← all deep dives

Heat: Pedal to the Metal

Deep Dive for an iOS Roguelite Translation

Project context: Translate the core loop of Heat: Pedal to the Metal (Asger Harding Granerud & Daniel Skjold Pedersen, Days of Wonder, 2022) into a single-player iOS roguelite. The "Dominion -> Slay the Spire" pattern, but for a racing deckbuilder. No direct digital competitor exists.

1. The mechanics, precisely

A turn from the driver's seat. You hold a hand of 7 cards drawn from your personal deck. First you choose your gear (1-4); you can shift one step up or down for free, or two steps if you pay 1 Heat per extra step. Your gear is your play count: gear 2 means play exactly 2 cards, gear 4 means play exactly 4. You play the cards face down (in multiplayer, simultaneous with everyone else), reveal, sum the speed values, and that is how many spaces your car moves this turn. Before moving, you may boost by spending 1 Heat to flip the top card of your deck and add its value (a mini push-your-luck — could be a 1, could be a Stress card with a random 0-4 outcome). Then you cross any corner lines: if your speed exceeds a corner's speed limit, you must pay Heat equal to the overage from your engine reserve to your discard. If you can't pay enough, you spin out — your car snaps back to before the corner and you start next turn in 1st gear. Finally you cool down (gears 1 and 2 let you move 1-2 Heat cards from hand back to the engine reserve), discard cards you don't want, and draw back up to 7. End of turn.

The heat economy. Heat is a closed-loop pollution currency. You start with 6 Heat cards face-up on your "engine" reserve, outside the deck. Every time you pay Heat (boosting, shifting two gears, exceeding a corner limit), a Heat card moves to your discard pile, where it will be reshuffled into your draw deck. Drawing it later is dead weight — Heat cards have no playable value and cannot be discarded; they sit in your hand clogging it. The only purges are: (a) Cooldown — being in 1st/2nd gear (or last place adrenaline bonus) lets you return Heat from hand to the engine reserve, and (b) certain upgrade cards. Overheating happens when you can't fill your gear's required play count with playable cards — you must dump Heat cards face-down to make up the count, your car does not move at all, and you immediately drop to 1st gear. So Heat is currency and deck pollution and tempo punishment, all in the same token.

Why the gear-shift one-step rule matters. Gears are sticky. You can't dump from 4th to 1st to dodge a corner — you'd pay 4 Heat in shifts, then probably overheat trying to play only 1 card from a hand built for 4. This forces cornering planning two to three turns ahead: you must downshift before the apex while your hand still has low cards. The whole strategic lattice of the game lives in this constraint — it converts the racetrack into a rhythmic deceleration puzzle and prevents the deckbuilder from collapsing into "always play the highest card."

Corner check & cooldown. The corner check is the push-your-luck heartbeat: each corner has a numeric speed limit, and going over by N costs N Heat per corner crossed. Hit two corners on one big turn and you pay both. Cooldown is the release valve — sandbag in 1st/2nd for a turn to flush Heat back to your engine, or grab the Adrenaline bonus the last-place car gets every round (extra cooldown, free boost, plus catch-up movement).

Sources: Rulebook PDF, BGA gamehelp, RulesPal.

2. What makes Heat work as a tabletop game

The "Sensation" aesthetic. Granerud's stated design intent is captured perfectly: "We are not making Heat the racing simulation game, we are making Heat the Hollywood movie." The dopamine moment is the late braking fantasy — you reveal a hand of three 4s while the leader played safe 2s, you scream through the chicane spending 3 Heat, and you nose ahead by half a length. The simultaneous reveal makes that moment a stunt rather than a calculation. Heat does not simulate a racing line; it simulates the feeling of nerve.

Simultaneous reveal tension. Everyone plans behind their player screen, then flips together. The drama comes from incomplete information: you saw the leader downshift to 2nd — did they sandbag a juicy hand for the next corner, or are they bleeding Heat? You commit before you know. This compresses 60 minutes of physical race into ~12 rounds of mutual second-guessing. Removing this is the central translation problem (see §4).

Heat as deck pollution — beloved because it's visible and self-inflicted. Compare to Quacks of Quedlinburg: in Quacks you draw from a bag and the bomb chips arrive randomly via purchases; the pollution is gradual and partially out of your control. In Heat, every single Heat card in your discard exists because you chose to spend it for a specific competitive gain — boost, shift, corner overage. That direct trace from "I pushed it" to "now my hand is choking" is a tighter feedback loop than Quacks' bag-build, and tighter than Trains' track-builder pollution (which is two layers removed from in-the-moment decisions). Granerud noted in interviews that the heat system went through "numerous iterations" precisely because "the more dilemmas you have in a decision space, the more regret you can have" — regret is the design target.

Designer commentary worth carrying forward. Granerud and Pedersen explicitly described Heat as the illusion of an F1 race, identified corners + straights as the only two structural primitives that mattered, and built upgrade variety so that "you couldn't go faster through a corner [otherwise] it would be a limited game space." The Legends solo automa was designed last and is widely praised — it provides a working template for how the AI in the iOS prototype could behave (one shared card per round driving multiple bots, with state-dependent movement based on distance to next corner).

Sources: Talking Shelf Space interview pt 1, BoardGameWire interview, Meeple Mountain review, BGG "Quacks comparison" thread.

3. Existing digital attempts

Official digital Heat: none. Days of Wonder ported Ticket to Ride (huge success on iOS/Steam) and Small World, but as of April 2026 there is no announced standalone Heat app. Online play is confined to Board Game Arena (browser, asynchronous-friendly turn-based, requires premium for some games) and a community Tabletop Simulator mod on Steam. Both are faithful ports of the multiplayer board game — neither attempts a single-player roguelite reframing.

Racing roguelite deckbuilders on Steam. The closest analog is Deck RX: The Deckbuilding Racing Game (still in "coming soon" / demo as of search date — no shipped reviews yet). Deck RX uses cards as discrete actions (accelerate, brake, drift, attack) with a 150+ card pool, six themed worlds, and a vehicular-combat second win condition. It is PC-only, has no Heat-style gear/heat economy, and leans on car-combat — closer to Roguebook meets Twisted Metal than to Heat's pure racing-rhythm feel. The Loopler (demo) is a roguelike racer but not card-driven. Wild City and StarVaders are deckbuilders without racing. The roguelite-deckbuilder genre is hyper-saturated (StarVaders, Balatro et al.) but the racing corner of it is essentially empty.

Mobile racing games. Universally physics/arcade — Real Racing, Asphalt, CSR, Mario Kart Tour, Drive Ahead. None are deckbuilders or hand-management. The sole hand-managed driving mobile game I can find evidence of is Road Rage card games and various non-roguelite ports of Mille Bornes (the 1954 hazard/remedy card game). There is a clear vacant lane: a tactical, turn-based, deckbuilder racing roguelite on iOS does not exist.

Verdict: Heat-the-IP is digitally unclaimed at the single-player level. The genre slot is also vacant on mobile. This is a real opportunity, not a saturated market.

Sources: Days of Wonder Heat page, BGA Heat, TTS mod, Deck RX on Steam.

4. The translation problem

What gets stripped.

  • Simultaneous reveal is the obvious casualty — there are no other humans to surprise. Replace it with opponent-first telegraphing: AI rivals declare gear visibly at the start of the round (you see the icon over their car), but their card values stay hidden until you commit your gear. This preserves the "commit-before-knowing" tension while letting the player optimize against partial public information.
  • Player-screen secrecy and physical card flipping — replace with rhythmic single-tap reveals once you commit; the AI's reveal animates in the next instant, mimicking the multiplayer reveal moment.
  • Slowdown/sandbag table-talk metagame (where humans bluff position) — gone, no replacement needed.

What gets added (the digital meta-layer — pick 2-3 of these for v1).

  1. Roguelite season run. A "season" is a 5-race campaign on procedurally chained tracks. After each race you draft 1 of 3 upgrade cards (tiered T1/T2/T3) and 1 of 2 garage relics (passive modifiers — e.g. "first corner of every race ignores +1 speed", "Stress cards now flip 0-3 instead of 0-4"). A run ends when you finish 5 races or are eliminated by finishing last twice.
  2. Career mode (long meta-loop). Persistent driver with XP, sponsors that grant stable perks, and a championship across multiple seasons. Slower burn for retention.
  3. Daily challenge. Fixed track + fixed opponents + fixed starting deck + a constraint (e.g. "you start with 4 Heat in deck"). Global leaderboard by finish time / position. This is the streamer/Twitch surface.
  4. Workshop / Garage. Let players modify a base car between runs: choose engine archetype (more boost, less heat-payoff; or vice versa) and tire profile (better cornering, worse straights). This is the "build identity" layer.
  5. Crew chief radio. Diegetic tutorial/AI rival barbs delivered as voiced one-liners to give the Hollywood-movie aesthetic Granerud talked about.

Recommendation: ship v1 with #1 + #4 + #3. Roguelite run is the spine, Garage is the build identity, Daily is the social hook. Career mode is overinvestment for a solo dev prototype; add it post-launch if retention demands it.

What cannot translate cleanly.

  • Asynchronous trash talk and the table moment when someone overheats in front of you — can be partially recovered through animated AI personalities and post-race replay clips, but the social texture is gone.
  • Negotiated drafting and slipstream micro-tactics (the slipstream rule rewards being directly behind another car) — works fine mechanically, but loses the "dad vs kid" dynamic. The rivals will need legible personalities (the cautious one, the aggressor, the wildcard) to make slipstreaming feel like more than a number on a screen.
  • The physical hand — 7 actual cards spread in your fingers — has tactile information density that 7 thumbnails on a phone cannot match. Mitigation: large, swipeable cards with strong silhouettes.

5. Concrete iOS prototype spec

Target: iOS portrait orientation, single-developer Unity or SwiftUI+SpriteKit, 4-6 month MVP.

60-second core loop. Tap a gear (1-4 buttons along the bottom). Your hand fans up; tap N cards (N = gear), they slide to a play queue. Tap commit. Opponents' chosen gears already showed earlier in the round; now their cards reveal. Camera pans along the track as the cars animate forward step-by-step, corner checks pop a "+2 Heat" floating chip if you over-speed. End-of-turn, tap to discard unwanted cards, deck auto-draws to 7, next round starts. A race is ~10-14 such rounds, ~3-4 minutes total.

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

  • Top 35%: 2.5D top-down/isometric track segment, showing your car + nearest 2 rivals; speedometer + gear indicator overlaid on your car.
  • Middle 15%: Engine status bar — visible Heat reserve (6 slots, filled/empty), current gear, next-corner speed limit warning icon when within 2 spaces.
  • Bottom 50%: Hand of cards (fan of 7, large enough for thumb-tap), gear selector (4 chunky buttons in a row), play/commit button.
  • Drag-up a card to inspect; drag-up the engine to see deck composition mid-race (key for purge planning).

Card system.

  • Starter deck: 12 speed cards (3 each of values 1-4), 3 Stress cards (random 0-4 with side effects), 3 starting upgrades. Mirror the board game exactly — proven balance.
  • 6 Heat cards in the engine reserve (out of deck initially).
  • Gear-shift UI: tap a gear button; if it's >1 step from current, a Heat-cost preview appears next to it ("Shift to 4 — costs 1 Heat"). Confirm with a second tap.
  • Heat card visualization: deep red border, no value number, a thermometer icon. When drawn into hand, sits at the rightmost slot and pulses subtly to make its drag on your tempo felt.

Run structure.

  • 5 races per run, each on a procedurally generated track (3 corner archetypes: hairpin / sweeper / chicane; tracks composed of ~3-5 corners + straights).
  • 3 AI rivals per race, drawn from a roster of ~8 personalities with distinct play biases.
  • Race win condition: finish 1st = full reward, 2nd-3rd = partial, 4th = lose a "tire" (3 tires = run over).
  • Run win condition: finish 5th race in podium position. Final race is a championship track with a named boss rival.

Meta-progression.

  • Between races: pick 1 of 3 upgrade cards (replaces a starter card or adds to deck) + 1 of 2 garage relics.
  • Between runs: unlock new cars (different starting deck profiles), new tracks (added to the procedural pool), new rival personalities, new relic pool entries. About 30-50 unlockables for a ~20-30 hour completion arc.

Monetization. Premium one-time purchase, $6.99-$9.99. F2P with cosmetic-only cars + paid expansion seasons is a viable alt, but avoid energy mechanics — they directly fight the run structure (a run is a session; gating it kills the loop). The game's natural ceiling is "Slay the Spire mobile" pricing, not "Asphalt" pricing.

Aesthetic direction. Lean into a stylized 80s-meets-modern-flat look — think Outrun palette (sunset oranges, electric blues, white track lines) but with the readable flat-vector silhouettes of Mini Metro. The physical Heat board is illustrated, semi-cartoony, period-1960s; the iOS version should reference that period-romance without copying the lithograph style. Diegetic UI sounds: tire squeals on corner over-speed, engine pitch shifts on gear changes. Music: synthwave, low-key during planning, swells on commit/reveal.

6. Risks and unknowns

Top design risks.

  1. Tension collapse without simultaneous reveal. AI rivals' commits are scripted — the player may feel they are puzzle-solving rather than racing. Mitigation: opponent-first gear telegraphing, animated "reveal" beat with sound, and AI personalities that visibly hesitate or aggress per archetype.
  2. Heat economy may feel punitive in a single-player vacuum. In multiplayer, overheating is hilarious because someone else benefits; solo, it just feels like you lost. Mitigation: generous adrenaline bonus when behind, visible "this was a great recovery" feedback, and tune the corner-overage costs gentler than the board game.
  3. Procedural tracks may produce trivially-easy or impossible layouts. Heat tracks are hand-designed; randomization risks breaking the strategic two-corners-ahead planning. Mitigation: procedural with constraint solver — guarantee minimum straight length between corners, validate a "Legends bot can finish in N rounds" check.
  4. Roguelite drafts may produce dominant strategies. Heat upgrades were balanced for hand-built decks; a draft pool can compound. Mitigation: tier rare upgrades, cap relic stacking.
  5. Mobile attention span vs. 4-min races. A race is longer than a Slay the Spire combat. Mitigation: fast-forward toggle on AI animations, optional "instant resolve" for known-safe rounds.

What needs validation, in order.

  1. Tabletop solo playthrough first. Play Heat solo with the Legends automa for 5-10 races before writing any code. Confirm the feeling survives single-player at all; if it doesn't survive on a table, it won't on a phone.
  2. Paper prototype of the iOS reveal flow. Mock the opponent-first-gear telegraph rule with index cards; check whether it preserves commit-tension or just feels like reading enemy intent.
  3. Greybox digital prototype of one race, one AI rival, no meta-layer. Validate that core-loop pacing holds at 3-4 min per race on a phone screen. Only then build the roguelite layer.

Three questions for the prototype builder.

  1. What is your appetite for AI work? A great Heat AI is the difference between "puzzle" and "race." If you don't want to build behavior trees / personality scripting, the project halves in scope but loses its biggest differentiator. Pick a side now.
  2. Solo premium or live-service ambition? A premium $7 game can ship in 6 months and make $30-100k in year 1. A live-ops game with weekly tracks and seasons is 18 months and a different financial bet. The roguelite spec above fits premium; pivoting later is painful.
  3. Are you willing to deviate from the board game's heat-as-deck-pollution to soften single-player pain? E.g. "Heat cards can be discarded for half-purge in 1st gear" — heretical to fans, but maybe necessary for iOS feel. How much fidelity is sacred?

7. References