Telleny

What a long book forgets

A manuscript grows past the part of itself a writer can hold at once. Nearly everything Telleny is follows from what you do about that.

A book outgrows its writer

By the later chapters a draft holds more decisions than any one reading of it can carry.

A short piece can be held whole in the mind. A novel cannot. Somewhere past the first third, a manuscript begins to contain more commitments than a writer can keep in view at once: a promise made in an opening paragraph, a fear a character admitted to nobody, a rule the world obeys, a date that has to stay before another date, a word that was used once and must not be used again.

None of that is a failure of memory. It is a property of length. The book is doing what a long book does, which is to accumulate obligations faster than any single act of attention can audit them, and it will keep doing it whatever tool sits beside it.

What changes between one writer and another is only where the accumulated material is kept: in a notebook, in a spreadsheet, in a series of increasingly specific document names, or in the uneasy sense that something three chapters back has stopped agreeing with the page in front of you.

Six loose gatherings of blank handmade paper laid in a row on dark wood, a single terracotta thread running through all of them and away to a small brass bell.

Memory is not continuity

Telleny does not keep a book's continuity. The difference is the whole design.

The obvious thing to build on top of that problem is a watcher: something that reads the draft, holds every commitment in it, and reports what no longer agrees. It is the feature the problem seems to ask for, and it is not what Telleny does.

Telleny does not keep your continuity, does not audit a manuscript for contradictions, and will not tell you that chapter thirty has broken a promise chapter three made.

It remembers what you have told it and retrieves what you have written. Those are smaller words on purpose. Remembering is holding material you gave it; retrieving is bringing a piece of it back when the passage in front of you calls for it. Neither is a claim about whether the two pieces still belong to the same book.

That last step is the reading, and the reading is where a book is actually decided. A tool confident enough to perform it would be wrong in a particular and expensive way: it would flag the deliberate inconsistency, the unreliable narrator, the fact a character is lying about, and the change of voice that is the point of the chapter. A writer who learns to trust that report has handed over the judgment that makes the manuscript theirs.

What retrieval is good for

Four modest things, each of which leaves the reading where it belongs.

  1. Surface

    Bring back the earlier passage the scene in front of you has started to argue with, so you can read the two together.

  2. Connect

    Set a decision you made months ago beside the one you are making now, without proposing which of them should win.

  3. Reflect

    Say back what a chapter appears to be doing, in plain terms you can disagree with, so the disagreement is worth something.

  4. Remember

    Hold a question you are not ready to answer until the manuscript has travelled far enough to answer it.

The unit is a sentence

Accept or reject is the wrong question to ask about a revision.

Most tools that propose a change offer two answers to it, and both of them are wrong most of the time. A revision to a paragraph is rarely wholly right or wholly wrong. It is usually a good second sentence attached to a first sentence that has lost the rhythm, or a clarification that costs the hesitation the passage needed.

Faced with two buttons, a writer either accepts a paragraph for the sake of one line inside it or refuses the line for the sake of the paragraph around it. Both are losses, and neither is visible afterwards. The draft simply ends up slightly further from the book, and nothing in the interface recorded why.

So a proposal in Telleny is marked span by span and decided the same way: keep two sentences, refuse the rest, and what you refused does not reach the chapter. That is a smaller mechanism than a smarter model, and it does more for a manuscript, because it lets the writer answer at the resolution the question was actually asked at.

A drawing of one thread folded down a page, and beside it the same thread with the same folds, cut in a number of places. Nothing was added between the two and nothing taken away; the second differs from the first only in how many places it stops. It stands for the difference the passage beside it draws — between answering a revision once, and answering it where its parts are.

Why it does not decide

The refusal is not modesty. A tool that decided would cost the book the thing it is for.

A tool that settled the question would be pleasant to use and quietly ruinous. Its judgments would be plausible, evenly distributed, and unattached to the one thing that makes a novel worth reading, which is that somebody chose this and not the near alternative for reasons they could not fully explain at the time.

Fluency is the trap. A well-formed alternative reads as a correct one, and the difference between the two is exactly the judgment the writer is being asked to keep. Refusing a fluent sentence is harder than accepting it, and a surface that makes acceptance the path of least resistance has decided the book by ergonomics.

This is why the same refusal is written into the product and into these pages. It does not grade a manuscript, score a version as better, or rank one reading against another. It can report what is on the page and bring back what is elsewhere in it. What the two mean together is not a fact, and it is not the tool's to state.

Speed is not the measure

Finishing matters. So does recognising what a draft needs before it can honestly move forward.

There are moments when the next step is clear and the work gathers speed. There are other moments when a writer needs to slow down enough to tell the difference between a temporary difficulty and a decision the book is asking them to make.

Neither pace is a moral achievement. The aim is not to protect hesitation for its own sake, nor to make deliberation look more serious than action. It is to give each part of the process the kind of attention it needs.

That is a quieter promise than a shortcut, but it is the promise Telleny can make honestly: to support a return to the work rather than treat speed as the only evidence that the work is going well.

Make room for the return

Sign in