Methodology and randomness

Purpose

The app treats a reading as a prompt for attention and reflection, not as a prediction or a source of facts about the future. You can write an intention, choose cards, record an initial response, and compare it with fixed guidance.

Three guidance timings are available: self-led hides guidance, reflect-then-compare reveals it on request, and guided shows it with the reading. These modes change when existing text appears; they do not generate a different answer for you.

How cards are prepared

• You choose the full deck, Major Arcana, or one suit.
• The app copies that pool and shuffles it in the browser with the Fisher–Yates algorithm.
• For a one-card reading it presents up to 12 face-down candidates. For a three-card reading it presents up to 15.
• You select the face-down candidates used in the reading.
• When reversals are enabled, the app makes a separate random yes-or-no choice for each selected card.

The app requests random values from the browser’s Web Crypto API when it is available. If that API is unavailable, the current implementation falls back to JavaScript’s Math.random(). The interface does not claim that a draw is suitable, favourable, or meaningful because of its random result.

Interpretation

Card guidance is a fixed catalogue shipped with the site. It is not generated from your question or journal text. The current catalogue offers Rider–Waite–Smith symbol, narrative, and elemental reading lenses.

The Rider–Waite–Smith lens links to A. E. Waite’s The Pictorial Key to the Tarot as a historical source. The app’s short prompts are editorial summaries, not quotations and not universal card definitions. See Sources and rights for the current source boundary.

Limits

The project does not certify the randomness implementation, audit a device’s entropy source, or test the statistical properties of each browser. A browser without Web Crypto uses the weaker fallback described above.