Re-verified on this request
The Descent · The Clockwork Descent
The no-cheat leaderboard. Procgen rows are re-executed from passage choices and required to reach the hoard. Browser-native rows are replayed under their distinct Lean-native verb/receipt rules and required to crown. A forged or unfinished run appears in neither lane: exclusion comes from re-verification, not a stored flag.
Day
today
Seed
b0a53e00
Warden HP
45 of 60
Corridors
2 of 3
Procgen choice-tape ruleset
Winning procgen runs
Each row re-executes its recorded passage choices in a fresh world drawn from this day's seed and is required to reach the hoard. depth is how deep that replay actually got. One row per human, not per account: a player who has proven that their web and chat identities are the same person ranks once, on their best run. An unlinked player is their own human, so nothing groups by accident.
| # | player | turns | depth | proof |
|---|---|---|---|---|
| 1 | emberattributed | 9 | 5 | verify this run |
Lean-native browser ruleset
Crowned native runs
These records are not converted into the procgen choice tape above. Each row replays its native delve / smite / loot / unlock / flee events and exact receipt, state, and journal-root envelope. Only a crowned settlement ranks. This lane groups one row per human on the same rule as the table above: the actor is taken off the replay, and a proven cross-platform link is what makes two accounts one row.
What score is, and what it is not. It is the relic depth a run carried out: every banked relic counts the floor it was minted on, read off this day's own drawn map, so the deep ones are worth what the rules make them cost to carry up. It is a read over facts the replay already committed: no bonus, no weighting, no tiebreak. ⚑ The ranking is still turns, not the score: deciding an order is a rule a player could argue with, and a rule belongs with the game's other rules rather than in the page that draws them. So read the score as the size of the haul and the rank as the speed of it; they are two different questions and this table answers both without pretending either is the other.
No crowned native browser run has re-verified for this day.
Re-verification decides which runs appear here, and it is about the moves: every row was re-executed under the rules and reached the end. Who played it is a separate question, and this board answers it separately. The mark beside each name says how that name was established, and it is checked on this page load exactly as the moves are: a run that arrived with a signature has that signature verified again, right now, against the run it came with. A run that arrived without one reads as attributed, which is the ordinary way a browser plays and counts for just as much. If you want your own runs to carry the stronger mark, let this browser sign for you.
How each name on this board was established
- signed
- Every move carried a signature made on the player's own device. This server never held the key that made it.
- signed by the bot
- Every move carried a signature, made with the key the chat bot derives for that account. Whoever runs the bot can derive that key too.
- attributed
- The name is the one the player was signed in as, recorded without a signature. This is how playing in a browser normally works.
All three are real play and all three count. The difference is what was checked, not how good the run was: a signed row is a stronger claim about who played it, and an attributed row is the ordinary way a browser plays. If you want your own rows signed, your identity page has the one switch that does it.