Lucidthread
Watch a reader’s mind move through your writing.
A fully client-side model of a reader’s evolving mental state, built on Centering Theory, coreference resolution, and discourse working memory — every number traceable to an explicit, inspectable rule, never a black-box score.
Live demo, looping automatically — the same components you’ll use on /analyze.
The science
Centering Theory
Grosz, Joshi & Weinstein (1995)
As you read, your attention tracks a “center” from sentence to sentence — usually whoever or whatever is the grammatical subject. When the next sentence keeps that same center in the spotlight, reading feels effortless. When it jumps to something else without warning, you have to stop and re-orient.
“Maria opened the door. She smiled.” keeps Maria in focus — easy. “Maria opened the door. The room was freezing.” shifts focus abruptly — harder.
Coreference & Ambiguity
standard discourse coreference literature
Pronouns like “she,” “it,” and “they” only work if the reader can tell what they point back to. When two people or things both fit, the reader has to guess — and sometimes guesses wrong, then backtracks once the sentence stops making sense.
“Maria met Sara at the cafe. She ordered a coffee.” — who ordered, Maria or Sara? Both are grammatically possible.
Working Memory & Information Structure
given/new information structure; discourse working-memory research
A reader can only hold so many people, ideas, and unresolved questions in mind at once. Sentences that pile on brand-new information, or introduce something before it's been explained, cost more effort — even when every individual word is simple.
“The committee approved the library, the gymnasium, the after-school program, and the transit subsidy.” — four new items in one breath.
Reading the timeline
Every visual choice below maps to a specific, inspectable rule — nothing here is decorative.
A distinct entity — a person, place, or thing — colored from the Okabe-Ito colorblind-safe palette in the order it first appears.
How present that entity is in working memory right now: thick when it's the sentence's center of attention, thinner as attention drifts, tapering to nothing once it's dropped out of memory (unmentioned for more than 3 sentences).
Continue — easiest possible link back
Retain — same topic, attention starting to drift
Smooth shift — topic changes, but cleanly
Rough shift — topic changes abruptly
No link to the previous sentence
A pronoun could plausibly refer to two or more different entities within the last 2 sentences.
The center of attention changed to something that isn't even this sentence's main focus — the most disorienting kind of transition.
This sentence shares almost no vocabulary with the one before it — a break in the flow of topic.
This sentence's load is unusually high relative to the rest of the paragraph.
A pronoun with two or more equally plausible antecedents. Dashed ghost branches point toward every candidate the pronoun could mean, alongside the tool's best guess.
Height = how much effort that sentence takes to process, from the working-memory load formula. Hover or tap any point for the exact breakdown — entities in play, unresolved references, shift penalty, and new information.
Who this helps
Writing instruction
Show students exactly where their prose loses the reader — not just “this is unclear,” but which sentence, which pronoun, and why.
Technical, legal & medical writing
In instructions and contracts, an ambiguous “it” isn't a style nitpick — it's the sentence someone gets wrong. Catch it before a reader has to guess.
ESL & multilingual writers
Reference and information structure are exactly the parts of English that don't transfer cleanly from other languages. See where a native reader's mental model might diverge from what you intended.
Readers with ADHD or working-memory differences
The cognitive-load curve makes visible what's usually invisible: which sentences ask the most of a reader's working memory, so writers can lighten exactly those spots.
Ready to see it on your own writing?