Fourfold guide

Ratings and assistance

Player and puzzle rating make fair matches possible while preserving a separate place to test new work.

Ranked

Ranked pairs signed-in players with established puzzles near their rating. Solve every quartet with no more than three mistakes and you win; the fourth mistake or giving up is a puzzle win. An eligible result moves both ratings.

Discovery and Casual play

Discovery helps new, human-approved puzzles find their level. An eligible result can move the puzzle's rating, but never the player's rating or ranked record. Daily streaks are separate: a qualifying daily win can count in Discovery too. First plays automatically use Discovery for calibrating puzzles and Ranked for established puzzles, including when opened from a profile or shared link. Casual is reserved for guests, your own puzzles, repeat plays, and puzzles needing revision. Unfinished games retain their original mode.

Finding your level

Guests and players without an unassisted rated result start with recommendations around 1,200 difficulty. This is an easier browsing target, not an earned player rating. Change the difficulty filters anytime. After rated play starts, recommendations use your actual unassisted rating. Starter practice never affects ratings.

What can be rated

Only a signed-in player's first exposure to a puzzle version can count. Guest play, an author playing their own work, previews, replays, and practice after a loss are unrated. Once you make a guess or reveal a spoiler in Ranked, resume or concede that attempt before starting another Ranked puzzle. In Discovery and Casual, you can move on whenever you like and resume unfinished puzzles later; moving on does not settle a result.

Assistance has its own track

Revealing a creator-provided quartet hint before settlement ends the unassisted leg as a puzzle win. You can then continue the same board against that version's assisted rating: solve before the fourth total mistake for an assisted win. Revealing specialist topics or a spoiler-hidden structure label does not affect rating. Fourfold shows both rating legs and their net effect separately, and reveals after settlement cannot change either result.

Elo methodology

Player Elo begins at 1,200 as a provisional calculation baseline. New puzzles begin at 1,300, 1,500, or 1,700 according to the creator’s difficulty estimate. Expected results use the standard 400-point Elo scale. Players move more quickly during their first 20 eligible results. Puzzle ratings move quickly while new, then become more stable after 30 and again after 200 eligible results. Assisted and unassisted puzzle ratings remain separate whenever creator hints are available.

Fourfold retains the rating history needed to audit a result and correct invalid outcomes without silently rewriting earlier matches.

Was this guide helpful?

Tell us what worked or what is still unclear. Feedback is private to the Fourfold team.