# Incremental Delve System - Design Document v2

## Core Concept
Player-controlled crawler descends through underground network. Progression constrained by pressure accumulation and structural stability. No combat, no hard failure states, no forced resets.

---

## Progression System

### Prestige Mechanic
- Player-triggered reset at any time
- **Total resources ever generated = prestige XP** (fairest, no tradeoff with upgrades)
- XP converts to skill points via exponential curve
- All purchased upgrades persist across prestiges
- Restarts crawler at depth 0 with improved capabilities
- Target prestige duration: 2-4 hours (prevent grinding feel)

### Upgrade System
NOT a PoE-style skill tree. Instead: **visible upgrade tracks with escalating costs**

```
Resource Generation      [Level 22] Cost: 1200 → +
Pressure Reduction Rate  [Level 8]  Cost: 450  → +
Movement Speed           [Level 15] Cost: 890  → +
Movement Cost Reduction  [Level 5]  Cost: 320  → +
Collapse Resistance      [Level 3]  Cost: 180  → +
Stability Threshold      [Level 11] Cost: 650  → +
Auto-Stabilize Upgrade   [Level 2]  Cost: 500  → +
Idle Efficiency          [Level 4]  Cost: 280  → +
```

- Each level costs more (1.15× multiplier or similar)
- All options visible from start
- Player allocates points based on current needs
- Can pump into one track or spread points
- No hidden unlocks, no blind choices

### Keystones (needs better name)
Separate from upgrade tracks - rare prestige unlocks or achievement rewards:
- **Rule modifiers with meaningful tradeoffs**
- Limited active slots: start with 1, unlock 2→3→4 via upgrades
- Examples:
  - "Pressure Immunity / Half Movement Speed"
  - "Auto-Stabilize / No Manual Control"
  - "Double Depth Rewards / Double Pressure Accumulation"
  - "Fast Retreat / Cannot Dig Laterally"
- Define playstyle, not raw power
- Always have downsides

### Level/XP Scaling
- Target: ~800 total levels for dedicated long-term players
- First 10-20 levels quick (hook phase)
- Level 200+ is where multipliers explode
- Exponential scaling (Cookie Clicker model: level 98-100 = same XP as 1-98)
- Must balance early hook vs endgame grind

### Fresh Start
- Complete new save, separate from prestige
- Zero progress, zero points
- Player choice for build experimentation
- No mechanical benefit

---

## Resource System

### Single Unified Resource
Generated by digging/traversing nodes. Used for:
- **Movement** - digging to adjacent nodes (cost scales with depth)
- **Pressure reduction** - stabilization actions (rate-limited)
- **Upgrades** - permanent improvements (persistent across prestige)

### Dynamic Allocation Model
Resources flow from generation and must be allocated:

```
Generation: 5.2/sec

Allocation (player-controlled):
→ Stabilization: 0.1/sec (98% effective at current depth)
→ Movement:      4.8/sec
→ Surplus:       0.3/sec (accumulates for upgrades)
```

**Depth scaling:**
- Same allocation (0.1/sec) becomes less effective deeper
- 98% → 95% → 90% → 80%...
- Eventually must reallocate more to stabilization (slower movement)
- Creates organic pacing: fast early, careful mid-game, crawling at extreme depth

**Design question for feedback:**
- Manual allocation (player sets sliders) vs automatic with toggle modes (aggressive/balanced/safe)?
- How much micromanagement is fun vs tedious?

### Generation
- Rate increases with depth (high risk = high reward)
- Both vertical and lateral digging generate resources
- Deeper nodes = better generation rates

---

## Movement System

### Vertical Descent
- Automatic downward progression at player-controlled speed
- Faster speed = more pressure buildup + higher per-node cost
- Movement cost scales with depth

### Horizontal Movement
- Cost scales with current depth
- Used for pressure relief and resource generation
- Safer than vertical at extreme depths

### Upward Movement
- **Always free** (anti-deadlock safeguard)
- Allows retreat to manageable depth
- Completed nodes stay completed
- **Cost**: partial progress on incomplete node is lost
- Implicit defeat (wasted time) without mechanical punishment

### Node Completion
- Only fully completed nodes count as progress
- Incomplete nodes reset if abandoned
- Creates risk/reward: finish node before retreating or cut losses?

---

## Pressure System

### Accumulation
Pressure increases from:
- Current depth (exponential scaling)
- Movement speed
- Traversing unstable nodes

### Effects at High Pressure
- Reduced operational efficiency
- Increased collapse probability
- Lower resource generation throughput

### Reduction Mechanics
- Costs resources (from allocation pool)
- **Rate-limited**: maximum reduction per time unit
- **Depth-scaled costs**: reduction at depth 500 costs far more than depth 100
- Cannot be bypassed by stockpiling resources
- Upgrades increase rate-limit by 1.15× per level

### Anti-Exploit Design
Prevents "dig laterally for days then blast downward":
- Lateral mining reduces current pressure + generates resources
- But pressure accumulation resumes immediately when descending
- Rate-limits prevent burning stockpiled resources for trivial progression
- Depth-scaling ensures shallow resources don't trivialize deep content

---

## Unstable Nodes & Collapse

### Unstable Nodes
- Marked nodes that rapidly increase pressure
- Higher collapse risk when overstressed

### Collapse Consequences
- Local area becomes inaccessible
- Surrounding areas destabilize
- Temporary pressure spike
- Requires rerouting

**Does NOT cause:**
- Progress loss
- Upgrade resets
- Game over

Collapse = topological challenge, not punishment.

---

## Active vs Passive Play

### Auto-Stabilize System
**ON:**
- Maintains stable pressure automatically
- Costs 20-30% more resources (inefficient AI)
- Good for idle/away time

**OFF:**
- Manual control required
- Optimal efficiency if played well
- Risk of pressure spike if neglected
- Rewards active engagement

### Other Potential Auto-Systems
- Auto-retreat: pulls up when stuck, costs resources
- Auto-upgrade: buys upgrades when affordable, picks sub-optimally
- Auto-allocate: balances resources conservatively

**Philosophy:** Active play = better efficiency. Idle = still progresses, just slower.

---

## Offline Progress

### Implementation
- Crawler stays parked at current node
- Generates resources at minimal rate (5-10% of active rate)
- No pressure changes, no movement, no risk
- Capped at 24-48 hours max

### On Login Message
"Crawler maintenance complete. Extracted 342 resources from debris and system cleanup."

**Simple calculation:**
```
offline_time = min(time_since_logout, 48_hours)
idle_resources = offline_time × base_idle_rate × depth_multiplier
```

**Philosophy:**
- Rewards active play (10-20× better)
- Casuals still progress
- Can't be exploited (time-capped)
- No complex simulation needed

---

## Upgrade Categories (To Be Expanded)

### Crawler Systems
- Dig speed
- Movement efficiency (lower cost per node)
- Durability (collapse resistance)

### Pressure Management
- Stabilization effectiveness
- Rate-limit increase
- Passive pressure decay
- Emergency vent systems

### Resource Processing
- Extraction efficiency
- Processing multipliers
- Storage capacity (if relevant)

### Detection/Automation
- Collapse prediction
- Auto-stabilize triggers
- Hazard scanning range
- Auto-retreat thresholds

### Infrastructure
- Supply line efficiency
- Backup power systems
- Redundant stabilizers

### Depth Unlocks
- New tech available only at certain depths
- Deep-specific upgrades

**Design goal:** Upgrades should feel qualitatively different, not just "number goes up"

---

## Engagement Problem & Solutions

### Identified Issue
Player engagement drops when progress shifts from hours → days/weeks per milestone.

### Proposed Solutions

**1. Multiple Parallel Progression Tracks**
- When prestige levels slow, skill points keep coming
- When skill points slow, depth unlocks appear
- When depth slows, keystone options unlock
- Always multiple things progressing at different rates

**2. Frequent Prestige Loops**
- Target 2-4 hours per prestige
- Reset before grind sets in
- Maintain dopamine flow

**3. Qualitative Unlocks**
- Every X prestiges: new mechanic/system/rule
- Not "10% more resources" but "unlocked auto-retreat"
- Gameplay changes, not just scaling

**4. Front-Loaded Design**
- Accept game may be 30-50 engaging hours
- Design for quality over infinite scaling
- Natural endpoint when content exhausted

---

## Core Design Loop

1. Descend at chosen speed
2. Pressure accumulates
3. Allocate resource flow:
   - Movement vs stabilization vs surplus
4. Monitor collapse risk
5. Make upgrade purchases
6. Choose when to prestige
7. Reset stronger with more options

**Strategic tension:** Speed vs safety vs optimization vs prestige timing.

---

## Design Constraints

- No permanent loss
- No hard failure states
- No forced idle waiting
- Active play always better than passive
- Older saves always stronger than fresh starts
- Progress measured by efficiency, not raw numbers

---

## Open Questions for Feedback

1. Manual resource allocation (sliders) vs automatic with mode toggles?
2. How many upgrade categories before it becomes overwhelming?
3. Prestige curve tuning - what feels right for 2-4 hour loops?
4. How to maintain engagement past 30-40 hours of play?
5. Balance between active rewards vs idle accessibility?
6. Is the pressure/stability system engaging enough as core constraint?

---