What’s changed

Telling Maestro it has got something wrong

Your reading of your own book outranks ours. These changes are about making that easy to say, and making it stick.

Disagree where the suggestion is made

You could always tell Maestro it had misread something, but the only door to it was Ask Maestro — a separate screen, a question to compose and a session spent. The writer most likely to want to say 'that repetition is deliberate' is the one reading the sentence that says it isn't, and at that moment the product offered nothing. There is now a reply button beside the finding itself.

Correcting your genre actually changes your genre

A correction was stored as a note for the next read while the project's own genre stayed as first inferred — so the craft lessons, the reader panel and the diagnostics all carried on using the wrong one. You now choose from the catalogue and that is what the whole product uses.

Change your mind, or take it back

Corrections and clarifications can now be edited or removed wherever they appear. Where a correction has already been written through to a character's profile, that stays as you set it, and the confirmation says so.

Checking we read your book correctly

Everything Novel Maestro tells you rests on having read the right text, divided the right way. These are the checks that make that visible to you rather than assumed by us.

Your Text — the manuscript as we parsed it

A new step in the First Take shows your book as Novel Maestro divided it: every chapter, every scene beneath it in order, each with the working title it was given. A check that cannot fail is theatre, and these do fail — one manuscript arrived as 25,000 words against a declared 40,000, and an earlier import produced nineteen empty chapters. A writer who looks at that and says 'that isn't my book' has been served; one shown a reassuring tick has been handled.

The questions your read raised, finally shown

Every read has always ended by raising three questions about your manuscript that only you can answer. They were being generated and displayed nowhere — 255 of them across 87 reads. They now have a page of their own, and your answers go into everything that follows.

A draft that has lost fifty thousand words is a question, not a statistic

A writer uploaded a single chapter of a 57,000-word novel as a revised draft. The figure went past as one more measurement, and ten minutes later a report congratulated her on the deletion. A suspected shortfall is now raised as a question on the page, before anything is read. The read still starts — being asked is enough, being blocked would be presumptuous.

How much changed, measured on sentences

Deciding whether a revised draft needs a full re-read means measuring what actually changed, and the obvious units all lie. Scenes are rebuilt after every import. The parser does not divide a book the same way twice — a 33,000-word novel parsed as 11 chapters one day and 23 the next. Even paragraphs do not survive a round trip through Word, which turned a 10% change into a measured 16%. Sentences survive all of it, so sentences are the unit.

The beta reader panel reads the same book the same way

Caught by a beta tester who ran the panel twice on two versions of one manuscript eleven words apart, and got two different readings. The panel's excerpts were chosen by scene number, so they moved whenever the scene count did, even when the prose had barely changed. They are now anchored to positions in the book.

When a read fails

A read that fails costs you nothing

And the message that follows now only says a credit has been returned when one actually was — it previously said so even after reads that had never taken one.

Your report on paper

The Developmental Report is now one document rather than twenty separate downloads, and a week of reading it closely turned up a good deal that needed fixing.

One document instead of twenty-odd files

The editorial assessment leads, and everything that was measured follows it — the diagnostics, your cast's arcs and the beta reader panel, all in one place. A tester had downloaded everything, then downloaded each section again to be sure the combined file was not merely a summary. It wasn't the right way to find out.

Findings in the report, evidence in the app

Each diagnostic chapter now carries its full reading and a sample of the measurements behind it, then points you at the page where the whole table can be sorted and followed back into your manuscript. The report had grown to eighty-odd pages for a 45,000-word novel, three quarters of it apparatus.

Tables that are really prose now read as prose

Where a column held whole paragraphs, the grid squeezed them into five-line slivers a dozen characters wide. Those entries are now set as text. Genuine tables — a cast list, a set of percentages — are unchanged.

Instructions you can find at a glance

In the lists that tell you what to do — what to focus on, dangling threads, scenes needing purpose — the instruction itself is now set in bold, with the reasoning after it.

A contents page you can click

Reported by a beta tester: the contents in our PDFs were a plain list. Every chapter is now a link.

Momentum was reading your book out of order

The scenes behind the energy curve were being read in the order the database happened to return them rather than in reading order, so positions, the opening and closing comparison and the quiet stretches were all computed over a shuffled manuscript. They are now read in the order you wrote them.

Momentum no longer contradicts itself

The chapter could open by saying there was not enough data to read your book's shape and then print its energy figures underneath. The measurements were there all along; the report simply never asked for them.

Long numbers, cut to size

An energy figure printed as 3.5416666666666665. It now reads 3.5 of 10, which is both shorter and clearer about what the scale is.

No more rows split across a page

A row that ran past the bottom of a page used to continue under a repeated set of column headings on the next one, stranded from the row it belonged to. Rows are now carried whole.

Missing characters

Arrows, ticks and a minus sign had no glyph in the report's typeface and were being dropped silently, which turned a scale running from minus two into one that appeared to start at two, and broke the word after every missing character. They now print.

Blank chapters and raw data

A diagnostic that found nothing produced a chapter heading over an empty page, and one chapter was printing a page of unformatted machine output. Neither appears now.

Your manuscript, your cast

Characters listed by how much they are in the book

The cast was ordered alphabetically within each role, which put a minor character with an early initial directly beneath the protagonist. Presence now comes first, with the alphabet as the tiebreak.

Characters present, limited to your actual cast

The scene panel could list dozens of names picked up from the text. It now shows only the characters on your cast list, each one clickable.

Exports

The beta reader panel can be exported on its own

It could be read on its page and nowhere else: no combination of choices in Export included it. It now has its own entry, and travels inside a combined report.

One report covering every character's arc

Asked for by a tester — 'I can pull these up individually, but why not let the computer do the work?' Looking into it turned up something larger: character arcs could not reach a combined report by any route at all.

Answers you can keep

An Ask Maestro answer can now be saved as a note you own — editable, taggable, filed with the rest of your work rather than held as our record of an exchange.

Every report as an editable Word file

Asked for by a tester who works on her notes rather than in the product: she strips the formatting, cuts what is not actionable and regroups the rest before building a task list. A PDF did not stop her — it made her fight the format while she did it.

Before the read

Room to tell us about your book

The note you write before a read was capped at 1,500 characters. A tester asked for double, and it is now 3,000 — with a box the size of the thing it is asking for.

Somewhere to say this is book two

There was nowhere to tell us a manuscript was part of a series, which matters: a thread left open for the next book is a decision, not a loose end. There are now fields for the series and which book this is, and the read treats a deferred thread accordingly.

Finding your way around

Scene references that look like links

A reference to a passage in your manuscript only underlined itself when the pointer was over it, and two testers independently failed to find it. On a touchscreen there was no way to discover it at all. References are now underlined at rest.

My Novels, in the trail

From a tester's note. The first step of the breadcrumb left the application for the public site, so following the trail back from a book landed you on the marketing home page. The trail now starts at your shelf.

A way back from Export

Adding a diagnostic to your reports sent you to Export and left you to find your own way back through the menu. There is now a single link to the section you came from.

Genres grouped the way you think about them

The genre menu is now arranged in the same families as our genres page, rather than one long list.

A dead button at the end of the First Take

Reported by a tester: a button reading 'Continue to Continue' that did nothing when pressed. For anyone with full access there was no step after Your First Move, so the last page of the First Take offered a door to nowhere. The page now ends somewhere real.

A revised draft no longer pretends to be finished

Importing a new version used to take you straight to a First Take describing the previous draft, with nothing to say a fresh read had started four seconds earlier. You now get what changed, stated as a measurement, with the read's progress beneath it and the way in offered once there is something to read.