ReplayUnderstand
Every match
is evidence.
A match is a seed and a list of actions, so it can be played again exactly. Replay steps through it event by event, names the rule behind each one, forks what-ifs and checks that the result still hashes the same.
LiveRe-executed in your browser
Scrub, fork,
compare.
If you played a match on the Table, it’s here. Otherwise this is a sample bot match on your current rules. Fork at any step to see how the game would end under the ruleset in Studio now.
Sample match · seed 1208
ruleset v0.4.0 · 29 actions · 100 events · Seat 1 wins
Seat 1 ended the turn. Combat resolved.
- Cinder Drake dealt 4 to Ironback.§3.1
- Ironback dealt 3 to Cinder Drake.§3.1
- Cinder Drake was destroyed.§3.3
- Spark Imp dealt 1 to Seat 2’s Keep.§3.1
- Seat 1 wins.§6.1
What if, from here?
Fork step 29 and let two ghost seats finish the match under the ruleset that is in Studio now (v0.4.0).
Break-It LabPlaytesting at scale
A thousand games
before lunch.
Choose how many games, press the button and watch the count. Findings name the rule involved and offer a fix. Apply it, run again, and the history table shows whether it helped.
Version compareComing to Replay
Two rulesets,
one diff.
Replay compares versions: the same seeds under v0.4.2 and v0.4.3, side by side, with the rule changes that explain the difference.
v0.4.2
first seat 61.7%
§2.2 The first player does draw on their first turn.
v0.4.3
first seat 50.1%
§2.2 The first player does not draw on their first turn.
The percentages are real: seeds 1,000–2,999 on this site’s engine. Reproduce them in the Lab: rule on A-1, apply both seat-balance fixes and run 2,000 games.