-ceiling attributes, a three-constraint spec whose third term is non-settleability, and a boundary that lies neither in quotas nor in grammar nor in scarcity. Section 2 cites no one; the literature arrives in Section 7, not to be argued with but to be assigned stations.
读全文 →Read → 王德生 著 · 三车道之一 · 形态（S） · 约 2.2 万字by Wang Desheng · Lane I of III · Form · ~22k退出成本陷阱与齿轮组策略：AI 时代教育系统变革的困境与出路The Exit-Cost Trap and the Gear-Train Strategy
教育系统不是不知道要变，是被锁在原地：教师的专用性人力资本贬值、家长对确定性的风险规避、评价产业链的寄生性依赖，三重力量构成一个第一个退出者被惩罚的均衡——观望成了所有人的占优策略。本文给出四齿轮的退出成本对冲机制，并提出把分析单位从"学校教育系统"撤换为"分布式认知发生系统"。这条车道的机制后来成了典范论文《寄食性裂变》的微观基础。Waiting is everyone's dominant strategy when the first to exit is punished.
读全文 →Read → 王德生 著 · 三车道之二 · 条件（E） · 约 2.1 万字by Wang Desheng · Lane II of III · Condition · ~21k教育是如何被"过度赋义"的：AI 时代教育困境的制度根源与范式转向How Education Was Over-Signified
困境的根源不在技术冲击，也不在教育系统内部失灵，而在一种百年演化的制度错位：社会治理体系持续把阶层流动、劳动力筛选、代际公平三大压力转嫁给教育，使其被过度赋义。AI 撕开了标准化评价等价物的最后合法性，代偿空间归零。核心主张："教育该如何卸下过度赋义"应替代"教育该如何适应技术变革"成为核心问题。这条车道后来在典范论文中被重命名为"功能寄生"。From reforming education to unburdening it.
读全文 →Read → 王德生 著 · 三车道之三 · 演化（D） · 约 2.4 万字by Wang Desheng · Lane III of III · Evolution · ~24k制度免疫：AI 时代教育系统自我维持的机制与演化路径Institutional Immunity
为什么一个功能被掏空的制度，能在全社会普遍怀疑之下继续运转？本文识别出五套相互配合的免疫技术——功能仿制、标准骑劫、符号挪用、成本转嫁、叙事重置——并论证演化拐点的触发条件不是替代方案的功能完备度，而是替代方案信用积累速度与旧系统免疫消化速度之间的竞赛。把经典问题从"为何改变如此困难"重新框定为"维持机制是如何被工程化的"。Not why change is hard, but how the maintenance was engineered.
读全文 →Read → 王德生 著 · 推论篇 · 教育专栏 × 本频道 · 已被律篇订正三处 · 网页 · PDFby Wang Desheng · Corollary · Education × this channel · Amended in three places · Web · PDF拒绝的社会生产：教育评价的测度外壳与外壳再分泌律The Manufacture of Rejection: The Measurement Shell and the Law of Shell Re-secretion
评价装置生产的不是关于学习的信息，是可接受的拒绝；测度是它分泌出来的壳。拆掉一层壳，装置立刻分泌下一层——USMLE Step 1 改为通过/不通过之后，压力原样搬到了 Step 2。修订稿补上了脱钩的时间机制：壳与被测对象脱钩得多快，由优化产业成熟得多快决定；人工智能不是揭穿了壳，是把再分泌的周期从两代人压到一个学年——于是壳换代而不消亡，从结果的不可争辩换到过程的不可争辩。期刊评审不是同一台装置的另一个物种，是同一台装置：双盲评审与科举糊名在结构上是同一样东西，理由也完全一致——它不提高判断的准确性，它消除可指认的施害者。预注册、数据强制公开、写作留痕，正在被读为科研诚信；本文读为换代。文末解剖了一片标本：本文自己为了在这台装置里活下来而分泌的那个式子。
An assessment apparatus does not produce information about learning; it produces acceptable rejection, and measurement is the shell it secretes. Strip one shell and the next is secreted at once — when USMLE Step 1 went pass/fail, the pressure moved intact to Step 2. The revision supplies the timing mechanism: how fast a shell decouples is set by how fast its optimisation industry matures, and AI has compressed that cycle from two generations to a single academic year — so the shell changes generation rather than dying, from unarguable results to unarguable process. Journal review is not another species of the same apparatus; it is the apparatus: double-blind review and the sealed names of the imperial examinations are structurally one thing, for identical reasons — neither improves the accuracy of judgement; both remove anyone who can be named as the culprit. Preregistration, mandatory data release and writing trails are read as research integrity; this paper reads them as the change of generation. It closes by dissecting one specimen: the formula this paper itself secreted in order to survive that apparatus.
读全文 →Read → 王德生 著 · 论文抄袭专栏by Wang Desheng · Plagiarism column原创抵押权：制度化知识生产中原创定义的自我取消The Originality Mortgage: How Institutional Knowledge Production Cancels Its Own Definition of Originality
制度把它管不了的认知发生，抵押给它管得了的产权标记——查重率、创新点声明、期刊级别。抄袭不是道德溃疡，是这笔抵押契约在最薄弱节点上的极端兑现。
The institution mortgages the cognitive event it cannot govern against the property markers it can: similarity scores, novelty statements, journal tiers. Plagiarism is not a moral ulcer but the extreme redemption of that mortgage at its weakest node.
读全文 →Read →