The comparison & result
The test policy switched to a discard-history heuristic whenever an opponent showed two melds. The current harness prefers a tile that appeared anywhere in the discard history; that does not establish safety against the threatening player. The normal policy continued to consider its own hand rather than folding at that fixed trigger.
The reflex-folding policy won less often without a reported improvement in deal-in frequency. Exposed melds are useful information, but this experiment does not support treating their count as a complete decision rule. Your distance from a legal win, hand value and available safe tiles still matter.
At your table
- Assess your own progress alongside the visible threat.
- A near-ready hand and a distant hand need not respond in the same way.
Check your table’s scoring rules before applying the finding. Read the Hong Kong reference.
How it was tested
Published experiment: 100,000 simulated Hong Kong hands, fold-at-two-melds versus the normal policy, with alternating seat assignments.
The players were software policies, not human participants. This report describes the published experiment; it does not establish the best move for every possible hand or opponent.
Results without a published uncertainty interval should be read as point estimates. Similar results do not prove two choices are equivalent.
Read the Lab’s method and limitationsPublished by Hé Fēn Mahjong. Ask about this finding or suggest a correction.