Why most writing feedback is useless

Most writing feedback is useless because it describes the reader instead of the text, and you can sort the useful from the useless in about ten seconds. The usual explanation is that people are too polite to be honest, or too cruel because they have nothing to lose. The evidence says something more precise than either.

Avraham Kluger and Angelo DeNisi audited the feedback literature for Psychological Bulletin in 1996. The average note was encouraging, and more than a third of measured feedback effects were negative: the feedback made performance worse. The pattern behind the failures held across studies. The further a note pulled attention toward the self and away from the task, the worse it performed. "You're so talented" and "you have a great voice" are self notes. They move nothing on the page.

John Suler named the machinery in 2004. The first two factors in his online disinhibition effect are dissociative anonymity and invisibility. You don't know me. You can't see me.

The obvious conclusion is that anonymous critics turn cruel. The trial evidence is more interesting. In 2000, Elizabeth Walsh and colleagues randomised 408 manuscripts at the British Journal of Psychiatry between reviewers who signed their reviews and reviewers who did not. Signed reviews came back higher in quality and more courteous. They also took longer to write.

That third finding is the one to take. Earlier BMJ trials found that signing alone made no measurable difference, so a name guarantees nothing. Time is the variable doing the work. Specificity is expensive, and accountability is what makes a critic pay for it. Nobody spends forty minutes locating the paragraph where a chapter loses its grip if being wrong is free.

Savagery and sugar are the same error in opposite directions. Both leave the page out of the frame.

Damon Knight, James Blish, and Judith Merril founded the Milford Science Fiction Writers' Conference in 1956, a peer conference with no teachers and no students. The procedure it produced, still called the Milford system, makes accountability a rule rather than a virtue. Everyone brings a manuscript. Nobody may attend or comment who has not brought one. The group forms a circle, each person critiques in turn, and the author sits in silence until the last reader has finished.

Read that entry condition again. You do not get read unless you read. Reciprocity is the door, and it has been the door since 1956.

The same lineage produced the tool that makes a critique checkable. Knight went on to found the organisation now called SFWA, which hosts the Turkey City Lexicon: sixty-nine named faults across seven sections, compiled by Lewis Shiner, revised by Bruce Sterling, free and uncopyrighted. "Fuzz" is motivation the author was too lazy to supply, and it ships with its own search term, because the tip-off is usually the word somehow.

A named fault is falsifiable. Tell a writer their third chapter has White Room Syndrome, an opening where someone wakes in a featureless space because the setting has not been invented yet, and they can go and look. They can also come back and say the setting is established in paragraph two, and then you answer for it. Vagueness protects the critic. A note with no location cannot be checked, and a critic who cannot be checked has risked nothing.

Everything above depends on a cost falling on the critic. Milford imposed it with geography and a chair. Most online feedback kept the format and dropped the cost, which leaves a stranger with no standing and no consequence, who types that he loved it and maybe tighten the middle.

Ars Calami rebuilds the cost. Critique is the currency, and it attaches to a name that accumulates a record.

• Give critique, earn credit, spend credit to be read. You get read because you read.
• Publishers see your critique history, so your notes become evidence of your judgement rather than a favour done for a stranger.
• Genre and rank matching puts your chapter in front of readers who know its conventions and write at your level.
• Each chapter goes to several readers, so you can separate a pattern from one person's taste. Four readers flagging one paragraph is a finding. One reader disliking it is a mood.
• Long work splits across multiple readers, each given context on chapters they have not read, which is how chapter 24 gets five readers instead of none.