← all deep dives

Bomb Busters

Deep Dive for an iOS Coop-Deduction Roguelite Translation

Project context: Translate the core loop of Bomb Busters (Hisashi Hayashi, Cocktail Games / Pegasus Spiele, 2024 — Spiel des Jahres 2025) into a single-player-or-coop iOS campaign roguelite. The "Dominion -> Slay the Spire" pattern, but for cooperative deduction. No nailed digital coop-deduction game exists on mobile: Hanabi has only stagnant browser/BGA implementations, Codenames Online is browser-and-async, The Crew lives on BGA but has no native mobile app. Bomb Busters is the freshest, most onboarding-friendly entry in the genre and ships with a 66-mission campaign that maps directly onto roguelite progression.

1. The mechanics, precisely

A turn from the bomb-tech's seat. Each player has a tile-stand with a hidden hand of numbered "wire" tiles drawn from a deck of 48 blue wires numbered 1-12 (four of each value), plus mission-specific yellows and reds. Critically: every player must sort their wires in ascending order from left to right, facing themselves. The values are hidden from teammates, but the position of each slot is public, and the sort rule means slot 3 is bounded above by slot 4 and below by slot 2. On your turn you point at a slot in front of any teammate (yourself or another) and call a number. If correct, the wire flips face-up and stays in the stand; when a face-up wire matches another face-up wire of the same value somewhere on the table, the matching pair is "cut" — that's how you score progress. If wrong, the detonator dial advances one space and an info token revealing the wire's true value is placed in front of it (so a wrong guess still leaks information). Run the detonator off the end of its track and the bomb goes off; cut all required pairs first and the team wins (rulebook PDF, Goblin Games how-to).

Wire colors are a difficulty ramp. Blue wires (1-12) are the bulk of the bomb. Yellow wires must be cut in a specific order — they enforce a sub-puzzle on top of the deduction. Red wires are sudden-death: a wrong guess on a red is instant mission failure, no detonator advance needed. The colors are not flavor; they are tiered risk gates that let the designer shape difficulty mission by mission (Merchants of Play guide, tabletopping.games review).

Communication is locked. You may not state the value of a wire in your stand, imply numbers, recall what was said last turn, or pre-announce a guess. You may discuss general tactics, equipment use, and special-rule reminders. The whole information channel is therefore: who points at what, the public face-up wires, the info tokens from wrong guesses, and a small set of equipment tokens (Wire Cutters, Scanners, Detection tools — used to peek, mark, or cut without risking the detonator) (BGG strategy thread, Tabletopping).

The campaign of 66 missions is the load-bearing structural feature. The box ships 66 numbered missions plus 5 sealed "surprise" boxes that introduce new wire types, twists, and rule modules over the course of a 30-50-session campaign. Missions 1-3 are training (reduced wire pool); the dial cranks until special wires, sabotage tiles, and constraints layer on top of each other. This is Pandemic Legacy-style scaffolding inside a 30-minute family game — and it is exactly the structure that mobile roguelite/campaign loops want (Cocktail Games product page, BGG entry, Spiel des Jahres page).

2. What makes Bomb Busters work as a tabletop game

The ascending-sort constraint is the elegance. Every guess narrows the possibility space for adjacent slots, so partial information cascades through the stand. Slot 4 was a 7? Slot 5 must be 7-12, slot 3 must be 1-7. This is positional deduction, not just card-content deduction, and it is what distinguishes Bomb Busters from Hanabi (where cards have no spatial relation) or The Crew (where the position of cards in hand carries no rule-bound information). The puzzle is half-revealed by the geometry before any words are spoken (Meeple Mountain review, Roll to Review).

The communication design is warmer than Hanabi. Hanabi's "you have a 4 / you have a yellow" formal hint protocol feels like data entry. Bomb Busters lets you talk about strategy freely while sealing only the literal wire values — and the color bands give players a soft vocabulary ("I'd be careful with anyone touching slot 3 right now") that maps onto natural human hedging. Reviewers consistently note that Bomb Busters feels less clinical and more connected than Hanabi despite enforcing similar information opacity (Shelf Gamer, Meeple Mountain, Meeple and the Moose).

Color bands are an onboarding ramp. New players guess by color band (yellow / blue / red / by gross value-tier); experts guess specific numbers. Same component, two skill ceilings — same trick Wingspan uses with bird abilities and Ticket to Ride uses with route length. It is the reason Bomb Busters can win Spiel des Jahres (a family-game award, voted by panel partly on accessibility) at weight 2.0 while still satisfying weight-3 deduction nerds (Spiel des Jahres announcement, BoardGameWire 2025 SdJ coverage).

The campaign fights the coop-deduction wall. Hanabi groups stagnate around session 10: optimal play converges, the convention metagame solidifies, and the social novelty drains out. Bomb Busters' campaign drips new constraints — sabotage tiles, double-ended wires, color rule inversions — every 3-5 missions for the first 60+ plays. The novelty curve is long enough to outlast the convergence curve, which is what Hanabi's "more cards" fan-expansions never achieved (Meeple Mountain, BGG quick-review thread).

Genre-mate comparison. The Crew is a coop trick-taker — same "limited communication" DNA but no positional deduction; players know their cards, the puzzle is in trick order. The Mind is silent, no deduction, pure rhythm. Hanabi shares the hidden-hand-deduction core but lacks the positional sort and the campaign shell. Bomb Busters' unique strength is the slot-bounded deduction + iconographic equipment + 66-mission scaffold combined (The Crew on BGA, Wikipedia: The Crew).

3. Existing digital attempts

Official digital Bomb Busters: none. As of April 2026 there is no Cocktail Games or Pegasus app, no Steam port, no announced licensee. Cocktail's broader digital footprint is essentially zero — they are a French paper-publisher with no app catalog (Cocktail Games). Board Game Arena does not list Bomb Busters in current catalog searches; given the SdJ win in July 2025 a BGA implementation is plausible-but-unconfirmed and would in any case be a browser port, not a mobile-native experience (BGA front page).

Hanabi on mobile — the cautionary tale. No first-party native iOS app. The community plays via BGA in-browser (BGA Hanabi) and a smattering of unofficial trackers like Hanabi Tracker on Google Play. The board-game world's most acclaimed coop-deduction title has never gotten a mainstream mobile port that retained players past month one (BGG iOS thread). The lane is empty because nobody has solved single-player coop-deduction.

The Crew is on BGA for online async play and has a Kosmos Helper App for rule animations — not a playable port. Codenames Online is browser-only, async-friendly, party-skewed; structurally a different genre. TTS mods for Bomb Busters likely exist (every BGG-top-100 game gets one within months) but require a desktop and four humans on Discord — not a mass-market loop.

Verdict: Bomb Busters-the-IP is digitally unclaimed, and the entire genre slot — single-player-or-async coop deduction with campaign progression on a phone — is unbuilt. This is one of the cleanest empty lanes in the BGG-top-100-to-mobile translation space.

4. The translation problem

The hardest part: communication is a constrained verbal medium. Around a table, "hmm I'd be careful with slot 4..." carries tone, eye-contact, and shared social context. On a phone with one human and three AI teammates, you have neither voice nor eyes — and forcing a multiplayer-only mobile game stacks adoption friction (find 3 friends, schedule, voice chat) on top of an already-niche genre.

What gets stripped (solo mode).

  • Free verbal hinting — gone. Replace with iconographic hint stickers that AI teammates place on slots: question mark (uncertain), exclamation (notable), thumbs-up (likely safe), warning triangle (do not touch), and a band-color chip (yellow/blue/red intuition). Five-icon vocabulary, no text, fully diegetic.
  • Reading other players' faces during your guess — gone. Replace with AI hesitation animations: the AI hand subtly shakes when it would warn you off a guess in progress.
  • Quarterbacking risk — Bomb Busters' tabletop design avoids the common coop quarterback failure (Meeple Mountain) because no one has perfect info. Solo digital must be careful: if AI teammates have full information, the player becomes a passenger. Solution: AI teammates simulate imperfect information (their own "hidden hand" is hidden even from the engine's hint planner).

What gets stripped (online mode).

  • Replace verbal hints with a structured chat-menu of phrases ("careful here," "I have info," "this slot is high") plus optional voice via Discord deeplink or in-app push-to-talk. Same five-icon sticker vocabulary as solo, available to humans too.

What gets ADDED for the roguelite v1.

  1. 66-mission campaign, ported nearly directly from the box, with cutscene framing per mission (3-line intel briefings, no voice acting required for v1).
  2. Procedural mission generator post-campaign — wire-pool size, color mix, equipment loadout, and special-tile combinations sampled from the campaign's mechanic library.
  3. Daily mission with a global success-rate leaderboard (everyone gets the same bomb).
  4. Rogue mode — a 7-mission run with permadeath: each failed mission costs you a teammate (down from 4 to 1), and the final mission must be solo. This is the Slay-the-Spire Ascension equivalent.
  5. Equipment as relics — mission-rewarded equipment becomes draftable between missions, mirroring relic drafts.

Recommendation: ship v1 with #1 + #4 + #3. Campaign is the spine, Rogue mode is the meta hook, Daily is the social surface. Procedural generator is overinvestment until campaign retention is measured.

5. Concrete iOS prototype spec

Target: iOS portrait orientation, single-developer Unity or SwiftUI+SpriteKit, 5-7 month MVP.

90-second core loop. Tap a teammate's slot, the number-pad slides up, tap a value (or a color-band shortcut), tap commit. Correct = green flash + the wire animates "cut" with a wire-snip SFX; if it pairs with another face-up wire, both fly off into a "defused" tray with a satisfying clack. Wrong = red flash, the detonator dial ticks one space (audible bomb-clock SFX), and an info token slides in front of the wire revealing its value. Repeat until the bomb is fully defused or the dial hits zero. A mission is ~12-25 guesses, 3-5 minutes.

Single-screen layout (phone portrait, 6.1" reference).

  • Top 35%: Teammate stands (1-3 AI or remote players, smaller card-row indicators with face-down wires shown as tile-backs, face-up wires shown as colored numerals; equipment slots underneath each).
  • Middle 15%: Detonator dial (analog dial visualization, ticks left as you fail), mission objective panel (pairs cut / pairs needed), special-rule banner for current mission.
  • Bottom 50%: Your stand (tap-to-reveal-to-self, hidden from screen-recording by default), action panel with number-pad and color-band shortcuts when a slot is selected, equipment row (tap to use a Scanner, Cutter, etc.).
  • Long-press a teammate slot to view all info tokens / face-up history for that slot.

Hint sticker UI. When AI teammates "would speak," they instead drop an iconographic sticker on a slot for 1 turn. Five icons: ? (unsure), ! (notable), thumbs-up (safe-ish), warning triangle (danger), color chip (band-tier intuition). Player can also place stickers to request info from teammates (which the AI then resolves on its next turn). This is the load-bearing UX innovation; paper-prototype this before any code.

Mission system.

  • 66-mission campaign with 3-line intel briefings and unlock gating (mission N+1 requires beating mission N, with a "skip after 3 fails" mercy option).
  • Procedural mode unlocked at mission 30, expanding at mission 66.
  • Daily mission, global pass/fail leaderboard with median-detonator-remaining as tiebreaker.
  • Rogue mode unlocked at mission 20.

Solo vs. multiplayer.

  • Solo with 1-3 AI teammates is the default mode and the primary design target.
  • Online async multiplayer (1-4 humans) added in v1.1 — turn-based with push notifications, like Words With Friends. The async-friendly nature of "your turn = one guess" makes Bomb Busters unusually suited to async coop.
  • Real-time multiplayer with Discord voice integration in v1.2.

Monetization. Premium one-time purchase, $9.99 base including the 66-mission campaign, all default equipment, solo + async multiplayer, daily mission. $4.99 expansion mission packs (themed: e.g. "Sabotage", "Double-Ended Wires", "Time Pressure") — 20 missions each, releasing every ~3 months. Avoid energy mechanics; avoid cosmetic-gacha for a deduction game (ruins the serious-puzzle tone).

Aesthetic direction. Cocktail Games' physical art is bright, cartoony, slightly retro-pulp (product page). The iOS version should lean slightly more grounded — think Mini Metro readability married to a Papers, Please tactile-grit feel. Wire tiles as chunky 3D plastic pieces with satisfying snip animations; detonator as an analog dial with hand-drawn imperfection. Audio: tactile clicks, wire-snip foley, low-bpm ticking-clock pad that picks up tempo as the dial advances.

6. Risks and unknowns

Top design risks.

  1. AI-teammate information asymmetry collapse. If AI teammates plan with perfect knowledge of the bomb state, the player becomes a button-presser executing AI suggestions. Mitigation: every AI teammate runs a constrained-deduction solver that only knows what a human in that seat would know (their own wires + face-up + info tokens). Hint stickers come from that bounded inference, not from omniscience.
  2. Mission-design throughput. 66 unique campaign missions is a lot of bespoke design. Mitigation: port the box's 66 missions verbatim for v1 (Cocktail's design has already been balanced through publication and SdJ jury review); save procedural and original missions for the expansion packs.
  3. Procedural ceiling past mission 66. Random combinations of campaign mechanics may produce trivially-hard or trivially-easy bombs. Mitigation: constraint-solver validator that estimates win-rate at the player's skill bracket; reject bombs outside a 30-65% target band.
  4. Color bands feeling babyish on a phone. Yellow/blue/red is great onboarding but reads as a kids' app on the App Store screenshot wall. Mitigation: aesthetic must lean grounded-pulp (Papers Please, not Candy Crush); a tense framing trailer; consider muted-saturation palette with the bright primaries reserved for the moment of the cut.
  5. Async multiplayer pacing. A 25-guess mission across 4 async players over 2 days may lose narrative tension. Mitigation: a 24-hour soft turn timer with a "rotate to AI" fallback that keeps the mission alive but flags it as ineligible for leaderboard.

What needs validation, in order.

  1. Tabletop solo with AI proxy. Play Bomb Busters solo using a hand-rolled "AI teammate" (yourself running 3 stands with separate notebooks of inferences) to verify the deduction puzzle survives without genuine social communication. If it doesn't survive on a table, it won't on a phone.
  2. Paper prototype of the 5-icon hint vocabulary. Index cards on a table, two players, one allowed only to place stickers, see whether the puzzle resolves at a similar rate to verbal play.
  3. Greybox digital prototype of one mission, one AI teammate, no meta-layer, on a real phone in a real hand. Validate that the 90-second core loop reads on a 6.1" screen and that the hint stickers don't crowd the slots.

Three questions for the prototype builder.

  1. How serious is your AI-deduction-engine appetite? Bounded-information AI teammates that play like a thoughtful human (not a solver) is genuinely hard — it's a partially-observable Markov decision process with a teammate-modeling layer. If you don't want to write that, the project halves in scope and becomes a glorified bomb-defusal puzzle. Pick a side now.
  2. Solo-first or multiplayer-first? A solo-first product can ship in 6 months and hits the empty-mobile-coop-deduction lane immediately. A multiplayer-first product needs servers, matchmaking, social/friend systems — 12-15 months minimum and a very different financial bet. Solo-first is the recommendation.
  3. Will you cleave to Cocktail's communication restrictions, or relax them for solo? The tabletop forbids stating numbers; in solo with AI teammates, that restriction is artificial (the AI can't "talk" anyway). One could let solo-mode AI teammates literally write a text hint like "slot 3 might be 7-9." Heretical to the design, possibly necessary for clarity. How much fidelity is sacred?

7. References