Tyrants of the Underdark

Deep Dive for an iOS Roguelite Translation

Project context: Translate the core loop of Tyrants of the Underdark (Andrew Veen, Peter Lee, Rodney Thompson; Gale Force Nine, 2016; Aberrations & Undead expansion 2017; Updated Edition / Refreshed re-release with the Drow & Demons / Elemental Evil & Drow Houses half-decks) into a single-player iOS roguelite. This is the "Dominion -> Slay the Spire" pattern, but applied to a hybrid deckbuilder + 4X strategy game. No direct digital competitor exists at any platform. The genre slot — deckbuilder fused with persistent territory control — is, as of April 2026, digitally vacant.

1. The mechanics, precisely

A turn from the matron mother's seat. You start each turn with a hand of 5 cards drawn from your personal deck. You play the entire hand at once, summing two resources printed on every card: Influence (the "buy" resource) and Power (the "fight" resource). You then spend those pools, in either order, in any combination, on a menu of seven actions: (a) Recruit a card from the market — six face-up slots backed by a 40-card market deck — paying its Influence cost into your discard; (b) Deploy a troop from your supply onto a site you already control or that connects to one you control, costing 1 Power per troop; (c) Move a troop along an edge for 1 Power; (d) Assassinate an opponent's troop on a site you control, costing 2 Power, sending it to your Trophy Hall (worth endgame VP); (e) Promote any single card from your hand or discard, removing it from your deck permanently and scoring its inner-circle VP value at game end; (f) trigger a card's Aspect ability (some require Influence/Power, some are free); (g) place a noble token to claim full control of a site. Then you discard everything, draw five fresh, and pass.

The dual-resource card. Every card in the 40-card market — and every starter — prints two numbers: an Influence value (top-left) and a Power value (top-right), plus an aspect icon (Warrior, Wizard, Priestess, Demon, Aberration, Undead, depending on which two half-decks were paired) and a unique text effect. A typical card might give 2 Influence + 1 Power + "draw a card." A combat-heavy card might be 0 Influence + 3 Power + "assassinate one troop on a connected site." This is the structural genius: every card is partial currency in both economies, so every recruit decision is simultaneously an "engine" decision and a "war" decision. There is no clean separation between deckbuilder math and area-control math; the card is the seam.

The Underdark map. A graph of named sites (Menzoberranzan, Blingdenstone, Sshamath, Araumycos, Gracklstugh, etc.) connected by edges. Each site has a numeric VP value (3-8) and is either a normal site or a city (cities also grant a VP token when first fully controlled). Each player has 10 troops + 3 noble tokens in supply; "control" means more troops than any other player on a site, "total control" means at least one troop and no enemy troops. Adjacency drives deployment and assassination range.

Promote — the signature. Promote permanently removes a card from your circulation, scoring its printed inner-circle number (typically 1-4 VP). It is the deckbuilder's classic "trash" mechanic re-skinned as in-fiction loyalty: the minion has been "promoted to your inner circle" and no longer participates in the bloody work below. Crucially, promote and assassinate are the two parallel VP tracks — promote is the lawful/political path, assassinate-and-trophy is the chaotic/violent path — and the best decks weave both. Most cards score 1 VP if left in the deck at game end, 2-4 VP if promoted; the differential is the entire reason to thin.

Endgame triggers. The game ends the turn after one of: (1) any single half-deck pile in the market is exhausted; (2) a player places their last troop from supply; (3) the market deck runs out entirely. (1) is by far the most common — typical games end in 8-12 turns. Score = sites controlled (face VP) + city VP tokens + cards in deck (1 each) + cards promoted (printed inner-circle value) + trophies (2 VP per assassinated troop) + bonuses for total-control bands.

The four (now six) half-decks. Each game uses two 40-card half-decks, shuffled together to form the market. The base game ships with Drow, Houses, Demons, and Elemental Evil; the Aberrations & Undead expansion adds two more; the community Hazards mod adds a seventh. Aberrations force opponent discards; Undead use a "devour" sacrifice mechanic; Demons offer raw aggression; Elemental Evil offers card-draw engines; Drow are balanced; Houses are politicking. Pairing changes the whole strategic shape of the game — Drow + Demons is a knife fight, Drow + Elemental Evil is an engine race, Aberrations + Undead is a self-mill nightmare. Six decks yield 15 unique pairings; with the Hazards fan deck, 21.

Sources: Rulebook PDF, Esoteric Order of Gamers reference, Topping the Table how-to-play, BGG entry.

2. What makes Tyrants work as a tabletop game

The dual-resource card is the most elegant decision-compression in modern deckbuilding. Compare to Star Realms, where each card is either Trade (buy) or Combat (attack) and the choice is mostly about deckbalance ratios. Tyrants prints both numbers on every card and forces the player to split a single hand between two competing economies that share no fungibility. You cannot "convert" Power to Influence. The result: a 3-Influence/0-Power card is a builder turn, a 0-Influence/3-Power card is a war turn, but the typical 1/1 or 2/1 card demands a tradeoff at the moment of play — do I push my engine or push my front line? Veen and Thompson didn't invent dual currencies, but they're the cleanest implementation in the genre. (Justin Gary's interview with Thompson on Think Like A Game Designer episode 26 is the closest we have to designer commentary; Thompson notes the team specifically wanted "every card to mean something on every turn," which is the dual-resource constraint stated as a design goal.)

Promote is the genius mechanic, and it's genius for one specific reason: it converts trashing from a chore into a payoff. In Dominion, trashing a Copper feels like cleaning your room — necessary maintenance, not exciting. In Tyrants, trashing a card scores points. The frustration of starter-card bloat — the dominant complaint in 90% of deckbuilders — is converted into the primary VP engine. As one reviewer put it: "Promotion is a hugely interesting mechanic that turns the normal flow of a deck builder on its head." This single design move means every deckbuilding decision is also a scoring decision, eliminating the engine/race tension that plagues less-tight deckbuilders (Legendary, Marvel Champions, etc.) where you build a great deck and then run out of game before it pays off.

Modular half-decks: infinite variety, tight balance. The two-half-deck combinatorial design is the same trick Magic uses with two-color pairs and Marvel Champions uses with hero+villain pairings, but Tyrants nails the granularity — 40 cards is enough to feel like a coherent strategy archetype without being so large that pairings feel diluted. The base box's six pairings each play meaningfully differently; Aberrations & Undead pushes that to fifteen. This is the replayability moat. (Reviewers consistently flag the second edition / Updated refresh as having fixed the printing issues that plagued A&U — card-back color mismatch, glossier stock — that made the original expansion shuffle-detectable.)

Spite vs. economy — the harassment dilemma. The most-discussed strategic question on BGG: when you sum your hand and have 4 Power, do you spend it deploying onto a contested site (pure economy — more VP, more board presence) or do you assassinate the leader's troop in your trophy hall (pure spite — they lose 1 VP from their site, you gain 2 VP in trophies, and they lose tempo redeploying)? This is the engine-vs-interaction tension that pure deckbuilders (Dominion) lack and pure area-control games (Risk) overdose on. Tyrants threads it because the cost of spite is denominated in the same Power pool as construction, so harassment is never free — you're trading your build for theirs. The game's emergent player politics — table-talk truces, mutual-deterrence pacts, retaliatory assassination spirals — all flow from this single shared currency.

Why it sits at BGG rank 223 with weight 2.54. It is heavier than Dominion (2.36) and lighter than Twilight Imperium (4.32). It plays in 60-90 minutes and supports 2-4. It demands strategic depth without rules-lawyer overhead. The community consensus: it is the best deckbuilder + area-control hybrid ever published, and reviewers like Big Boss Battle and Roll to Review routinely call it a "masterpiece."

Sources: There Will Be Games — King in the Shadows deep dive, Player Elimination — 1,941 Days Later reflection, Think Like A Game Designer #26 — Rodney Thompson, Board Game Quest — Second Edition review.

3. Existing digital attempts

Official digital Tyrants: none. Gale Force Nine is a physical components and miniatures publisher first; their digital footprint is essentially zero. They have no app studio, no Steam catalog, no announced port. Wizards of the Coast (the D&D licensor) has digital irons in the fire — D&D Beyond, Baldur's Gate 3 (Larian), Solasta, and the cancelled-then-resurrected MMO conversations — but no D&D-branded deckbuilder of any kind has ever shipped on mobile or PC. The IP is sitting on the shelf.

Community digital play. The active surface is Tabletop Simulator, where there are at least four major scripted mods, including a fully-automated "Tyrants of the Underdark (+exp) [scripted+modified]" workshop entry with deck-selection UI, market auto-refill, and end-of-turn cleanup. There's also an unofficial Hazards Expansion fan deck. There is no Tabletopia listing, no Board Game Arena port, no mobile companion app — the contrast with Star Realms (which shipped on iOS in 2014 and has run continuously) is stark.

Hybrid deckbuilder + area-control digital, the genre survey.

Star Realms (digital app, iOS/Android/Steam) — pure combat deckbuilder, no map, no territory. Closest in spirit but missing the entire 4X half. Excellent reference for deck-only mobile UX.
Marvel Snap — three "locations" with modifiers, simultaneous play, six turns. Lane-control, not territory-control; the "map" is a 1×3 grid. Brilliant cardplay design (see Ben Brode's GDC talk) but the spatial element is intentionally tiny — a locations affix system, not a map.
Inscryption — single-player roguelike deckbuilder with horror narrative; sacrifice-based card-play, no map. Reference for atmospheric campaign framing of a deckbuilder.
Dune: Imperium (Direwolf, 2023) — official digital port of a deckbuilder + worker placement hybrid. Workers go to spots on a map; agents claim influence with factions; some territorial element via Conflict cards but not persistent zone control. The closest published precedent — and tellingly, it sold well on Steam and iOS, validating that hybrid deckbuilders can land digitally.
Mahokenshi — samurai deckbuilder with hex-tile movement on small maps; closer to a tactics game than a 4X. The map is a board for combat, not a contested territory.
Total War: Three Kingdoms — Mandate of Heaven — the "deck of cards" hook in the marketing turned out to refer to unit cards, not a deckbuilding mechanic. False lead.
Nowhere Prophet — deckbuilder with overworld map but the map is a roguelike node graph, not contested territory.

The verdict. A turn-based, deckbuilder + persistent-territory-control digital game does not exist. Dune: Imperium is the closest published reference and it's worker-placement, not territory-control. Tyrants' specific mechanical signature — Power as a shared currency for both troop deployment AND inter-player aggression on a connected map — is genuinely unclaimed. Combine that with the unclaimed D&D Forgotten Realms / Drow IP angle and the opportunity is real, not imagined.

Sources: TTS scripted mod, Star Realms digital, Marvel Snap GDC talk, Inscryption on Steam, Mahokenshi.

4. The translation problem

What's structurally easy. The map's adjacency + control logic is trivial code — a graph with sites, edges, and per-site troop counts. Card resolution is straightforward state mutation. The market's six-slot face-up + draw-pile model is one of the most common deckbuilder UX patterns in the world (see Star Realms, Hearthstone Mercenaries, Slay the Spire shop). The promote action is one button.

What's structurally hard: the dual-resource UI. Every card prints two numbers, and every turn the player must mentally sum both pools and partition each across multiple actions. On a tabletop, you stare at your hand, count Influence, count Power, plan. On a phone, that act of summing-while-planning needs to be surfaced or it collapses into illegibility. The right answer: two persistent pool counters at the top of the screen — a purple Influence orb and a red Power flame — that update live as the player taps cards into the play zone. Each card displays its two values as colored split-icons (purple wing, red sword) on the card frame, not buried in text. When the player drags Influence onto a market card, the orb depletes; when they drag Power onto a map site, the flame depletes. Make the math visible.

Promote needs a slick gesture. On the table it's a satisfying physical motion — pluck a card from your hand or discard, slide it to your inner circle pile. Digitally: a long-press on any card in hand or discard reveals a "Promote" action. The card animates upward off the playfield into a glowing inner-circle tray at the top-right, with a particle effect and a number popping up showing accumulated promote VP. The tray must be always visible — minimized to a number-badge — because forgetting promote is the single most common new-player error and the game collapses without it.

Map representation. The Underdark map is irregular and graph-shaped, not a hex grid. A 2D top-down isometric view with stylized cavern art works better than a flat hex map — it preserves the Forgotten Realms cartographic vibe. Sites are clickable nodes; edges are drawn as glowing tunnels. Troops are stacked icons on each site with a count badge. A second tap on a site opens a detail tray showing controllers, VP value, and adjacent sites highlighted.

What gets ADDED for the iOS roguelite/campaign meta-layer.

Underdark Conquest — a 7-9 map campaign with escalating boss houses. Each map randomizes the half-deck pairing (e.g. Map 3 is Drow + Demons; Map 4 is Aberrations + Elemental Evil) so the player sees most pairings across a run.
Persistent house upgrades — between maps, draft 1 of 3 House Perks (e.g. "+1 starting troop," "first promote each game gives +1 VP," "your noble tokens cost 1 less Power").
Draftable Matron Mother abilities — once per run, choose a faction matriarch with a passive ability that flavors the run (Lolth: assassinations score +1 VP; Vhaeraun: every 3rd recruit is half-price; Eilistraee: promote bonus +50%).
Daily Underdark — a fixed map + fixed pairing + fixed AI rivals + leaderboard by VP margin. Streamer-friendly daily challenge.

What can NOT translate cleanly: the negotiation table-talk metagame. The single best moment in a four-player Tyrants game is the implicit treaty: "If you assassinate me here, I will assassinate you back next turn — we both lose tempo to the third player. Don't." This is structural deterrence. AI opponents cannot table-talk. Replacement: visible Grudge meters per AI rival — an icon next to each rival that fills when you attack them, and at certain thresholds triggers visible retaliation behaviors ("Vhaeraun's House is plotting against you — they will prioritize assassinating your troops next turn"). This converts implicit table negotiation into legible UI feedback. It's a smaller dynamic than the human version, but it preserves the shape of the dilemma.

Other translation losses. The physical tactility of shuffling, fanning a hand, and dropping cards onto the play area is gone — mitigation is haptic feedback on every card draw and a satisfying ASMR-grade shuffle sound. The end-game tension of "is the market half-deck about to run out?" is preserved trivially via a counter UI.

5. Concrete iOS prototype spec

Target: iOS landscape orientation (the map needs horizontal width), single-developer Unity or SwiftUI+SpriteKit, 5-7 month MVP. Landscape, not portrait — this is the major UX departure from a Heat-style port.

90-second core loop. A turn opens with your hand of 5 cards laid along the bottom. Tap a card to flip it into the play zone; both pools (Influence orb, Power flame) tick up. Once you've played all 5, a "Spend" mode unlocks — drag from the Influence orb to a market card to recruit, drag from the Power flame to a map site to deploy/move/assassinate. Long-press any card in hand or discard to promote. Tap "End Turn" to discard, draw 5, and pass to the AI. AI rival turns animate in 3-5 seconds with visible action callouts ("House Baenre recruited Driderkin," "House Mizzrym assassinated your troop at Sshamath"). One full game: 8-12 turns, ~10-15 minutes.

Three-zone landscape layout (6.7" reference):

Top 45%: The Underdark map. Stylized 2D isometric cavern art. 12-16 sites visible, troops as small drow figures with count badges, glowing tunnels for edges. Tap a site to deploy/assassinate; drag a troop from one site to an adjacent one to move.
Middle 15%: The market (6 face-up cards in a row), flanked by the persistent Influence orb (left) and Power flame (right) showing current pool. Promote tray icon top-right with running VP badge. Endgame trigger counter (cards left in market half-decks) top-left.
Bottom 40%: Your hand of 5 cards, large enough for thumb-tap, with the play zone above them. Discard pile and draw deck as small icon stacks far-right; tap to inspect.

Card system.

Starter deck: 7 cards (matching the boardgame Drow starter — typically a mix of low-Influence and low-Power cards plus 2-3 generic). Mirror published balance.
Market: 40 cards drawn from two half-decks shuffled together. Each card has Influence (purple wing icon, top-left), Power (red sword icon, top-right), aspect symbol, ability text, recruit cost, inner-circle promote VP.
Card art: D&D Forgotten Realms aesthetic — drow priestesses, demons, illithids — in a high-contrast painted style. Aspect color-codes the card frame (Drow = violet, Demons = crimson, Aberrations = teal, Undead = bone, Elemental = ochre, Houses = silver).
Promote zone visible at all times as a glowing crest top-right, with a "+VP" badge that pulses each time a card is promoted.

Run structure: Underdark Conquest.

7 maps per run, escalating site count (8 sites on map 1, 16 sites on map 7) and AI count (1 rival map 1-2, 2 rivals map 3-5, 3 rivals map 6-7).
Random half-deck pairing per map; the player sees the pairing pre-game and gets a 30-second mulligan window to study.
After each map, draft 1 of 3 House Perks + 1 of 2 Matron abilities (only first map for matron).
Final map: a named boss house (House Baenre, House Do'Urden, House Vandree) with a unique passive (Baenre: starts with +2 troops; Do'Urden: gains 1 Influence per turn; Vandree: assassinates score double).
Run ends in victory (highest VP on map 7) or elimination (last place on any map after map 3).

Meta-progression.

Unlock new house banners (cosmetic), persistent assassin perks (small passive bonuses across all runs), new Matron candidates, new boss houses.
Daily Underdark challenge with global leaderboard.
About 40-60 unlockables for a 25-35 hour completion arc.

Monetization. Premium one-time purchase, $9.99. Paid expansion seasons at $4.99 each: Season 1 = Aberrations & Undead deck halves, Season 2 = Hazards, Season 3 = a custom-designed "Lolth's Web" pairing not in the boardgame. Avoid energy mechanics and lootbox card-pack mechanics — this is a Slay the Spire / Inscryption pricing model, not a Marvel Snap one. The audience for a D&D strategy deckbuilder pays upfront.

Aesthetic direction. Forgotten Realms cavern aesthetic — deep violets and obsidian blacks, glowing faerzress mushroom-light, drow purple-grey skin tones, demonic crimson accents. UI chrome in carved obsidian with rune-light. Music: dark ambient with tribal drum + dulcimer (think Baldur's Gate 3 Underdark zone). Voice: a drow matron narrator delivers turn prompts and rival barbs in Forgotten Realms cadence — "House Baenre will not forget this slight."

6. Risks and unknowns

Top design risks.

Area-control AI is materially harder than combat AI. A combat deckbuilder AI has clear move-tree pruning — minimax over damage outcomes works. A territory-control AI must reason about emergent equilibria, positional value, and threat coalition (in 3-4 player games, who attacks the leader is a critical question that has no obvious heuristic). Star Realms' AI is mediocre because it doesn't need to be great; Tyrants' AI must be at least credibly aggressive at the right moments or the player will dominate. Mitigation: hand-authored personality archetypes (Aggressor — always assassinates; Builder — only buys; Politician — targets the leader) and a coalition heuristic where AIs preferentially attack whoever has the most VP. This is closer to Civilization AI work than Slay the Spire AI work — budget accordingly.
Dual-resource cards may overwhelm new players. Showing two numbers on every card is more cognitive load than the typical deckbuilder. Mitigation: the first 2-3 maps of the campaign use Drow + a curated subset of Houses (the most balanced, least gimmicky pairing), tutorialized one mechanic at a time. Promote is introduced on map 2, not map 1. Assassination on map 3.
Promote risks getting lost in UI. If promote isn't constantly visible and inviting, new players will play 3 maps without ever using it, then complain the game is shallow. Mitigation: the promote tray glows and gently animates whenever a card in hand has a printed inner-circle value ≥3 — a soft prompt without a tutorial pop-up. Track usage in analytics and pop a one-time hint if a player completes a map without promoting.
Half-deck balance is hard to validate digitally. The boardgame had years of organized-play data; new digital-only pairings need internal playtesting equivalent to a CCG balance pass. Mitigation: ship with only the 6 published half-decks for v1; reserve "Lolth's Web" and other designer-original pairings for post-launch when telemetry exists.
Run length vs. mobile session length. A full Underdark Conquest run is 7 maps × ~12 minutes = ~80 minutes — too long for one sitting on mobile. Mitigation: aggressive cloud save between maps; maps are individually save-resumable.

Validation order.

Paper map first. Before any code, play 5-10 solo Tyrants games with a "dumb AI" rule sheet (no negotiation, fixed priorities) to confirm the game survives single-player at all. The boardgame has no published solo mode; this is real risk.
Greybox single-encounter prototype. One map, one AI rival, no meta-layer. Validate that the dual-resource UI actually reads on a phone — that players can sum and partition pools without rage-quitting. This is the make-or-break test.
Campaign loop prototype. Three maps strung together with House Perk drafts. Validate that the meta-progression feels rewarding without overpowering the per-map balance.
Daily challenge + leaderboard. Ship feature.

Three questions for the prototype builder.

How willing are you to invest in territory-control AI? A mediocre Slay-the-Spire-tier combat AI ships in 4 weeks; a credible Civilization-tier coalition AI is 3-4 months and possibly the largest line item in the project. If you don't have appetite for it, the project halves in scope but the strategic depth halves with it. Pick a side now.
Premium $9.99 with paid seasons, or free-to-play with card-pack monetization? The boardgame audience expects premium and will revolt at gacha. A F2P model would unlock 10× the install base but compromise the design integrity. The roguelite spec above fits premium; a F2P pivot would require redesigning the half-deck pairing system into a card-collection system, which is a different game.
D&D license: do you pursue it, or build in a reskin? Wizards of the Coast licensing is slow and expensive and may not say yes for an indie team. A reskin (your own dark-elf-equivalent IP — "Children of the Black Web" in some made-up subterranean realm) costs nothing legally but loses the search-volume marketing benefit of "D&D" in the title. Decide before art direction.