三长

Sixteenth of February, 2039, a Wednesday, seven in the morning, Chengdu, the north of the city. Old Zhong's print shop at the end of an old street: an old courtyard, two storeys, a magnolia at the gate in furred bud like small clenched fists. The afternoon I landed I went first to Well Lane. The same lane, old walls set with old photographs and bricks. Ten years ago there was a door in that wall, and inside it a teahouse and a stout long-bearded man scalding cups. I stood before the wall a long time. A wall is a wall. Two jades warm in my left shirt pocket. Not yet. Upstairs on the long table, three stacks of proofs, each weighted with a brick, and three people: Guli, in from Urumqi with Yasen, who was downstairs learning to trim paper; He Yuan and Eli by train from Aksu, three days and two nights, Eli's canvas bag holding Yiming's notebook from two Julys ago. A month of laying out pages online; last night's final proof till two. "Uncle Ma, don't look yet. Let us tell you how it's arranged." First, the hundred-odd rules, each with its maker's name as before, and under each a new line: where it works, where it doesn't, who said so. East wall first now reads: "Watch which way the wind comes; hang the windward wall first. Abdurahman." And beneath: Turpan yes; Hami, wind from the north, damp for a week, said Nurlan; after the change, Hami yes. Second, the supplements: thirty-one letters folded in under the rules they touch, each with the writer's name: Ziyang sweet potato, Grandpa Chen, written down by Xiao Liu at the copy shop; Hami's calendar, Nurlan; Guangzhou's blindfold tasting, Boss Liang; Wushi's nights on the canal, Kurban. Third, the thinnest stack: errors.

Up the creaking stairs came a thin, slightly stooped man of fifty in a worn blue gown, a scroll under his arm, who looked at the proofs before the people. "Liu Zhiji, styled Ziyuan, of Pengcheng. A compiler of history. I compiled all my life, till I didn't want to: too many supervisors, one saying write it this way, one that, till the writer was just a copyist. So I went home and wrote a book on how history ought to be written. This is a history." It's a book about hanging grapes. "People who've hung grapes for decades, who hung them, how, where they went wrong, who corrected them, set down one by one. If that's not history, what is?" Someone once asked me why writers are many and historians few. I said: a historian needs three strengths, talent, learning, judgement; no one has all three. You three have one each." To Eli: "You have materials: an old man's fifty years, written down cell by cell at his side. That's learning: good land, gold, timber." To He Yuan: "You can arrange: a hundred rules, thirty-one letters, laid out so the eye finds them. That's talent: the craftsman's hand." To Guli: "You can sort: praise from trial, trial from failure; you know what works at home doesn't count. That's judgement. Learning without talent: a fool on a hundred acres of good land. Talent without learning: a master carpenter with no timber. Both without judgement: a house built without knowing which beam is rotten."

He Yuan, red-faced: "I didn't want the third stack. Print the errors and people will say the book's unreliable. I wanted to cut the eleven rules that failed, so that what's left all works. We argued three nights." Eli opened Yiming's notebook to a page in his blue hand: "1991: can't remember. 1994: can't remember. 1997: can't remember." "Seven of fifty-two he couldn't remember. I asked: can we leave them blank? He said: write them down as can't remember. Leave them blank and later people think you missed them. Write can't remember and they know someone asked and got no answer." Liu Zhiji: "Beyond the three strengths, something more important: love of the upright; good and bad alike must be written. Cut the eleven, and every rule left works, and a reader who sees every rule work believes none of them. Keep them, with where and who, and he knows the ones that say they work, do." He Yuan: "Then my three nights were a mistake." "Write that down too." He pencilled it in. At eight the press ran: sixty-seven names from the fifth edition and nineteen new letter-writers; first, Abdurahman's six trembling characters; second, two Uyghur words in one unshaking stroke, and under them, small: Yiming, cut from the last page of his notebook, "printed as it was, not one stroke retouched." My name is not on the cover. Old Zhong's list: Aksu station sixty copies, on the training line, Liu Guifang; Shanghai forty for the tasting room; the Fresno cooperative twenty through Carol; Boss Liang thirty for his warehouses. "Places that use it pay by the copy; people who take it for themselves still pay in what they know. This time the printing isn't out of your pocket." This is my ox. — M.*

Memo to file. Chengdu, 16 February 2039. Good and bad alike.

1. The wall. Well Lane on the day I landed: where the teahouse door was ten years ago, a wall. Two jades in my pocket. Not yet.

2. The structure. Rules with makers' names, plus where each works and where it doesn't, and who said so; supplements from thirty-one letters, each signed; an errors section.

3. Three strengths. Liu Zhiji: talent, learning, judgement; no one has all three. Eli has learning (Yiming's fifty years, cell by cell); He Yuan has talent (he can lay it out); Guli has judgement (she knows home success doesn't count).

4. The missing third. "Learning and talent without judgement: a house built without knowing which beam is rotten."

5. The argument. He Yuan wanted to cut the eleven failed rules so all that remained worked. Three nights.

6. Yiming's blanks. "1991: can't remember." Seven of fifty-two. "Leave them blank and they think you missed them; write can't remember and they know someone asked."

7. The principle. "Love of the upright; good and bad alike must be written." Cut the failures and a reader believes nothing; keep them with where and who, and he believes the rest.

8. He Yuan's line. "Then my three nights were a mistake." "Write that down too." He did, in pencil, on the last page of the errors.

9. The cover. Eighty-six names. Abdurahman's six trembling characters first; Yiming's own signature in Uyghur second, cut from his notebook, not retouched. Not mine.

10. The ox. 1,200 copies, paid by the places that use them: Aksu (training line), Shanghai, Fresno, Guangzhou. Individuals still pay in what they know. The tin box in Aksu now says "small repairs"; the book now has a source.

11. The Wall Street version. An old fund manager whose annual letter had a page titled "Where I was wrong this year": how often, how much, why, what changed. Laughed at for exposing himself; his investors stayed twenty years. "He tells us the misses, so we believe the hits." And Liu Zhiji: "A history of only the right things is a eulogy. Write right and wrong, and who corrected them, and later people dare to use it. What they dare to use is history."

12. My page. Four new errors since the fifth edition: wanting this year's vine to repay last year's cut; "wait for me"; the printing from my pocket; mistaking in-sample success for good rules. And: "This edition was edited by Guli, He Yuan and Eli. I am one of the people who use it." Next: Alim: "10 March, the weighing. Every morning my father has me wheel him to the cupboard to look at the bag, and pats the door with his left hand."

"A historian needs three strengths, talent, learning and judgement; and beyond them, good and bad alike must be written down."
Liu Zhiji, the Tang historian who wrote the first systematic critique of Chinese historiography, argued that historians were rare because few people combine talent, learning and judgement. Learning without talent is good land in a fool's hands; talent without learning is a master builder with no timber. Above all three he placed the duty to record both good and bad.
The same holds for any body of knowledge that others will rely on. A manual that lists only successes reads like advertising, and readers discount it accordingly. A manual that records where each practice failed, who reported it and how it was corrected gives readers a way to calibrate the rest.
Investors have long recognised this in the managers who publish their mistakes alongside their wins. Disclosure of misses is not self-abasement; it is what makes the claimed hits credible. Even honest blanks belong on the record: "cannot remember" tells a future reader that someone asked.
"A history of only the right things is a eulogy."