Highlighting a single word now speaks for you. If you pick just "cats" while reading a manuscript, Anketta reads it as "I like this concept" and gently leans your feed toward people who also write positively about cats. Crossing out a single word is a stronger signal — if you cross "sauna", authors who clearly love sauna (whether they said it in their manuscript or crossed/liked it themselves) stop appearing in your pool. The soft edge to this: if another reader also crossed "sauna", you two still match — shared dislikes count as compatibility, not as avoidance. You'll see your word choices collected in Settings under "Hidden words" and "Liked words", where each entry is one tap away from being removed if you change your mind. To protect against accidental highlights: a crossed word needs to be reinforced a few times before it actually starts filtering your pool, there's an hourly cap on new words per direction, and any cross you stop reinforcing for a couple of months softens automatically and is eventually removed — a moody evening from last quarter doesn't become a permanent filter.