Planning and Executing the Perfect Heist

Framing the Heist: Ethics, Fiction, and Advanced Play

This module treats heists purely as a fictional / game-design problem, not as real-world criminal instruction.

You’ll approach heists the way a systems designer, speedrunner, or tabletop GM would:

• Modeling risk vs. reward like an optimization problem
• Treating guards, locks, and patrols as game systems
• Designing routes, contingencies, and fail-states for complex stealth missions

We will lean on examples similar to The Elder Scrolls V: Skyrim Thieves Guild questlines and Nightingale powers, plus modern stealth game design (Dishonored, Hitman, Payday 3, etc.). The focus is on planning, analysis, and execution discipline—skills that also map to cybersecurity, operations research, and project management.

You should already be comfortable with:

• Stealth mechanics (light, sound, line of sight)
• Thieves Guild progression and fences
• Nightingale powers and late-game stealth tools

Core idea: A perfect heist in this context means:

> Maximizing value extracted per unit of risk and time, while preserving optionality and minimizing chaos, within a fictional or simulated environment.

We’ll build this as a pipeline:

1. Define the objective space (what counts as success?)
2. Perform reconnaissance and layout modeling
3. Run a target selection and risk assessment
4. Engineer entry, main route, and extraction
5. Layer contingencies and escape trees
6. Stress test with edge cases and failure modes
7. Execute with checklists and real-time decision rules
8. Review, learn, and iterate

Use this like an operations manual for late-game Thieves Guild / Nightingale-style heists.

Step 1 – Defining Success: Objective Hierarchy and Constraints

Before any map drawing or sneaking, define what “perfect” means for this mission.

1.1 Objective hierarchy

Think in three layers:

1. Primary Objective(s)

Non-negotiable goals.

• Example (Skyrim-style): Steal the Jeweled Crown of Barenziah from a Jarl’s private vault.

1. Secondary Objectives

High value but optional.

• Empty 2–3 high-value display cases on the way out
• Pickpocket the steward’s master key for future jobs

1. Tertiary / Style Objectives

Constraints that make the heist clean or elegant.

• No kills, no alerts, no bounty
• No use of loud tools (explosives, obvious magic)
• Complete within a single in-game night cycle

Write this as a stack:

```text

Primary: Crown of Barenziah

Secondary: 3x display cases in gallery, steward’s master key

Tertiary: Ghost (no detection), no bounty, no deaths, 1-night window

```

1.2 Constraints and resources

List your hard constraints and available resources.

Constraints examples

• Guard shifts change at 2:00 and 6:00 (in-game)
• Vault only opens when steward is present
• Magical wards dispel invisibility near the throne room

Resources examples

• Nightingale powers: Shadowcloak of Nocturnal, Nightingale Subterfuge
• Skills: 100 Sneak, 100 Lockpicking, 90 Pickpocket
• Gear: Shrouded armor, muffle enchantments, invisibility potions, paralysis poisons

1.3 Success metrics

For advanced play, define quantitative metrics:

• EV (Expected Value) of loot (gold equivalent)
• Risk score (probability of detection, bounty cost if caught)
• Time-to-complete (in-game or real-time)

You’re starting to treat heists as optimization problems: maximize EV subject to risk and time constraints.

Exercise – Build an Objective Stack

Imagine a high-value target in a fictional city: a noble’s manor that holds a unique enchanted gem and various artworks.

Task (write this out for yourself):

1. Define one primary, two secondary, and two tertiary objectives for this heist.
2. List 3 constraints (e.g., time windows, magic wards, specific NPC schedules).
3. List 5 resources you have (skills, tools, contacts, powers).

Use this template:

```text

Primary:

• ...

Secondary:

• ...
• ...

Tertiary:

• ...
• ...

Constraints:

• ...
• ...
• ...

Resources:

• ...
• ...
• ...
• ...
• ...

```

Reflection: After you write it, ask:

• Are any tertiary objectives conflicting with primary/secondary? (e.g., “no magic” but your only viable entry requires a teleport spell)
• If so, which constraints or style goals would you relax first?

Step 2 – Reconnaissance and Layout Modeling

Recon is where most perfect heists are actually won. In advanced play, you treat recon as building a stateful model of the location.

2.1 Multi-pass recon

Instead of “look around once,” think in passes, each with a focus:

1. Macro pass (outside-in)

• Map building footprint, entrances/exits, roof access, sewers.
• Note guard posts, watchtowers, line-of-sight corridors.

1. Meso pass (public / semi-public areas)

• Follow NPC routines; log times and routes.
• Identify choke points, blind corners, and sound-propagation risks (e.g., stone corridors vs. carpeted halls).

1. Micro pass (near target zones)

• Lock types, trap placements, warded doors.
• Lighting sources, cover objects, noise sources you can exploit (fountains, machinery).

2.2 Building a functional map

Don’t just draw rooms; annotate functionally:

• Nodes = rooms / key positions (e.g., Gallery, Steward’s Office, Vault Antechamber).
• Edges = routes between nodes, with attributes:
• Visibility: exposed / partially covered / fully covered
• Sound risk: quiet / normal / noisy
• Guard density: low / medium / high

Represent this as a graph:

```text

[Roof Access]

| (quiet, low guards)

[Upper Hallway]

| (exposed, med guards)

[Gallery] --- (quiet, low guards) --- [Steward’s Office]

| (noisy, high guards)

[Main Stair] --- (warded) --- [Vault Antechamber]

```

2.3 Time dimension: patrols and windows

Add time to your model:

• For each guard, record a loop: `location → path → pause points → loop time`.
• Note synchronization points (times when multiple patrols overlap a corridor).
• Identify windows: time intervals when a route’s effective guard density is minimal.

Example annotation:

```text

Guard A: Gallery loop (40s)

• 0–10s: Enters from Upper Hallway
• 10–20s: Walks past display cases
• 20–30s: Looks out window (back turned)
• 30–40s: Exits to Upper Hallway

Best window to cross Gallery center: t = 20–30s (back turned, stationary)

```

2.4 Information gaps

List what you still don’t know:

• Unknown: Is there a silent alarm rune in the vault?
• Unknown: Does the steward carry a key or use a spell?
• Unknown: Are there hidden pressure plates near the gem pedestal?

These unknowns will shape your risk assessment and contingency planning later.

Checkpoint – Recon Strategy

Test your understanding of structured reconnaissance.

Step 3 – Target Selection and Risk–Reward Modeling

Now you know the space, decide what is worth taking. In expert play, you don’t loot everything; you optimize.

3.1 Categorize potential targets

For each item or area, estimate:

• Value (V): Gold value or strategic value (e.g., blackmail documents, keys, maps).
• Access cost (C): Time + tool usage + resource expenditure (potions, charges).
• Risk (R): Chance of detection or permanent consequence if something goes wrong.

Example table (conceptual):

| Target | V (gold) | C (time units) | R (0–1) |

|----------------------------|----------|----------------|---------|

| Crown of Barenziah | 5000 | 5 | 0.30 |

| Gallery display case A | 800 | 2 | 0.10 |

| Gallery display case B | 1200 | 3 | 0.25 |

| Steward’s master key | 0 now / 3000 future | 4 | 0.20 |

3.2 Expected value under risk

A simple heuristic:

• Expected Value (EV) ≈ `V × (1 − R)`
• Efficiency ≈ `EV / C`

Using the table above (ignoring future value nuances):

• Crown: EV = 5000 × 0.7 = 3500; Efficiency = 3500 / 5 = 700
• Case A: EV = 800 × 0.9 = 720; Efficiency = 720 / 2 = 360
• Case B: EV = 1200 × 0.75 = 900; Efficiency = 900 / 3 = 300

Here, the Crown is both the primary objective and the most efficient, but Case A is a solid secondary. Case B is marginal.

3.3 Strategic vs. immediate value

Some targets have delayed or strategic value:

• Keys and access tokens (open future routes, reduce future C and R)
• Blackmail material (unlock new fences, safehouses)
• Ledger books (enable better prices, guild upgrades)

Assign them a discounted future value based on your campaign horizon:

• If you plan many future heists in the same city, a master key has high effective V.
• If this is a one-off mission, its V might be near zero.

3.4 Pruning the loot list

Based on EV and your tertiary constraints, decide:

• Must take: Primary targets and any extremely high-efficiency loot along your main path.
• Opportunistic: Only if conditions are perfect (guards desynced, extra time).
• Skip: Low EV or high R relative to payoff.

This is how you avoid greed deaths—getting caught because you couldn’t resist one more shiny object.

Exercise – Compare Two Loot Paths

You have two possible secondary routes after grabbing the main objective.

Route A – Gallery Sweep

• 3 display cases, each V = 600
• C per case = 1 time unit
• R per case = 0.10 (independent)

Route B – Treasury Detour

• 1 chest, V = 2500
• C = 4 time units
• R = 0.40

1. Compute approximate EV and Efficiency for each route (treat each case in A as independent, then sum).
2. Decide which route better fits a ghost-style, low-risk heist.
3. Decide which route you’d pick if your tertiary constraints allowed moderate risk but demanded maximum profit.

Use this scratch template:

```text

Route A:

• EV_case = ...
• EV_total = ...
• Efficiency = EVtotal / Ctotal = ...

Route B:

• EV_total = ...
• Efficiency = EVtotal / Ctotal = ...

For ghost play, I choose: ... because ...

For max profit with moderate risk, I choose: ... because ...

```

Step 4 – Designing Entry, Main Route, and Extraction

With objectives and targets chosen, design the spine of the heist: entry → approach → objective → exit.

4.1 Multiple entry options

Common fictional entry vectors:

• Front-door social infiltration (disguises, forged invitations)
• Vertical entry (rooftops, skylights, balcony grapples)
• Subterranean (sewers, forgotten tunnels, cistern access)
• Arcane / special powers (short-range teleport, illusions, Nightingale Shadowcloak)

Evaluate each by:

• Initial risk (R₀)
• Tool dependency (lockpicks, spells, allies)
• Reversibility (can you retreat via this path if something goes wrong?)

4.2 Main route as a constrained pathfinding problem

Treat the interior path as a shortest path with constraints problem:

• Nodes: All rooms / key junctions.
• Edges: Routes with attributes (time, risk, noise, visibility).
• Constraints: No-kill, no-alarm, max time, etc.

Conceptually, you’re solving:

> Find a path from Entry → Objective → Exit that minimizes a weighted sum of time and risk, subject to constraints.

You might:

• Prefer slightly longer paths with more cover and fewer patrol crossings.
• Intentionally route past resource nodes (healing potions, extra lockpicks) as safety nets.

4.3 Extraction first mindset

Professional planners often design extraction before entry:

• Identify at least two viable exits (e.g., balcony rope, sewer grate).
• Ensure your main route naturally leads toward an exit after the objective.
• Avoid “dead-end” vaults unless you have a guaranteed escape trick (e.g., teleport scroll).

4.4 Route annotations

On your functional map, mark:

• Primary route: solid line
• Backup route(s): dashed lines
• High-risk segments: highlight; note which powers/tools you’ll pre-commit (e.g., use Shadowcloak at exactly this corridor crossing).

Example (conceptual):

```text

[Roof Access]

| (Primary)

[Upper Hallway]

| (Primary, use Invisibility at t=0)

[Gallery]

| (Primary, low risk)

[Steward’s Office] --(Backup to window exit)--> [Balcony]

| (Primary, high risk, use Subterfuge)

[Vault Antechamber] -> [Vault]

| (Extraction via sewer hatch)

[Sewer Tunnel] -> [Canal Exit]

```

Step 5 – Contingency Trees and Escape Routes

Perfect heists assume imperfection. Someone will cough, a patrol will desync, or you’ll misclick. Advanced players design contingency trees.

5.1 Detection tiers

Define levels of compromise:

1. Tier 0 – Ghost: No one suspects anything.
2. Tier 1 – Suspicion: NPCs mildly alert, searching, but no global alarm.
3. Tier 2 – Local Alarm: Area guards mobilized, but city-wide systems not triggered.
4. Tier 3 – Full Alarm: Everyone hostile, exits may be blocked, reinforcements incoming.

For each tier, specify allowed and forbidden actions. For example:

• Tier 1: Allowed – distraction items, short sprints, temporary hiding spots. Forbidden – lethal force.
• Tier 2: Allowed – limited non-lethals, smoke bombs, one-time escape powers.
• Tier 3: Abort heist, prioritize survival and minimizing permanent consequences.

5.2 Decision points and branches

On your route map, mark decision nodes:

• “If guard in corridor A is off-cycle → use backup corridor B.”
• “If gallery search begins → retreat to roof access and wait 1 loop.”
• “If vault room is unexpectedly warded → abort secondary objectives and extract immediately.”

This forms a contingency tree:

```text

Approach Gallery

├─ If patrol normal → proceed via Gallery

│ ├─ If no suspicion → loot Case A

│ └─ If suspicion Tier 1 → skip Case A, move to Steward’s Office

└─ If patrol desynced → use Upper Hallway backup, avoid Gallery entirely

```

5.3 Escape route design principles

Your escape routes should be:

• Redundant: At least two viable exits that don’t share the same chokepoint.
• Asymmetric: Prefer exits that guards rarely use (tunnels, rooftops) over main gates.
• Tool-aware: Map which tools are reserved for escape (don’t spend your last invisibility potion on a minor risk earlier).

5.4 Hard abort criteria

Define in advance when you will abandon the heist:

• If you reach Tier 2 before obtaining the primary objective.
• If you have to consume more than X critical resources (e.g., both Shadowcloak and last healing potion) before the vault.
• If a key NPC (e.g., your inside contact) is missing or hostile.

This prevents sunk-cost fallacy—throwing more risk at a mission that is already compromised.

Checkpoint – Contingency Planning

Evaluate a contingency design choice.

Step 6 – Worked Example: Nightingale Heist on a Jarl’s Vault

Let’s walk through a condensed, high-level example applying the previous steps in a Skyrim-like setting.

6.1 Objective stack

• Primary: Steal the Jarl’s Jeweled Crown of Barenziah from the palace vault.
• Secondary:
• Loot 2 high-value gallery displays.
• Secure the steward’s master key for future city jobs.
• Tertiary:
• No detection (Ghost) if possible.
• No kills; minimal bounty.

6.2 Recon summary

• Macro: Palace on a hill; sewer outlet into river; roof accessible via adjacent tower.
• Meso:
• Guards rotate every 2 in-game hours.
• Gallery has one patrolling guard with a 40-second loop.
• Micro:
• Vault antechamber has a ward that dispels invisibility.
• Pressure plates near the vault door linked to silent dart traps.

6.3 Target modeling

• Crown: V=5000, C=5, R=0.30 → EV≈3500, Eff=700.
• Each gallery display: V=800, C=1.5, R=0.12 → EV≈704, Eff≈469.
• Steward’s key: V (future)=3000, C=4, R=0.20 → EV≈2400 (if you expect many future jobs).

6.4 Route design

• Entry: Roof access via grappling/climb from adjacent tower at 1:30 in-game.
• Primary route: Roof → Upper Hallway → Gallery → Steward’s Office → Vault Antechamber → Vault.
• Extraction: Sewer hatch in vault floor → river exit.
• Backup exit: Double back to roof and descend via tower.

6.5 Contingency tree (simplified)

• If Gallery patrol normal → cross at t=20–30s, loot first display only.
• If Guard enters early (Tier 1 suspicion) → skip all displays, move directly to Steward’s Office.
• If Steward absent → skip key, proceed to vault; mark key as future objective.
• If Ward stronger than expected (dispels Shadowcloak) → prioritize crown, skip displays on exit, use sewer extraction immediately.

6.6 Tool allocation

• Reserve Nightingale Shadowcloak for the Upper Hallway + Gallery crossing.
• Reserve Nightingale Subterfuge for emergency Tier 2 local alarm near Steward’s Office.
• Keep one invisibility potion unused unless you hit Tier 2 during extraction.

This example illustrates how objective hierarchy, recon, risk modeling, routing, and contingencies interlock into a coherent, high-precision plan.

Exercise – Stress-Test Your Plan

Return to the fictional noble’s manor you outlined earlier.

Task: Introduce two unexpected complications, then adjust your plan.

Choose any two:

1. A new magical ward appears on the main staircase, dispelling invisibility.
2. One guard’s patrol loop lengthens randomly by 50% every third cycle.
3. The side entrance you planned to use is now locked with a higher-tier lock than you can reliably pick.
4. Your inside contact (a servant) has been dismissed and is no longer present.

For each complication:

1. Identify which part(s) of your plan it affects (entry, main route, extraction, specific objectives).
2. Decide whether to reroute, change timing, add a tool, or downgrade an objective (e.g., move a secondary to “skip”).
3. Note whether any of your hard abort criteria should trigger.

Use this template:

```text

Complication 1: ...

• Affects: ...
• Adjustment: ...
• Abort? Why / why not?

Complication 2: ...

• Affects: ...
• Adjustment: ...
• Abort? Why / why not?

```

This is the same mindset used in professional risk management and complex operations: you pre-imagine failure modes and rehearse your responses on paper before acting.

Key Concept Review

Flip these conceptual cards mentally or in your notes to reinforce terminology and structures.