← all games

Eleusis

#5217BGG ↗

1956 · 4-8 players · 20min · weight 2.93 · 234 ratings

v2 v3 v4 wide v4 deep

BGG raw

ID
5217
Name
Eleusis
Year
1956
Rank
7175
Min players
4
Max players
8
Playing time
20
Min playtime
20
Max playtime
20
Avg weight
2.9333
Num weights
15
Bayes avg
5.66113
Average
7.31709
Users rated
234
Num owned
242
Wanting
10
Wishing
129
Num comments
103
Fetched at
Wed Apr 29 2026 05:34:56 GMT+0000 (Coordinated Universal Time)
Mechanisms (2)
InductionPattern Recognition
Categories (3)
Card GameDeductionEducational
Description (1630 chars)

Eleusis (and later New Eleusis, 1976) is a card game that simulates the scientific method in action. You need two packs of cards, a table and some paper and pencil to play. The 4-8 player range works best. One player ("God", "Nature" or the "Demiurge") thinks up a rule (a "law of nature") that governs the correct play of the cards. Usually, the rule only involves the most recently correctly-played card, but it can be more complex. The other players ("Scientists") take turns playing cards ("performing experiments") and race one another to see who can come up with a good hypothesis about the rule. The first player to correctly deduce the rule scores big; God also scores based on how long it took for the players to figure the rule out. God should not choose a rule that is so hard to find out that nobody does, because then he doesn't score. Additionally, a player with an hypothesis can declare himself to be a "prophet" who can predict the results of the other player's experiments. Other players can then try to bring about the overthrow of the prophet by trying to find experiments whose results cannot be predicted (falsification), thus gaining a chance to become prophet themselves. A very simple rule would be "Alternate between black and red cards". A simple but hard to discover rule would be "Alternate between cards which have closed loops in their number or letter designations (e.g. 4, Q), and cards which don't (e.g. 2, K)". The game was first published in Martin Gardner's Scientific American column in June 1959. A revised version appeared in Gardner's October 1977 Scientific American column.

LLM v2 (wide)

Not yet enriched at v2 (wide pass).

LLM v3 (deep)

Not yet enriched at v3 (deep pass).

LLM v4 wide (controlled-vocab primitives)

Not yet enriched at v4 (wide pass).

LLM v4 deep (archetype fit)

Not in the v4 deep-pass top-20% slice.