Resources for Writers

The books do not begin as abstract publishing ideas or market-led concepts. They begin with a course, a practical project, or a sustained question that has already required real work.

The project comes first. A book follows when the work reveals a problem worth documenting, explaining, or testing more deeply. This keeps the manuscript anchored in an actual body of research and practice rather than in a premise chosen solely because it appears commercially attractive.

Research commonly begins through an online course, a practical build, or both. As the work develops, a gap often becomes visible: an important question that the course material, available guidance, or existing literature does not adequately answer.

That gap becomes the subject of independent investigation. The work may draw on public datasets, documented examples, and incidental data - information collected for another original purpose but potentially useful for testing a related question. Incidental data is treated cautiously: relevance, limitations, provenance, and the difference between association and proof all matter.

Sources, observations, decisions, and changes are logged while the research is underway - not reconstructed retrospectively. The aim is to preserve an auditable record of how a conclusion emerged, including uncertainties, contradictory evidence, and findings that did not support the original assumption.

Each book starts with a hand-written outline: the central spine of the argument, the order in which a reader needs information, and the boundaries of the project. The outline is then tested against practical measures such as cohesion, readability, scope, evidence coverage, and whether each chapter advances the reader's understanding.

Before the manuscript is drafted in full, the research is developed into a white paper. This establishes the evidence base, key definitions, claims, methods, limitations, and source trail before the work is expanded into a book-length form.

The manuscript then grows outward from that foundation. Rather than writing chapters as isolated pieces of content, the process builds a connected argument: context first, evidence next, interpretation after that, and practical application where it can be supported.

Editing begins away from the screen. A printed draft is read and marked up by hand using highlights, margin notes, structural annotations, and source checks. Every factual change is verified against the relevant source before it is accepted into the manuscript.

Finished drafts are shared with three trusted readers and one independent external reviewer before being treated as final. Their feedback helps identify unclear passages, missing context, weak transitions, repetition, unsupported assumptions, and errors that are difficult for an author to see after extended work on a text.

AI is used only at the final parsing stage and only with constrained, documented prompts. That boundary comes from experience: an earlier uncontrolled AI-assisted pass caused material damage, including lost formatting and duplicated text. AI assistance is therefore never treated as invisible editorial labour. Where it has materially contributed to final-stage parsing or checking, that use is disclosed.

Manual proofing remains essential even where software is used. Amazon KDP itself recommends manual proofreading in addition to spelling and grammar tools, and its production requirements distinguish the manuscript file from the separately supplied cover file. [kdp.amazon]

Before any file goes to conversion, the manuscript now goes through a separate structural QA pass, run the same way every time: a placeholder and leftover-text scan, a check for blank or empty heading styles, table-of-contents and bookmark integrity, and a cross-check that every ISBN and AI-use disclosure matches across every file for the edition. This catches a class of defect a normal read-through misses entirely, because it is structural rather than a prose problem - and it exists because one of those defects was found on a real manuscript, not a hypothetical one. Pre-Publication QA Checklist →

Covers are designed personally in Canva after researching the book's category, audience expectations, visual conventions, typography, and technical production requirements. The purpose is not simply to make a cover attractive, but to make it legible, appropriate to the subject, and recognisable at small sizes as well as in print.

AI-generated imagery is avoided unless there is a clear and necessary reason to use it. This is a creative and risk-management choice: Canva permits the use of designated AI-generated content under its licence terms, but such material is identified as AI-generated and its use remains subject to the relevant terms. Where third-party fonts, images, graphics, or templates are used, their commercial rights and licence conditions should be checked for the specific design and publishing context. [canva]

Files are prepared for the relevant format rather than treated as interchangeable. Print books require a full cover design and production-aware layout; KDP specifies bleed requirements for book covers, while print specifications such as trim size, paper choice, margins, page count, and bleed affect the final file setup. The uploaded files are then reviewed in the platform's preview process before publication. [kdp.amazon] [kdp.amazon]

Print interiors and covers are now built to a fixed technical spec rather than eyeballed - PDF/X-1a or PDF/X-3, CMYK throughout, fonts fully embedded, bleed and margins locked in before a cover template is even generated. Ebook files are converted to EPUB 3.0 and run through an independent validator before upload, with every metadata attribute and image size checked against the platform's own file spec rather than assumed correct. Both processes are documented for reuse on the next title: Print File Specifications → and EPUB Conversion & Validation →

Each title is now set up as two separate records - print and ebook - each with its own ISBN, and published wide: direct through Amazon KDP for the Amazon market, and through IngramSpark for bookstores, libraries and every other retailer. Category, keyword and description fields are written and checked by hand rather than accepted from a platform's AI-assist suggestion, which has repeatedly missed half a book's actual subject matter when left uncorrected. Book Metadata & Categorization → · IngramSpark Title Setup → · KDP + IngramSpark Dual Distribution →

Publication is not treated as the point at which a book becomes beyond revision. Readers are given a visible route to report factual errors, broken links, unclear passages, or other problems they find.

Confirmed corrections are recorded openly on mindfulmachinespress.com rather than being quietly overwritten or hidden. Each correction should identify what changed, why it changed, and when the correction was made, while preserving a clear relationship to the affected edition or text.

This approach follows a simple principle: readers should be able to see when a published record has changed and understand the nature of the change. Publishing-ethics guidance similarly emphasises transparency, clear correction notices, and links between corrections and the original work. [publicationethics] [wiley] [csescienceeditor]