What it is. The game has six kinds of money and the player can name what none of them does. Every new feature bolted on its own token, and the wallet became a HUD of orphaned counters.

Why it hurts. Currency is a promise: earn this, spend it there. When the map from earn → spend blurs, players hoard everything, spend nothing, and stop feeling the pull of a reward. A resource nobody can price is not a reward — it is clutter with a number on it. See system-economy for the sinks/faucets discipline this violates.

The smell

A resource whose spend the player can’t state without opening a menu. If “what’s this for?” needs a wiki, the currency has no job.

How it happens

• Feature-first minting. Each system ships its own currency for isolation — antipattern-feature-soup wearing an economy hat. Gems for the shop, shards for crafting, cores for upgrades, dust for rerolls. Nobody merged them because each shipped in a different sprint.
• Monetization layering. A soft currency, a hard currency, a battle-pass currency, an event currency, a limited-time currency — see system-live-ops-and-seasons and antipattern-pay-to-skip for how this metastasizes.
• Fear of one economy. Designers split currencies to “prevent” a dominant strategy, then never notice the split just moved the confusion into the wallet, not out of it.
• Reward-schedule padding. More counters look like more progression. It’s system-reward-schedules faked with noise instead of paced with signal.

The tell (check YOUR design)

• List every currency. Next to each, write one sentence: “You earn X by ___, you spend X on ___.” Any currency that needs two sentences or a “usually” is spaghetti.
• Two currencies share the same decision horizon (both spent in the same shop, same session, on the same axis of power). They are one currency wearing two coats. Merge.
• A currency’s balance only ever goes up — a faucet with no sink. It’s a score, not money; relabel it or cut it.
• Players hoard. Full wallets, empty carts. When nobody spends, the currency isn’t valued — it’s illegible.
• Your onboarding (system-onboarding) needs a dedicated screen just to explain the wallet. The tutorial length is the symptom; the wallet is the disease.

Twist seams (X but Y)

• A deckbuilder economy but one currency, three timescales (twist vector: reuse the same coin at run / shop / meta so the player learns it once). anchor-slay-the-spire runs gold in-run and one meta unlock track — not five wallets.
• A city-builder but resources ARE the puzzle, not the clutter (twist vector: every resource has a visible sink you can point at on the map). If a resource has no on-screen consumer, it’s spaghetti; if it does, it’s genre-city-builder design. See system-resource-loops.
• A roguelite but the premium currency is diegetic and singular (twist vector: one thing to bank between runs, spent on one board). anchor-hades banks a small, legible set with obvious spends — contrast a mobile port that would add four.

The fix

One currency per decision horizon. A horizon is a distinct question the player answers with money: what do I buy this fight, what do I unlock this run, what do I keep forever. Give each horizon exactly one currency; merge the rest into it.

Guardrails while you cut:

• Three is the ceiling for most games; two is healthy; one is often enough. Every currency past three must earn its slot by owning a horizon nothing else touches.
• Each currency needs a named faucet and a named sink the player can point to — the sinks/faucets model in system-economy and the loop discipline in system-resource-loops.
• Put the wallet where the spend is. Currency shown far from its shop reads as noise. This is system-inventory-and-ui and pattern-readability work, not an economy tweak.
• If a currency exists to stop a dominant build, fix the build with system-counter-systems or system-build-diversity — don’t tax the wallet to patch balance.
• A currency the player never chooses to spend is antipattern-fake-choice in ledger form. Real money buys a tradeoff; illegible money buys a shrug.

Seen in…

• anchor-slay-the-spire — the counter-example. Gold + one meta track; you always know what each buys. Its clarity is why the economy disappears and the choices stay foregrounded.
• anchor-hades — several currencies, but each has a loud, single, diegetic spend (Darkness → mirror, Keys → weapons/talents). Legible because roles don’t overlap.
• anchor-loop-hero — resource sprawl that flirts with the smell: many mats feed the camp, and early players genuinely can’t say which mat gates which building. Saved by a tight sink list, not by the count.
• Free-to-play RPGs (broadly) — the textbook offender: gold, gems, crystals, stamina, arena coins, guild coins, event tokens, shards per hero. The wallet becomes a spreadsheet; spending becomes a chore. This is where antipattern-pay-to-skip and system-live-ops-and-seasons compound the mess.
• anchor-stardew-valley — mostly one currency (gold) doing enormous work; the few special tokens (Qi gems, tickets) arrive late, isolated, and clearly labeled. Restraint is the design.

Verify / guard

Design-time gate: enumerate currencies, write the one-sentence earn→spend for each, and count distinct decision horizons. Currencies > horizons means merge before handoff. Then prove the sinks/faucets balance and that no currency is faucet-only — the economy checks in system-economy, the wallet-readability check in system-inventory-and-ui, and the run-through in docs/VERIFICATION.md.

Before you add currency N+1, answer in one line: which horizon does it own that no existing currency owns? If you can’t, you’re not adding money — you’re adding spaghetti.