>The One-Line Rejection

A good error message names the exact failure and stops. The compilers' sharpest reviews work the same way — a single line of classical Chinese that identifies the fault and moves on.

The Siku Quanshu Zongmu Tiyao is criticism in the strongest sense. Its ~10,000 reviews span careful praise to polite reservation to open contempt. The best are minor literary masterpieces — a complete verdict on a text's worth in fewer words than most modern reviews use for an opening paragraph.

What makes the demolitions entertaining is the contrast between form and content. Classical Chinese is formal and dignified; the compilers wrote under imperial supervision, in a document preserved in seven manuscript copies. Every word was deliberate. And yet, within that formal framework, they found ways to be absolutely devastating. The following are drawn from across the catalog — the Yijing commentaries, the military section, the divination reviews, the philosophical categories — a small anthology of scholarly destruction.

The prize for most exasperated review may belong to the texts attributed to Zixia (子夏), Confucius's disciple, in the Yijing section. The Zixia Yizhuan had been attributed, debunked, re-attributed, and re-forged so many times that the compilers finally threw up their hands: 未有如此書之偽中出偽，層層造作而不已者 — "none has generated forgery upon forgery, layer upon layer of fabrication without end, like this text." The original was lost in the Han; someone forged a replacement, someone else a different one, a third yet another, each claiming to recover the genuine ancient text. By the Qing the layers were so deep that sorting them needed a full paragraph of bibliographic archaeology just to explain what had happened.

The military cunmu (存目) holds some of the catalog's most colorful dismissals. Late Ming texts were a particular target. One proposed mounting wooden figures on horseback with firecrackers in their bellies as a battle tactic. The verdict: 殆於兒戲 — "virtually a children's game." Another proposed making ship decks slippery with egg whites and tung oil so boarders fall; the compilers didn't dignify it with extended analysis. A third proposed astrological timing for campaigns; the compilers observed that actual generals had won on inauspicious days and lost on favorable ones, which rather undermined the premise. The contrast is the judgment: Sun Tzu gets detailed analysis, these fantasists get a sentence. The brevity says the text isn't worth the words required to explain why it's bad.

One of the most effective techniques was the bibliographic silence argument. If a text claims antiquity but appears in no dynastic bibliography for centuries, the formula appears dozens of times, always devastating: 隋唐志皆不著錄 — "not listed in either the Sui or Tang dynastic bibliographies." This is the scholarly equivalent of checking someone's résumé and finding their claimed employer has no record of them. The registries were compiled by officials who systematically recorded every known text; if a supposedly ancient book appears in none for five hundred years, either it was so obscure no government bibliographer in the empire knew it, or it didn't exist yet. The compilers consistently chose the simpler explanation.

The divination section (術數類) is the richest source of one-line demolitions — roughly three of four divination texts failed review, and the compilers developed a specialized vocabulary of contempt. On a physiognomy manual attributed to Gui Guzi (鬼谷子): 其書鄙俗，全無古意 — "the book is vulgar and crude, with nothing of antiquity about it." On a dream guide: 穿鑿附會，殊不足據 — "forced and fabricated, entirely unreliable." On a feng shui text claiming to predict the rise and fall of dynasties from burial topography: 其說荒誕，而行世甚久，術家奉為秘本 — "its claims are absurd, and yet it has circulated widely for ages; practitioners revere it as a secret text." That last is characteristic: the compilers note not just that a text is bad but that its badness has had consequences. A fraud nobody reads is a curiosity; a fraud practitioners treat as authoritative is a problem.

Not every demolition targets forgery; some texts are dismissed for simply being bad. The Yijing cunmu damns through faint description: 大旨不出先儒之說，而文筆蕪雜，無所發明 — "the main ideas do not go beyond what earlier scholars said, the writing is sloppy, with nothing original." Derivative, poorly written, no new insight — any one flaw might be forgivable; all three earn the cunmu. Worse than derivative is incompetent. For authors who clearly didn't understand their material the compilers reserved their sharpest line: 其人蓋未嘗讀書者也 — "this person has evidently never read a book." From scholars who spent their careers reading books, writing a review meant to endure in the imperial library, this is a permanent verdict on a person's competence, delivered in ten characters.

Sometimes the most devastating reviews start with praise. The compilers had a gift for the compliment that turns into a knife: 其文頗有可觀，惜乎立論全非 — "the prose is quite readable. Unfortunately the arguments are entirely wrong." Acknowledge the craft, destroy the content — their way of saying: this person can write, which makes it worse that they have nothing worth saying. The Han Feizi review runs a larger-scale version of the same pattern — brilliant prose in service of a condemned philosophy.

Beyond their entertainment value, the demolitions are quality control. The Chinese textual tradition was enormous — tens of thousands of texts over two millennia, many derivative, many fraudulent, many attributed to sages who had nothing to do with them. The Siku project was the first systematic attempt to sort the genuine from the fake comprehensively across every field. The demolitions are the negative space: they tell you not just what the compilers valued but what they rejected, and why. Two and a half centuries later, a scholar can check the Siku verdict and know immediately whether a text is authentic, sound, and worth reading.

What makes these demolitions literary rather than merely critical is precision. The contempt is always earned, always specific, delivered in the minimum number of characters required. Learned, concise, lethal — the tradition eating its own tail, and magnificent.