%2Fpic4597095.jpg&w=3840&q=75)
1999 · 2-2 players · 30min · weight 1.47 · 50,813 ratings
At a glance — v4 wide
Controlled-vocabulary primitives + 8-axis MDA aesthetic vector. Vocab v2.
Play a card to an expedition column or discard to a pile, then draw; each started column must clear its -20 threshold.
- [3]hand_management_under_draw— “play one card up an expedition or discard, then draw; committing to colors that must clear -20 threshold”
- [2]secret_objective_card— “investment_multiplier_handshake; wager cards multiply final score if expedition clears threshold”
- [2]push_your_luck_escalation— “ascending_sequence_lock; each new card commits deeper into the expedition, can't abandon”
- [2]partial_observability— “discard_pile_recycling; opponent's discards reveal their expedition strategy”
Archetype fits — v4 deep
How well this game shape maps to mobile archetype templates. Composite is a weighted sum of the 10 fit dimensions.
| Archetype | Composite | LTF | Session | Combo | Arc | Share | 5in | Onboard |
|---|---|---|---|---|---|---|---|---|
| Balatro ascending_sequence_lock + push_your_luck_escalation with -20 threshold is a sharp commit-or-fold engine, but no multiplicative joker-stacking — combo_scaling_depth at the Heat-tier mid. session_length_fit 9 (10-min play). Multiple polished iOS ports prove the loop translates. | 6.60 | 8.0 | 9 | 5 | 7 | 5 | 8 | 7 |
| Snap Two-player commit-card-or-discard timing has Snap-adjacent reveal tension and clean phone-native UI. session_length_fit 7 because 10-min is closer to Snap's match envelope than most. | 6.10 | 7.0 | 7 | 5 | 6 | 5 | 8 | 7 |
| Wordle Two-player commitment puzzle has tight win/lose moment and could produce a final-board emoji-grid (5-color completion summary). Sessions are 10-15min which exceeds Wordle's 60-90s but not catastrophically. | 5.90 | 7.0 | 7 | 3 | 7 | 6 | 8 | 7 |
| Cozy Open tier — head-to-head zero-sum race, push-your-luck punishment for failed expeditions caps loss_tolerance at 5. Tone is contemplative and there's no combat, but the -20 penalty bites and the win/lose framing isn't cozy. | 5.20 | 5.0 | 8 | 4 | 5 | 3 | 8 | 7 |
| Coop No shared fail-state — strict 2-player adversarial. The discard-pile-recycling does create indirect information signaling but not coop coordination shape. | 5.10 | 5.0 | 7 | 3 | 6 | 4 | 7 | 6 |
Translation candidate
Composite fit_score = bayes×0.30 + wish×0.18 + compress×0.17 + difficulty×0.20 + headroom×0.15.
Two-player card game with simple inputs; multiple polished ports exist on iOS, Android, Steam.
Rules card
Synthesized from sources below. Readiness: ready. Confidence: 0.94.
Readiness
ready (confidence=0.94, rules=0.90, fun=1.00). BGG rank: 344; year: 1999; weight: 1.47; playtime: 30 min
| Source | Quality | Role | Note |
|---|---|---|---|
llm_memory | 0.90 | draft synthesis | sonnet-self-rated-9 |
bga_tutorial | 0.85 | rules authority | BGA implementation rules summary |
bgg_comments | 0.75 | player voice | positive/player-voice sample |
github_code | 0.60 | implementation signal | GitHub implementation signal |
wikipedia | 0.55 | context/reception | bare-title |
Core Loop
Lost Cities is a two-player card game in which players build numbered expeditions in up to 5 color-coded archaeological sites. The deck contains 60 cards: 8 numbered cards (2–10) per color and 3 investment wager cards per color (5 colors total). On each turn, a player must play exactly 1 card to either their own expedition tableau (in ascending order within that color) or a shared discard pile for the corresponding color, then draw 1 card from either the main deck or the top of any color's discard pile.
Cards played to expeditions must be played in ascending order (e.g., 2, then 5, then 7 — you cannot play 4 after 7). Wager cards must be played before any numbered card in that expedition. Starting an expedition immediately costs 20 points; the sum of cards in the expedition is your gross, minus 20 is net. Three wager cards double (or triple) the entire expedition's final value (including the -20 penalty). Only expeditions you have started score; unstarted expeditions score nothing.
Turn Structure and State
- BGA tutorial is present; useful for exact turn flow and implementation gotchas.
- BGG description anchor: Lost Cities is a card game in the Kosmos two-player series. The game originally consisted of a single deck of cards of rank 2–10 in five different colors with three special "handshakes" ("HS" in scoring examples below) in each suit, but as of 2018 the game now includes six colored suits, with the sixth color being optional for gameplay. A game board is included to organize discarded cards and help players [...]
Win Condition and Arc
Most points after the deck is exhausted wins. You play until the last card is drawn from the main draw pile; the game ends immediately. VPs are summed from all expeditions (sum of cards minus 20 base cost, then multiplied by 1+number of waggers). Bonus: any expedition with 8 or more cards scores +20 bonus. The arc is a hand-management tension: too many expeditions spread you thin (all take the -20 penalty); too few miss scoring opportunities. The discard piles create a secondary market — drawing a discarded card is giving your opponent information about your hand.
Decision Primitives
BGG mechanisms: Hand Management, Melding and Splaying, Ordering, Push Your Luck, Score-and-Reset Game, Set Collection
Memory-derived primitives:
- Hand management (7-card hand, strictly ascending play within each color)
- Forced discard / discard pile interaction (every turn you must discard somewhere)
- Wager multiplier mechanic (double/triple the entire column score including the penalty)
- Investment threshold scoring (starting cost = -20; need 20+ to break even per expedition)
- Controlled information (discard piles are public; deck composition can be tracked)
v4 controlled primitives: hand_management_under_draw, push_your_luck_escalation, secret_objective_card, _other:ascending_sequence_lock, attrition_clock
Top iOS archetype fits: balatro 6.6, snap 6.1, wordle 5.9.
Why It Is Fun
Lost Cities is a masterpiece of minimal rules with maximal tension. Every wager card played is a commitment and a gamble — if the expedition scores poorly, the wager makes it much worse. The ascending constraint forces forward commitment; you can't take back a played card. Watching the deck dwindle creates endgame urgency, and the discard-pile draw option creates genuine card-reading cat-and-mouse.
Player-voice evidence:
- My son and I really like this game a lot. There's strategy, good length of game, that is variable due to ability to play different number of rounds, and it is easy to understand.
- This was the game that introduced me to Euro-style games. I loved it right away and still do, although I don't play it as much now. It's a great couples game, one of the few I can get my wife to play regularly. My mom even likes it. Very...
- Great 2-player game, and more interaction than its detractors give it credit for. Will be gard to find anyone who won't gladly play this, although I prefer more confrontational fare myself...
- Very simple and very enjoyable. The theme is a laugh, the board looks great when hanged on a wall. Can be played with a standard deck of cards and still be cool.
- A great two-person game. I've played with kids and adults. With children the theme of exploring and seeing how the artwork progresses is nice and mysterious. With adults its just quick and fun. My favorite Knizia.
Friction and Failure Modes
- Treat Sonnet-memory edge rules as draft until confirmed by manual, BGA, or transcript.
Translation and Design Hooks
- Use this card to ask: which primitive carries the fun if theme/licensing is removed?
- For iOS, look for short-session compression, clear state visualization, and a digital-only twist.
- For new tabletop design, look for the tension source and decide whether to preserve or invert it.
Edge Rules and Gotchas
- The "8 card bonus" (+20) applies to the full expedition (after multiplier); with 3 wagers, completing an 8-card expedition is worth massive points
- Wager cards played to an expedition count toward the 8-card threshold
- Drawing from a discard pile denies your opponent the card but tells them you wanted it (information leakage)
- Starting a poor expedition is sometimes forced — if your hand fills with low-value cards in a color you've started, you must continue it even knowing it'll score poorly
- Turn timing: the game ends the moment the main deck is exhausted (not at turn end), so careful card-count awareness affects when to start risky expeditions
- The 3 wager cards per color are entirely separate from the numbered cards — they must precede ANY numbered card of that color; you cannot play a wager after a number
Sources Used
[
{
"kind": "llm_memory",
"path": "data/llm_memory_sonnet/50.md",
"quality": 0.9,
"note": "sonnet-self-rated-9"
},
{
"kind": "bga_tutorial",
"path": "data/bga_tutorials/50.md",
"quality": 0.85,
"note": "BGA implementation rules summary"
},
{
"kind": "bgg_comments",
"path": "data/bgg_comments/50.txt",
"quality": 0.75,
"note": "positive/player-voice sample"
},
{
"kind": "github_code",
"path": "data/code_implementations/50.md",
"quality": 0.6,
"note": "GitHub implementation signal"
},
{
"kind": "wikipedia",
"path": "data/wikipedia/50.md",
"quality": 0.55,
"note": "bare-title"
}
]
Sources (5)
Inputs to rules-card synthesis. Click any pill with ↗ to open the original source.