# Original platform-decision exercises

Six original bilingual case sets and blank private worksheets. All people, post counts, plans and outcomes are constructed. These are not actual account tests, native uploads, customer outcomes, reader observations or another research study. The official-help retention conflict is documented, not resolved by these examples.

Read the Markdown files directly; no installation is needed. Keep completed notes private. No form here sends your notes, and no account settings, schedule or analytics connection is changed.

Optional fixed-fixture reproduction: extract the complete ZIP into a folder and, with an existing Node.js 22 or newer, run:

    node reproduce-examples.mjs

No package installation, network or account operation. It checks six constructed equal-age rows, medians, a declared effort condition, three date-specific timezone conversions, two repeated-hour instants, limited scope arithmetic and file identities. It uses the installed runtime's timezone data; future rule changes need review.

It does not validate completed records, underlying truth, consent, permissions, notification delivery, restored viewer history, actual scheduling, privacy guarantees, language, accessibility or audience effectiveness. A higher constructed impression median is not a causal gain. A current source link is not proof that your account exposes a feature or that a metric remains accessible. No real private files or account credentials are needed.