Help and FAQ

Young people aged 10–17, through a family created by their parent or legal guardian, and adults 18+ in the separate Open Division. People take part from any country.

One shared station makes every error comparable and fair. Boston's winter weather is genuinely hard, and its data is public and well documented. Where you live never changes your score.

The cutoff is 18:00 EST (23:00 UTC) on the day before the target day. Boston uses standard time for climate days all year, so the UTC time never moves. Every round card shows the cutoff in three lines: Boston (EST), your local time as your device reports it, and UTC. The countdown uses the server clock, not your device clock.

Most of the world thinks in Celsius, and one decimal lets you express fine judgment. We store your forecast as an integer number of tenths (14.0 becomes 140), so there are no rounding surprises. The observation stays in exact whole °F and we compute the error as an exact fraction. If you think in °F, the forecast form has a converter that shows the °C value it will submit.

error = |9t − 50F + 1600| / 90 °C, where t is your forecast in tenths of °C and F the observed whole °F. Example: 14.0 °C against 57 °F gives exactly 1/9 °C, shown as 0.11. The Open data page has a calculator.

Yes, as often as you like before the cutoff, up to 20 versions per round. The latest version received strictly before the cutoff counts. Every version keeps its own receipt.

Nothing bad happens. Missed rounds are never penalised or filled in. Your summary shows how many rounds you completed out of how many you could have joined. Reach 60% coverage to earn the consistent participation badge.

The round shows "Waiting for data" and is scored as soon as the first complete NWS climate report arrives. If data is more than 48 hours late we say so on the round page; after 14 days without data the round is void for everyone and does not count against coverage.

A result is provisional for 7 days after we take the observation snapshot. If NWS corrects its report in that window, we re-score and add a new version. After 7 days the result is final.

Yes. Any public source, model or AI is allowed. You must say honestly, with each forecast, whether you used AI. The answer never changes your score; it helps everyone learn how people forecast.

Within 5 calendar days after a result is published, the guardian (or an Open Division participant) opens History in the family hub and presses Appeal on that round. We re-run the public scorer on the frozen inputs; the re-run decides, and any change is added as a correction version.

Sign in with the Family Key, open Participants, choose Export JSON first if you want a copy, then Delete and type the codename to confirm. Identifying links are removed within 72 hours; only anonymous hashes remain in the ledger.

We cannot recover it, because we keep no email or name to prove who you are. Create a new family. The old family's data stays private and is de-identified on schedule.

In the family hub, choose Issue Student Pass for that participant and give the code to your child. It only allows forecasting and viewing history for that one participant. Issuing a new pass cancels the old one.

A few rounds are mostly luck, and comparing children before there is enough data is unfair. Season 1 shows each person their own errors, reference lines and coverage. Any future ranking needs a new rules version with strict gates published before the season.

Download the receipt after you submit. On the Verify page your browser recomputes its sha256 and the platform checks it against the ledger. After the cutoff, the round page links to a frozen manifest of all on-time receipt hashes.

No. It is free for families, and paid tools never add points.