Most accessibility standards target one need at a time. Large print for low vision. Plain language for cognitive load. Alt text for screen readers. Each works on its own, but they rarely combine into a single writing practice. CARAS treats accessibility as a property of the text itself, not a layer added afterward. It governs how a sentence is built, how a paragraph is paced, and how a chapter is structured, so the same content reaches readers with very different processing needs without producing five different editions.
CARAS combines five integrated sub-standards. Each addresses a specific reader profile, and the rules overlap in ways that reinforce one another.
Roughly four percent of readers have aphantasia, the inability to form mental images. Standard prose that leans on visualization can read as empty noise to them. APAS requires every passage of imagery to carry a literal anchor, a plain restatement of what the imagery is doing in the argument. Readers who visualize get the picture. Readers who do not get the meaning.
ADHDS controls cognitive pacing. Sentences stay short enough to land. Paragraphs end on a beat, not a clause. Headings are frequent and load-bearing, so a reader who looks away can rejoin the text without rereading. Lists are used where lists belong, and not used where prose works better.
CVDS governs every use of color in a Cinderpoint title. Color never carries meaning on its own. Every chart, diagram, and callout uses shape, label, or position alongside hue. Palettes are chosen so that protanopia, deuteranopia, and tritanopia all preserve the same information.
LEAS sets the typographic environment. Body type runs at sizes and line heights that hold up across long sessions. Contrast is high but not harsh. Background fields avoid pure white where the medium allows. Footnotes, captions, and metadata are sized so a reader does not have to lean into the page.
TLAS protects meaning across languages. Idiom, regional metaphor, and culture-specific shorthand carry a literal restatement nearby. Sentences are built so that a translator, human or machine, has a clean structure to work from. The result is a text that survives translation without losing the argument.
CARAS recognizes that not every Cinderpoint document is a book. The standard defines two modes, and a title selects one based on its purpose.
Narrative Mode is for books and long-form papers. Prose flows. Voice is present. Imagery is used freely, but every figurative passage carries its APAS anchor. American Soccer Cartel is written in Narrative Mode.
Instructional Mode is for tools, manuals, and reference material. Sentences are short and directive. Steps are numbered. Voice recedes so the reader can act. SAFEMACHINE documentation is written in Instructional Mode.
The mechanism that ties CARAS together is dual-layer writing. Layer one is vivid. It carries imagery, voice, and rhythm, the things that make a sentence worth reading. Layer two is literal. It restates, plainly, what the vivid sentence means. The two layers sit close together in the text, often in the same paragraph, sometimes in the same sentence.
For an aphantasic reader, layer two is the meaning. For an ADHD reader, layer two is the anchor. For a translator, layer two is the source of truth. For a fluent reader with no accessibility need, layer two reads as natural reinforcement, the kind of writing that simply feels clear.
CARAS is written for readers with aphantasia, readers with ADHD, readers with color vision deficiency, readers managing eye strain, and readers who encounter the work in translation. In practice, those groups overlap, and the techniques that serve any one of them tend to improve the experience for everyone else.
CARAS is at version 4.2. The full guidebook is maintained internally at Cinderpoint and governs every title released under the Press imprint. A public reference edition is planned.
Founder of Cinderpoint Systems LLC. M.S. Artificial Intelligence (MSAI), M.S. Management (MSM). Researches how systems fail under speed, opacity, and scale.