The Thief Problem
Published 2/24/24
Last updated 5/4/25
Table of Contents
Introduction for
2nd Version
The central question of this essay is: How can a story deal with character motivations that seem independent of the plot?
This essay aims to do the following:
-
Describe a framing device - a computer game model's pseudo-code documentation, including a play environment and character classes (Sections 1-4)
-
​Present the central question in context as motivations for/implications of a particular play strategy, the eponymous "Thief Problem" (Sections 5a and 5b)
-
Suggest multiple narrative tropes and counter-strategies as "solutions" for the "problem" (Sections 5c, 5d and Appendix)
I'm not a programmer making a feasible game. I'm a narratologist making an argument about a story archetype.​
The generalized argument uses a framing device to show its applicability for a lot of stories. This means the argument ignores certain valid storytelling elements that I reintroduce later in Sections 5b and 5c (e.g. superheroes have different superpowers and personalities, which impact how they rescue civilians; supervillains might use grand and elaborate evil plans but end up targeting specific characters; the characters have their own complex interpersonal relationships with each other, etc.).
Because the examples I provide throughout deal with fundamental archetypes and plot patterns, assume all the examples, especially in Section 5b, are spoilers for their respective media.
This second version of the essay elaborates on and clarifies the applicability of the Thief Problem.
​
Patch Notes:
-
Revised introduction
-
Summary comments on Table 1 and Section 5b about the Thief character class
-
Restructured Section 5b with more examples
-
Added six additional strategies (Strategies 11-16) to Section 5c
-
Reformatted Section 5c to distinguish between descriptions of solutions and commentary
-
Incorporated Appendix into main body of essay instead of external link
-
Other minor revisions and re-wordings
Section 1:
Core Assumptions
The model makes the following core assumptions:
-
The model represents a given finite two-dimensional (2D) Environment and a given period of time called "active play".
-
The Environment is populated by characters with different character classes. (See the diagrams in Section 2 for a basic illustration of what the Environment and characters might look like.)
-
​​Characters move throughout the Environment during active play and interact with each other. Assume that the model can calculate characters’ position coordinates, variable 360-degree direction vectors, constant character velocities, 360-degree character hitboxes of constant size, and any other data needed to facilitate two-dimensional movement within the set Environment.​
-
Character interactions can vary based on each character’s class.
-
-
“Dangerous” events can start spontaneously within the model.
-
Dangerous events can be expressed by a nominal change in location inside the Environment. This change corresponds to characters “in the dangerous event” or “not in the dangerous event”.
-
Characters who can respond to dangerous events have some intrinsic, implicit motivation to respond which is not expressed in the pseudo-code.
-
The model represents a zero-sum one-player game. If the player character wins, the other non-player characters lose.
​
The tropes and counterstrategies we explore later are assessed based on their consistency with these Core Assumptions - I treat strategies that would require a change to a Core Assumption as less useful than a strategy that doesn't.
​​
Variables in the Environment are expressed in bold. Narrative comments are expressed in italics until Section 5b, at which point it is not necessary to distinguish between the two.
Section 2:
Basic Elements
The model consists of the following elements based on Section 1:
Diagram 1: a character populating the Environment
Section 3:
Location Variables
“Perilous” is a nominal location variable assigned to each character during active play. (See Endnote 1.)
-
“Perilous” is expressed as a status “In Peril” when the character is within the Peril area.
-
“Perilous” is expressed as a status “Not In Peril” when the character is not within the Peril area.
Diagram 2: a character populating the Environment but not Peril
Diagram 3: a character populating Peril
“Ability to enter Peril” is a nominal ability constant assigned to each character during active play and defines conditions for Perilous changes.
“Ability to enter Peril” is expressed as “Free Enter” when a character Not In Peril can become In Peril without a character interaction. A character with this ability who is Not In Peril can enter Peril at any time.
“Ability to enter Peril” is expressed as “Force Enter” when a character Not In Peril can become In Peril only during a character interaction.
“Ability to exit Peril” is a nominal ability constant assigned to each character during active play and defines conditions for Perilous changes.
“Ability to exit Peril” is expressed as “Free Exit” when a character In Peril can become Not in Peril without a character interaction. A character with this ability who is In Peril can exit Peril at any time.
“Ability to exit Peril” is expressed as “Force Exit” when a character In Peril can become Not in Peril only during a character interaction.
​
A character with Free Enter and Force Exit can go into Peril alone but can't leave Peril without a buddy.
​
A character with Free Enter and Force Exit can leave Peril alone but can't go back into Peril without a buddy.
Diagram 4: A character with Free Enter and Force Exit.
Diagram 5: A character with Free Enter and Force Exit.
Section 4:
Character Classes
Character classes are defined by character location variables.
​
1. “Civilian”
Non-playable character class “Civilian” Force Enters Peril and Force Exits Peril. Civilians don't willingly go into harm's way and can't get out of danger by themselves. Core Assumption 3 suggests that Civilians might end up in Peril spontaneously.
2. “Hero”
Playable character class “Hero” Free Enters Peril and Force Exits Peril, to account for a moral incentive to save as many Civilians as possible. The model counts every instance a Hero and a Civilian interact, allowing for duplicates.
​
a) When a Civilian In Peril and a Hero interact, both the Civilian and the Hero Force Exit Peril.
i) Ratio variable “Power of Good” tracks and assigns a positive ability weight factor to the number of Civilian Force Exits Peril during active play.
b) When a Civilian Not In Peril and a Hero interact, there is no change.
c) A Hero has a class objective “Safety” to Force Exit as many Civilians in Peril as possible during active play (with a maximum of every Civilian in the Environment). Remember this is a narrative motivation about what a character wants to do in a story, regardless of what emotional motivation may have inspired the character’s actions.
i) A Hero also achieves Safety if all Civilians are Not In Peril.
ii) If a Hero achieves Safety, the Hero wins and active play ends.
3. “Villain”
Playable character class “Villain” Force Enters Peril and Free Exits Peril, to account for a moral incentive to hurt as many Civilians as possible and/or take hostages. The model counts every instance a Villain and a Civilian interact, allowing for duplicates.
a) When a Civilian In Peril and a Villain interact, there is no change.
b) When a Civilian Not in Peril and a Villain interact, the Civilian Force Enters Peril.
i) Ratio variable “Power of Evil” tracks and assigns a positive ability weight factor to the number of Civilian Force Enters Peril during active play.
c) A Villain has a class objective “Takeover” to Force Enter as many Civilians in Peril as possible during active play (with a maximum of every Civilian in the Environment).
i) A Villain also achieves Takeover if all Civilians are In Peril.
ii) If a Villain achieves Takeover, the Villain wins and active play ends.
4. “Thief”
Playable character class “Thief” Free Enters Peril and Free Exits Peril. The model counts every instance a Thief and a Civilian interact without duplicates.
a) The Thief has a conditional ability “Pickpocket”.
When a Civilian In Peril and a Thief interact, a coin is flipped.
The Civilian has a 50% chance of Force Exiting Peril, and the Thief has a 50% change of Force Exiting Peril, to account for not getting caught pickpocketing versus getting caught pickpocketing, respectively.
When a Civilian Not in Peril and a Thief interact, a coin is flipped.
The Civilian has a 50% chance of Force Entering Peril, and the Thief has a 50% chance of Force Entering Peril.
b) Ratio variable “Sneakiness” tracks and assigns a negative ability weight factor to the number of the Thief’s Civilian Force Enters Peril or Civilian Force Exits Peril during active play, to account for Civilians/Heroes/Villains noticing the Thief more as the Thief pickpockets everyone. (See Section 4a for more about Sneakiness.)
c) A Thief has a class objective “Loot” to interact with as many Civilians as possible during active play (with a maximum of every Civilian in the Environment). This would account for the Thief stealing from everyone they can and escaping without getting caught. If a Thief achieves Loot, the Thief wins and active play ends.
Table 1: Summary of Character Location Variables
Basically, the Civilian has to have a buddy character to go in or out of Peril.
The Hero has to have a buddy character (Civilian) to go out of Peril.
​
The Villain has to have a Civilian to go into Peril.
​
The Thief does not need a Civilian to move between Peril and the rest of the Environment.
Section 4a: Battle
When two non-Civilian character classes interact, they “Battle”. Their ability weight factors are recorded.
​
The character with the higher ability weight factor (“the winner”) experiences no change.
The character with the lower ability weight factor (“the loser”) experiences the following penalties:
-
The loser Force Exits Peril to a random location not in Peril AND
-
The loser cannot move in the Environment for ten (10) seconds AND
-
The loser’s ability weight factor resets to zero (0).
​​
A draw (the characters’ ability weight factors are equal) is considered a double loss: both characters are losers and receive all penalties listed.
​
To mitigate some of the existing Sneakiness penalty, the Thief has a dormant automatic ability “Unlock” which reduces the time a losing Thief cannot move in the Environment to 5 seconds. All other penalties listed apply.
Diagram 6: Two characters Battling.
When the example active play starts, the model generates the following characters:
-
A given positive odd number of Civilians in Peril (at least 5), to account for random everyday emergencies.
-
One (1) Hero
-
One (1) Villain
-
One (1) Thief
​
This is a 1-player game. The player can choose to play as the Hero, the Villain or the Thief per session of active play.
Civilians and the characters the player does not choose are considered non-player characters controlled by the computer. These characters can identify when the player is near them, but they do not automatically turn their attention towards the player. Their class objectives apply.
​​
There is no time limit. Active play continues until a playable character wins.
Section 5:
Reasonable Strategic Conclusions
We can draw multiple strategic conclusions from this model:
​
1. The Hero is incentivized into Peril to achieve Safety but not to confront the Villain until the Hero has achieved a significant Power of Good. The Hero won’t Battle the Villain until the Hero thinks they can win.
a. Because some Civilians pre-populate Peril, we can frame the Villain bringing more Civilians into Peril and the Hero removing Civilians from Peril as “acting and reacting”.
2. Corollary to Conclusion 1: The Villain is incentivized into Peril to achieve Takeover but not to confront the Hero until the Villain has achieved a significant Power of Evil. The Villain won’t Battle the Hero until the Villain thinks they can win.
3. The Thief has the explicit class objective Loot to interact with Civilians. But both Pickpocket and Sneakiness reward future interactions with Civilians and punish future interactions with Heroes and Villains. So the Thief is incentivized to interact only with Civilians: not Heroes and not Villains, unless they absolutely have to. For example, see EVA is section 5b.
a. For the same reason, should other Thieves populate later stages of active play, Thieves are also not incentivized to interact with other Thieves. “Unlock” only makes sneaky Thieves sneakier.
4. A Thief is equally likely to make progress towards Loot when interacting with Civilians in Peril and Civilians Not in Peril. The Thief is also equally likely to end up in Peril themselves as not. This incentivizes the Thief to be strategic about location and timing when interacting with Civilians - strategic in ways that don’t apply to the Hero or the Villain.
5. Because Loot incentivizes the Thief to interact with all Civilians in the Environment, the Thief has a sense of moral ambiguity during active play. The Thief can Loot and support either the Hero’s Safety or the Villain’s Takeover - or both - or neither.
6. Because the Thief moves in and out of Peril during active play, and because neither Safety nor Takeover explicitly depend on interaction with the Thief, the Thief is not always the focus of either the Hero or the Villain. Or: neither the Hero nor the Villain have a win condition that requires or incentivizes interaction with the Thief during their short interactive periods inside/outside Peril.
7. The penalty for the Thief from the Power of Good and the penalty for the Thief from the Power of Evil are equal. Therefore, the Thief cannot afford to risk their Sneakiness being either equal to or lower than their respective opponent’s Power of Good or Power of Evil in case of a Battle.
8. Safety reaches its logical conclusion when all Civilians in the model are not in Peril.
9. Takeover reaches its logical conclusion when all Civilians in the model are in Peril.
10. Loot reaches its logical conclusion when the Thief has interacted with all Civilians in the model, regardless of whether they are in Peril or are not in Peril.
Section 5a: "The Thief Problem"
Therefore, a Thief strategy presents itself, “the Thief Problem”.
The player characters will tend to not interact with each other unless they think they can win a Battle. The Thief will try to get stronger in preparation for Battling, but that preparation puts them at risk of a costly Battle with the Hero or Villain before they're ready.
One version of the strategy: The Thief can make the Hero and the Villain Battle each other throughout active play at some point close to the center of Peril, while the Thief goes to the border of Peril and pickpockets as many Civilians as possible.
​
A second version of the strategy: the Thief may choose to interact with either the Hero or the Villain - but only when necessary and only when they are confident their Sneakiness is higher than either the Power of Good or the Power of Evil. The Hero and Villain stay in the middle of Peril with each other while the Thief can escape at any time.
This is not the only available strategy or even the most rational strategy in all active models. However, the strategy is viable and illustrative. Distraction, sabotage, snares and sniping from a distance are all valid Thief strategies. They are all expressions of the problem.
The Thief forces the Hero to spend time and resources on a third goal (catching the thief) instead of increasing Power of Good or achieving Safety. Likewise, the Thief forces the Villain to spend time and resources on catching the Thief instead of increasing Power of Evil or achieving Takeover.
The Power of Good and the Power of Evil eventually have dramatic impacts on the later stages of active play. This gives any given run of active play a soft time limit. However, the Thief Problem is at least one Loot strategy that is quicker than either Safety or Takeover.
​​​
It is easier for the Thief to achieve Loot than for either the Hero to achieve Safety or the Villain to achieve Takeover. The true soft time limit of the model is Loot.
As long as Takeover and Safety are active incentives in the model, the Thief Problem is possible. The "character motivation that seems independent of the plot" is Loot.
​
​
Another way to express the problem: Safety has a maximum of every Civilian in the Environment.
Superheroes have to do triage, making choices about which civilians to save and when.
If a supervillain is not around to take over the world, superheroes are incentivized by volume to save the comparatively larger population of civilians from peril at the cost of missing the middle-level, sneakier, and less common threats of thieves.
If a supervillain is in the Environment to take over the world, superheroes are incentivized by scale to stop the takeover at the cost of missing the middle-level threats of thieves. Villains simply require too much time and attention from a Hero. When the Hero is busy saving the world, they cannot focus on Thieves or help the medium-term quality of life for Civilians.
Interacting with the Thief is a long-term goal with medium-term consequences for either the Hero or the Villain - and the Thief Problem means the Thief doesn’t just know or acknowledge that but is actively strategizing with it in mind.
The Thief can’t be in two places at once, but neither can any other character.
Thieves are often used as a method for escalating drama. A rookie Hero first fights street thugs, then Thieves, then Villains. But Villains require so much time and attention from a Hero that by the time the Hero and the Villain have Battled each other, the Thief can Loot and win the game.
The Thief’s ideal scenario involves distracting both the Hero and the Villain. So, another viable version of this strategy is to bring both the Hero and the Villain into Peril during the early stages of active play without necessarily Battling them - instead, the Thief just brings them together. If the Hero and Villain both lose (because of low Power of Good and low Power of Evil), or even one of them loses, or if they flee from each other, the Thief is busy increasing Sneakiness or working towards Loot.
The Thief Problem is that Thieves are both skilled enough to achieve their objective and quiet enough to not get caught.
The Hero and the Villain have objectives that are not just in direct conflict with each other but are mutually exclusive - when the Hero takes action towards Safety, that is also an action against Takeover (and vice versa). Conclusions 5 and 6 show that this exclusivity does not apply to Loot.
The Thief is not a Hero’s distraction from the Villain. The Villain is a Hero’s distraction from the Thief.
A Thief with an active class objective is not an optional side quest for a Hero or Villain - they are a part of a fundamental triage. Unless the Thief is not just caught but constantly managed during active play, they can achieve Loot and win.
​
In short, "The Thief Problem" is: you can't win by going after the Thief, but you can lose by ignoring them.
​
We assumed the former part (you can't win by going after the Thief) from defining the character classes, and we concluded the latter part (you can lose by ignoring them) through strategy.
​
Does the Hero Battle the Thief - who is more difficult to catch but more likely to win, and who might be helping the Hero? Or does the Hero Battle the Villain - who is causing obvious and large-scale Peril, who is actively undoing the Hero’s work towards Safety, and who has their own incentive to Battle the Hero?
​
Can the Hero achieve Safety dealing with both the Villain and the Thief at the same time?
Section 5b: Examples of “The Thief Problem”
This is a non-exhaustive list of examples from video games, movies, TV shows and other media. The list documents moments where the Thief is successful because the Hero is distracted by Safety or the Villain. It sorts across media by most exemplary (inspires or codifies the trope) to least exemplary (an example with some exceptions).
​
In active play, the Thief uses strategies that align with other archetypal thief, trickster and spy characters in fiction.
​
If a character at any point expresses a sentiment to the effect of "I'm not on anyone's side but my own", they're probably a Thief. If the character can be described as a "femme fatale" for their actions more than for their aesthetic (although the term is not limited to femme-presenting characters), they're probably a Thief.
​
Many of the classic examples also develop one or more solutions to The Thief Problem.
​
SPOILERS AHEAD.
​
Bond girls and femme fatales:
-
Black Cat, the professional thief - in many of her appearances, but especially in the Marvel’s Spider-Man (PS4, 2018) DLC, during which she leads Spider-Man on a wild goose chase that itself becomes part of a heist - and the game's 2023 sequel, during which she leads Spider-Man on another chase after stealing a magic item and tries to escape New York City. A classic example of the trope.​
-
Catwoman, the professional thief - in many of her appearances, has an on-again-off-again romance with Batman that only sometimes stops her from stealing. A classic example of the trope.
-
Irene Adler, the professional opera singer/blackmailer/thief - in many of her appearances, but especially the Guy Ritchie Sherlock Holmes movie (2009), in which she works with Holmes and Watson to foil the main villain’s plan - but her real goal is to steal part of the doomsday device for another villain. A classic example of the trope.
-
The Thief is a common role for “Bond girls” in the James Bond books/movies. Casino Royale’s Vesper Lynd (1953), Skyfall’s Moneypenny (2012), and Spectre (2015)/No Time to Die’s Madeleine Swann (2021), among others, have Thief qualities.
-
EVA, the spy in Metal Gear Solid 3 (2004), during which she disguises herself as an NSA agent who defected to the Soviet KGB, aids Naked Snake while pretending to be the main villain’s lover, and (believes she) steals the intelligence everyone was after the whole game. She also borrows these elements from the classic Bond girls that inspired her character. See Ocelot below for comparison.
Classic and archetypal tricksters:
-
Tom Sawyer, the young prankster in Adventures of Huckleberry Finn (1884), distracts Huck and Jim in an elaborate performance to rescue Jim from bondage - when Tom knows the whole time that Jim is already a free man. This, and his antics in his own novel (Adventures of Tom Sawyer (1876)), make him a classic example of the trope.
-
Part of the Fellowship of the Ring’s (1954) strategy includes gathering an army to attack Mordor while Frodo Baggins sneaks into Mount Doom to destroy the Ring. But Frodo's role extends throughout the Lord of the Rings trilogy into more of a classical hero.
-
Bilbo Baggins serves a similar role in Thorin Oakenshield’s dwarf party to recover his mountain treasure from the dragon Smaug in The Hobbit (1937).
-
-
In a traditional African folk story, the brer rabbit is caught by the brer fox. The rabbit begs and pleads for any fate except being thrown in the sharp, tangled briar patch. The cruel fox tosses the brer rabbit into the patch, not knowing the rabbit lives in the patch and can now escape into the protective thicket. Likely one of the original versions of this trope.​
-
Monsieur and Madame Thénardier, the innkeepers from Les Misérables (the 1862 novel and the 1980 musical), are literal comic relief thieves. They take advantage of situations like Fantine’s death and the student’s deaths during the rebellion for profit.
-
Superheroes like Spider-Man and Ant-Man are usually Thieves when they are not explicitly the protagonists of a given story, especially in their Marvel Cinematic Universe appearances.​
​
Manga and animation:
-
The Thief is also a common role for characters in the Bleach manga (2001-2016). The Soul Reapers Kisuke Urahara, Yoruichi Shihoin, Mayuri Kurotsuchi, Shinji Hirako, Aizen Sosuke and Gin Ichimaru each are patient enough to wait until opportunities present themselves towards their goals and quick enough to exploit those opportunities. In other circumstances, these characters are mentors (Kisuke, Yoruichi) or allies to the heroes when their motives align (Mayuri; Shinji says he and the Vizoreds are allies of the hero Ichigo and enemies of Aizen but are not allies of the other characters in the Fake Karakura Town Arc; Gin betrays Aizen at the end of the Fake Karakura Town Arc, Aizen himself fights against Yhwach at the end of the Thousand Year Blood War Arc). Their battle tactics rely on speed, distraction, and layered attacks within attacks. Their actions help instigate or increase stakes in the plot (especially Kisuke and Aizen). Those actions are often not revealed until after they have happened, especially during battles.
-
Blackbeard, the pirate in the manga One Piece (1999-present), starts the story with no bounty. He betrays members of his own crew to become a Warlord of the Sea and get immunity from the World Government that wants to eliminate all pirates. He uses that immunity in the Impel Down/Marineford War arcs to break into a giant prison, recruit its most dangerous prisoners for a crew under his name, and steal the power of his former captain Whitebeard. He goes from no bounty to one of the most dangerous characters in the world in a matter of days. This is also Blackbeard's transition from Thief to Villain both because of his newfound power and ideological opposition to the Hero, Luffy.
-
In the same Marineford War Arc, Luffy is also a Thief from the Marines’ perspective. He only allies with Whitebeard because of their common ally Ace. He’s a pirate with more reputation because of his family than because of his own strength compared to the Emperor crew/Warlords/Admirals on the battlefield. He takes advantage of other distractions the Whitebeard Pirates provide to free Ace, but he still boasts about being Pirate King in front of Whitebeard himself. And Luffy is not strong enough to successfully defend Ace from the Admirals.
-
-
Greed, the Homunculus anti-hero in the manga Fullmetal Alchemist (2001-2010), is as loud and attention-seeking throughout the story as his name implies. However, in much of the final battle against the villain Father, Greed hides in the background. He sacrifices himself to turn Father into brittle carbon and give the heroes an opening to defeat Father.
-
Aizawa, the teacher in the manga My Hero Academia (2014-present) has the power to turn off other characters' superpowers. After he does, he can capture the characters when their guard is down. Smart villains prioritize defeating Aizawa as soon as they can for this reason, but then they can leave themselves open to counterattacks.
-
In the final battle against Azula in Avatar: The Last Airbender (2005-2008), Katara’s strategy relies on defense, diversion away from the injured Zuko, then immobilizing and capturing Azula.​
​
Other examples:
-
Revolver Ocelot, the pistol-twirling spy from the Metal Gear Solid video game series (1998-2015) is often a double- or triple-agent who puts on a show of working for the villain of a given game before betraying them for someone else. He, at various points, works as an NSA defector, a GRU commander, a KGB spy, a CIA spy, and a La-Li-Lu-Le-Lo operative - often more than one of these per game.
-
The DC comic book storyline “JLA: Tower of Babel” (2000) and an animated movie adaptation Justice League: Doom (2012) use the Thief Problem itself as a plot structure. Supervillains from the Justice League member’s gallery each invoke the Thief Problem to sabotage their respective heroes. And each villain is a part of Ra’s Al Ghul’s/Vandal Savage’s overall Villainous plan.
-
Milo Minderbinder, the Air Force camp’s mess officer in the novel Catch-22 (1961), treats his duties to acquire fresh ingredients for the mess hall like a business. He makes money hand over fist but doesn’t seem to put the main character Yossarian in danger at first. He eventually ends up contracting missions for both the American airmen at the base and their German counterparts - during World War II - and bombs his own base just to keep profits up.
-
Many of the proxies/accomplices in Slenderman stories play the role of Thief until they get caught, often by the hero.
-
Totheark in the webseries Marble Hornets (2009-2014) is willing to harass and terrify Jay but also allies with Jay to kill Alex. One of the members of totheark dies, but the other successfully kills Alex and seemingly stops the Operator.
-
HABIT in the ARG EverymanHYBRID (2010-2018) is cryptic enough about his goals that he can influence Vinny to help him break out of a reincarnation cycle even after killing or possessing all of Vinny’s friends.
-
Vinny himself ended up knowingly leading a lot of his friends to their deaths.
-
-
-
This is not limited to a specific wrestler, but WWE has a “Money In The Bank” pay-per-view that offers its winner an opportunity to commandeer an existing scheduled match. When the Money in the Bank holder cashes in their contract, they turn the event into a championship match. This lets the holder enter the match at their convenience - usually (in kayfabe) by surprise and towards the end of the match against two already exhausted wrestlers - and cheat the “original” winner out of their win.
-
One propaganda technique from the novel 1984 (1949) suggests that the rebels against the INGSOC party are somehow both constantly losing to the INGSOC military and also may shift from ally to enemy on any given day (not that a member of the party should think that much about it in the first place). An example of a Villain creating a Thief deliberately to invoke The Thief Problem on the Hero.
-
An inversion of this trope: The Indiana Jones movies (1981-2023) usually feature a sequence of Indiana Jones solving the puzzles and escaping/deactivating any of the traps - which makes it easier for the villains to swoop in afterwards and snatch the artifact. Considering the villains also inevitably die to their own hubris, Indy is an example of both a Hero and a Thief.​
​
It’s important to remember I’m classifying these characters as Thieves based on their skill sets and activities in the plot. Many of these examples are certainly “protagonists” (who go through character arcs) or “antagonists” (who oppose the protagonists’ arc) of their respective stories, and many are not so easily sorted. Regardless of their moral alignments as “hero”/”villain”, their archetypal roles as “protagonist/antagonist”, or their actual occupations in the story, these characters use Thief strategies to accomplish their goals - this makes them useful as examples of Thieves, even if they aren’t literal thieves who steal MacGuffins.
Section 5c: Proposed Solutions to “The Thief Problem”
Writers (comic book writers especially) develop contextual strategies for patching the a-contextual Thief Problem exploit - as in, ways to stop the Hero/Villain from focusing on the Thief. These are well-known tropes in their own right. I think it is useful to frame them through this common origin.
​
Some strategies are cataloged without exhaustion here for reference. They are sorted as variations on strategies as they arise.
​
Strategy 1: “The Bohemian Scandal”
The Hero (or the Villain) lets the Thief approach Loot intentionally in a low-stakes conflict to advance their own class objectives, much like Irene Adler tricks Sherlock Holmes. Batman and Catwoman’s relationship is another classic example.
Counterpoint: This is not a sportsmanlike strategy because it is collusion in a zero-sum context, which breaks Core Assumption 6.
​
Counterpoint: This strategy does not solve the Thief Problem - the Thief still wins, which is a predicted outcome of the problem. Long-term stories can run into well-established difficulties trying to justify why exactly this strategy is in play.​
​
Strategy 2: “Fridging”
Villains can directly attack the Hero’s friends, family or partners. While the Hero feels fear, concern, urgency, grief, etc., they can either remove themselves from Peril to recover or focus even more on the Villain (while the Villain stays calm and focused on Takeover).
We solve the Thief Problem by ignoring the Thief with an emotional incentive.
Counterpoint: This has its own problems. especially for tone. Inflicting trauma onto side characters to motivate your main characters is a common trope, especially trauma onto women of the cast. It’s useful as an origin story to motivate a Hero into Peril, but Civilians who get similar trauma without the requisite superpowers just have to deal with that in therapy. This is a lot of emotional damage and baggage that your characters have to deal with now.
Counterpoint: This strategy is an extrinsic motivation, which breaks Core Assumption 5.
​
Strategy 3: “Serial Escalation”
Across a long-term model, a Hero experiences larger obstacles, stronger Villains and Thieves, and more dramatic adventures. Rescuing a kitten from a tree turns into saving a train from crashing and turns into destroying a killer robot that will destroy the city.
We solve the Thief Problem by ignoring the Thief with a physical incentive.
Counterpoint: Assuming audience interest in the story is growing along with your stakes means everything is as dramatic as possible all of the time. This exhausts both your characters and your audience.
Part of the original motivation for this essay was the comment from the character Vision in the Marvel movie “Captain America: Civil War” that both the number of [characters] with superpowers and the number of potentially world-ending events [in the franchise] have risen at the same rate. This is a stupid comment that confuses correlation with causation and also ignores that each superhero has their own independent rouge’s gallery with independent vendettas.
In short, this strategy is unsustainable.​
​
Strategy 4: “The Sky is Falling”
A variation of Serial Escalation.
A new threat appears that forces Heroes, Thieves and even some Villains into cooperation just to stop it. Whether it’s a craving hunter or a meteor or something in between, it changes the character relationships.
We solve the Thief Problem by making the Thief an ally.
Counterpoint: A falling sky is temporary and does not change the character’s long-term class incentives. If it does, it breaks Core Assumption 2, that the characters can be categorized with distinct classes.
Counterpoint: Depending on how a story uses this type of apocalyptic scenario, a falling sky could also break Core Assumption 4 by creating a scenario in which there is no meaningful difference between Civilians in Peril and Civilians not in Peril. If the threat is big enough that the entire Environment is in Peril (because all characters will be hit by the falling sky unless the threat is stopped), this is an automatic win condition for the Villain.
Counterpoint: A writer may not find the sky falling - or even Takeover itself - a desirable outcome. Regardless of their feasibility during active play, a doomsday scenario that turns from impractical to impending means the original motivations for active play are irrelevant. Like Fridging, we ignored the character’s extrinsic motivations. The end of the world is the ultimate extrinsic motivation. But once the world is not ending, the Thief needs some other intrinsic motivation to stay an ally.
Counterpoint: The same tone maintenance reasons why Fridging is not always desirable also apply here.
Counterpoint: See the “Paint the Town Blue“ DLC appendix for more about why this strategy may be unsustainable.
​
Strategy 5: “Cast Crunch”
A variation of Serial Escalation and a possible response to Fridging.
The Villain kills the Thief. This is popular in sequels for establishing how dangerous a new Villain is. The Hero thinks they’re dealing with a Thief who steals precious metals but they’re actually dealing with a black ops agent with a grudge.
Counterpoint: This strategy inherits the tone problems of Fridging and the lore-dependent artificial drama of Serial Escalation.
​
Strategy 6: “Cast Bloat”
The inverse of Cast Crunch.
Some superheroes fight street thugs in the city. Others fight international pro villains. Still others save the whole universe. Each of these heroes may have their own cast of villains and thieves. A smaller-scale hero can deal with an annoying thief.
We solve the Thief Problem by making more Heroes.
Counterpoint: I’ve mentioned before that “superhero fatigue” is an expression of hostility towards this and standard Serial Escalation. And introducing more small-scale heroes can mean making more Thieves for them to battle.
​
Counterpoint: Introducing more small-scale heroes can mean making more Thieves for them to battle. This is also an unsustainable strategy. See the "Paint the Town Blue" DLC.
​
Strategy 7: “Vigilante Cooperation”
A variation of Cast Bloat.
Some settings involve a formal policing organization (e.g. a police department) or a formal emergency response organization (e.g. a fire department or paramedics). This organization can work with the Hero, assisting many everyday Civilians in Peril and battling some Thieves.
We solve the Thief Problem by limiting Civilians in Peril in the first place.
Counterpoint: Cooperation between superheroes and police, or outright replacement of law enforcement with superheroes, brings cultural baggage about how your audience currently thinks about law enforcement. Be mindful about the subtext you may be introducing about this real-life analogue.
​
Counterpoint: This creates a new character who has a similar role to the Thief and just happens to have a closer moral alignment to the Hero. This doesn’t solve the Thief Problem - it just makes a new, weaker (for the Hero) instance of it. See the “Paint the Town Blue DLC” for further in-depth discussion of this strategy.
​
Strategy 8: “Kleptomania”
Especially in game contexts, writers can encourage player exploration and discovery through the Environment. The player has more choice about when to enter Peril at all. There are other side missions and collectibles and achievements outside of Peril. The player deliberately puts Safety on hold and searches for the Thief at their leisure.
​
Another version of this is making Safety the only win condition - there are Civilians in Peril and possibly other environmental hazards, but no Villains or Thieves.
​
We solve the Thief Problem by making the Environment more interactive in ways that do not include the Thief. We deprioritize the Thief.
​
Counterpoint: In non-game contexts, this can turn into dangerous amounts of filler. If a player is railroaded into side content away from Peril, they can get annoyed. Drama in Peril that turns into slice-of-life Safety can also get boring because of lowered stakes. This means “Kleptomania” and “Serial Escalation” are opposing strategies.
Counterpoint: This is an extrinsic solution like “Fridging” - the writers have not solved the Thief Problem within the plot. This strategy’s viability changes fundamental aspects of the model.
​
Strategy 9: “Spoiled for Choice”
A variation of Kleptomania.
In a population-dense Environment, at some level of increasing population in active play, it is not practical for a Thief to achieve Loot.
We solve the Thief Problem because there are simply too many Civilians to steal from and not enough time.
Counterpoint: This can also make Safety and Takeover impractical for the same reasons. We do want a character to win at some point. For the purposes of this model, the strategy is both extrinsic and self-defeating. This means it breaks Core Assumptions 5 and 6.
​
Strategy 10: “The Practice Dummy”
A variation on The Sky is Falling, Cast Crunch, and Kleptomania.
Safety incentivizes - but does not require - Battling the Villain. Takeover incentivizes - but does not require - Battling the Hero. A valid Hero (or Villain) strategy is to target the Thief and consistently remove them from active play [from the clip start to about 14:19] while benefiting from their Sneakiness farming. if either the Hero is strong enough that they can easily win a Battle (or that neither the Villain nor the Thief can reasonably get as strong as the Hero).
We solve the Thief Problem by bum-rushing the Thief and removing them from active play - or at least minimizing their impact - by force as much as possible.
Because this strategy is a variation of previous strategies, it also inherits their problems.
Counterpoint: Bum-rushing the Thief and then trying to recover active play with direct competition between the Hero and the Villain is unsportsmanlike conduct. This is a similar criticism of “The Bohemian Scandal”.
Counterpoint: The strategy doesn’t get the Hero or Villain closer to their own objectives. This is a similar criticism of “Fridging” and “The Sky is Falling” based on Conclusion 6.
Counterpoint: The strategy adds other intrinsic objectives to active play. This is a similar criticism of “Kleptomania”. “The Practice Dummy” is a cathartic solution, but it doesn’t help advance active play - it is a circular solution that reinforces the Thief Problem. If this were a real game, a developer would not want to incentivize turning a lone player into a practice dummy for the other players’ stats grinding.
​
Strategy 11: “Home Security System”
Place some artificial ceiling on the Thief’s progress so Loot is not possible. This could reject Conclusion 6 by forcing the Thief to interact with the Hero and the Villain. This strategy was developed for the second version of this essay.
Counterpoint: In the context of a game model, this nerfs a player character so they can’t win (see Strategies 2, 8, 9 and 10)
​
Counterpoint: This is the opposite of Strategy 1. It’s as unsportsmanlike for the Thief to give up as it is for the Hero or Villain to give up. The strategy recreates, instead of solves, the problem.
​
Strategy 12: “Super-Speed”
A variation on “Home Security System”. The writer gives a character situation-specific powers to deal with the Thief. When a Thief is so powerful that their appearance in Peril immediately shifts the progress of active play in favor of Loot, or especially if a Cop is approaching Protect and Serve, this strategy attempts to rebalance the weight factors of other characters in response during active play. Instead of “Home Security System” changing the Thief themselves, “Super-Speed” forces other characters to gain new supplementary powers or otherwise cheat to make progress. This strategy was developed for the second version of this essay.
See, for example, the Bleach manga character Yamamoto and his fights, which include, in order: three powerful characters collaborating to reflect Yamamoto’s attacks and heal each other; a weaker henchman villain given a strength boost specifically to distract Yamamoto and seal away the majority of his attacks; and a henchman disguised as Yamamoto’s nemesis who exhausts Yamamoto’s most powerful attacks before the real nemesis arrives for a deadly sabotage.
See, for another example, the Jojo’s Bizarre Adventure manga character Jotaro Kujo. After his run as the protagonist of the Stardust Crusaders story arc, Jotaro has the ability to briefly pause time, leaving his opponents extremely vulnerable. So those opponents use strategies that reward traps, diversion and no-win sacrifice situations that Jotaro can’t prevent.
In the right hands, this can make a Thief character feel overpowered and their opponent feel desperate or even cunning - the opponent has to resort to new, creative methods of dealing with the Thief because they’re otherwise unstoppable. For this example, Yamamoto and Jotaro are some of the strongest characters in their series. The plot has to make character-specific countermeasures just to stop them from winning constantly.
Counterpoint: In the wrong hands, the writer has a difficult time hiding that they’re cheating. The strategy isn’t limited to a super-speed ability but can occur in any ability without reasonable upper bounds. Any ability set that is unbalanced to the point the character has to be nerfed or their opponent has no reasonable strategy but to cheat applies here.
Counterpoint: “Super-Speed” also calls an audible to rebalance a character during active play - which does not explicitly break any of the Core Assumptions. But I would consider it unsportsmanlike for consistency purposes and for this reason would want to avoid the strategy when possible.
​
Strategy 13: “Heroic Sacrifice”
Assume Conclusions 1, 3 and 7 are not true. The Hero will confront the Villain before they are ready (or the Thief will confront the Villain before they are ready) as a stalling distraction while the other farms sufficient Power of Good/Sneakiness to win. This prioritizes Conclusion 5 because the Hero supports Loot and the Thief supports Safety. This strategy was developed for the second version of the essay.
Counterpoint: This creates another “Paint the Town Blue” scenario where the Hero + Thief team has a significant advantage against the Villain.
Counterpoint: This is another possible variation on the Practice Dummy because the Hero + Thief team is engaging in unsportsmanlike targeting.​
​
Strategy 14: “Steal From The Rich, Give To The Poor”
In active play, Sneakiness is a negative weight factor that punishes the Thief the more they interact with Civilians and approach Loot. This is the inverse of Power of Good or Power of Evil, which reward the Hero/Villain the more they interact with Civilians and achieve their class objectives. This strategy was developed for the second version of this essay.
​
Assume Sneakiness is instead a positive weight factor like Power of Good or Power of Evil. This would mean Conclusion 7 is not true and that the Thief is not necessarily punished more for Battling than the Hero or the Villain. Sneakiness is designed this way in active play to incentivize the Thief to be strategic about when and where they interact with Civilians: the player has to work towards Loot as a class objective but gets noticed more and will have to deal with the Hero or Villain more if they’re sloppy. If Sneakiness is a positive weight factor, there is less incentive for the Thief to be more strategic than the Hero or Villain. The character classes are more equally balanced – it is not necessarily easier for the Thief to achieve Loot than for the Hero/Villain to achieve Safety/Takeover. And the Thief is not necessarily incentivized to pit the Hero and Villain against each other as a distraction.
​
Instead of punishing the Thief for getting caught, SFTRGTTP rewards the Thief for not getting caught.
This is a complex solution to the Thief Problem as a game strategy. One example of this could be the mythologized Robin Hood, a thief stealing from the rich royalty but having a good reputation by giving his literal loot to the poor.
​
The Off the Cuff weight factor from “Paint the Town Blue” is another version of this solution with a weight factor that can be either positive or negative.
​
Counterpoint: We have now introduced the possibility of a three-way stalemate that encourages collusion in order to break the stalemate.
​
Counterpoint: This new behavior is less of a “Thief” and more of a “Bounty Hunter” who might deliberately try to stand out and target Civilians that were just moved by the Hero or Villain. The character class’s rebalancing doesn’t change the game player’s self-interest. The Thief Problem is not related to the Thief character’s literal sneakiness.
​
Strategy 15: “Reciprocity”
A variation on Steal From The Rich, Give To The Poor.
In the original game, Battling only punishes the loser of the Battle. “Reciprocity” creates a mechanic to calculate collateral damage for both characters who Battle. When characters fight, they typically both get injured. This strategy was developed for the second version of this essay.
The strategy assumes Conclusion 4 is not true. The Thief is not especially incentivized to be more strategic than other players because now both players are damaged by Battling.
Counterpoint: This is a practical method for rebalancing the gameplay, but the strategy offers less insight into the narrative problem.
​
Strategy 16: “Fish or Cut Bait”
The Villain puts the Thief in a Morton's Fork-style scenario where they have to either collaborate with a teammate or make progress towards their class objective without a teammate. The Thief is not necessarily choosing to be in this situation, like they might with Heroic Sacrifice.
​
For examples:
​
Pucci, the priest from the manga Jojo's Bizarre Adventure: Stone Ocean (1999-2003), forces the heroes to repeatedly choose between sacrificing one of their allies/other long-term goals OR letting Pucci escape to continue his plan. He exploits these pressures to make progress. This invokes the Thief Problem even though Pucci is the main villain of the arc.
Similarly, Thanos from Avengers: Infinity War (2018) invokes the Thief Problem when he forces characters like Loki, Star-Lord, Gamora and Wanda (and then Iron Man in Avengers: Endgame (2019)) to make sacrificial plays choosing between a character's life in one hand and an Infinity Stone in the other hand.
​
Counterpoint: This has Cast Crunch problems if you are serious about letting characters get killed.
​
​Section 5d: Final Note/Conclusion
Unfortunately, most of these proposed solutions have their own problems. I’ve provided counterpoints for them above in turn, but in summary:
-
They change some fundamental assumptions of the model. This usually takes the form of
-
Changing population sizes of character classes - see “Cast Crunch” and “Cast Bloat”;
-
Changing character relationships - see “The Bohemian Scandal” and “Fridging”;
-
Changing class objectives - see “Serial Escalation” and “Kleptomania”;
-
Making class objectives mutually inclusive - see “The Bohemian Scandal” and “Vigilante Cooperation”; or
-
Unsportsmanlike conduct among players like collusion, cheating etc. - see “Spoiled for Choice”, “The Practice Dummy”, the “Paint the Town Blue DLC” and “Super-Speed”
-
-
They require certain dramatic tones (especially violent or even apocalyptic tones) which are not suited to all stories and for that reason lose some applicability - see “Fridging” and “The Sky is Falling”;
-
Alternatively, they require certain non-dramatic tones that can impact a story’s pacing - see “Kleptomania”;
-
The strategies cross-contaminate each other. The use of one strategy often inherits similar complications from similar strategies.
The Thief Problem does not necessarily happen in every active play session. But because the Thief Problem can happen at all, we can also use strategies that can counteract the problem. It works as a heuristic for making sure a plot is character-motivating and character-motivated.
Long-term serialized stories with large casts of characters can be uniquely susceptible to both the Thief Problem and ineffective counter-strategies. The more Thieves in a cast, the more difficult it is to justify why those Thieves are not active in the plot, the more writers need excuses to neutralize them.
Writers may choose to deliberately invoke the Thief Problem as its own narrative choice for reasons that include:
-
To establish stakes, like Luffy out of his depth in the Marineford War and Wano arcs. Luffy's typical headfirst strategies stop working in these situations. But beware Serial Escalation!
-
To force a dramatic choice that reinforces an emotional relationship, like Mav saving Rooster or Hangman's return in Top Gun: Maverick (2022). But beware the Bohemian Scandal and Heroic Sacrifice!
-
To treasure-hunt or bounty-hunt. These plot structures include multiple parties making progress towards the same goal. But beware Steal From The Rich, Give To The Poor!
​
Conceptualizing these tropes as methods for solving a problem (and, using our framing, methods for patching an exploit) means recognizing them as part of the story’s design. Writers can still use these strategies for telling compelling stories, even if they have complications, by solving their problems in turn or finding other creative methods.
The zero-sum nature of the framing device discourages redundant or circular strategies that don’t actually solve the Thief Problem. The ways you solve - or choose not to solve - the Problem become interesting narrative choices. Your story is not bad just because you have to deal with the Thief Problem. This essay explores why the problem happens in the first place and ways to not let it overrun your plot.
In summary: the Thief Problem is a good heuristic for plot coherency, a test to make sure characters use their abilities for maximum drama. Your audience will not be satisfied if you add a Thief to active play and then forget about them - especially if you don’t give a reason to forget them.
Endnote 1
The variables are designated according to the nominal, ordinal, interval and ratio scales of measurement common in social science research. I used these for two main reasons:
-
This allows for a consistent way to systematize features like Peril and the character weight factors without relying on the specific syntax of a programming language. A casual reader can hopefully understand the structure of the essay without prior computer software knowledge.
-
I am familiar with using these scales of measurement. Like I said, I’m not a programmer making a feasible game - I’m a narratologist making an argument about superhero stories.
For the purposes of this essay, I will define each of the applicable scales of measurement here.
Nominal variables have multiple values but not a continuous scale (e.g. on and off).
Interval variables can have multiple values that compare with each other on a given numerical scale (e.g. a population of 2 characters is greater than 1 character but less than 3 characters).
Ratio variables can have multiple values that compare with each other on a given numerical scale, like interval variables, and also have a contextual zero value (e.g. a population of 2 characters is greater than 1 character but less than 3 characters AND it’s possible to have a population of 0 characters).
Appendix: "Paint the Town Blue DLC"
​In short, the Thief Problem is a narrative tendency for a hero character to get sidetracked by a side character. Assuming a superhero/supervillain/thief dynamic, it is easier for the thief to achieve their goals than for the hero to “save the day” or a villain to “take over the world”. So the hero/villain either 1) needs to spend time and resources to stop the side character from achieving their own goals - at the cost of the story’s main plot - or 2) has to be okay with the side character achieving at least some of their goals - along with the moral and narrative consequences that come with that.
Writers often have creative ways to solve or ignore this problem, some of which I overview in the main essay. One of the proposed solutions to the Thief Problem (“Vigilante Cooperation”) encourages the writer to introduce emergency response characters that can work with the Hero to move Civilians in Peril out of Peril. This is usually police but can also be e.g. firefighters, paramedics/doctors, or military personnel, if necessary. Because these institutions also exist in our real world, Vigilante Cooperation is one of the more common Thief Problem solutions.
The specs for this “DLC” explore in greater detail why Vigilante Cooperation can often recreate the problem it’s trying to solve.
The Thief Problem computer game model adds a new downloadable content package, named “Paint The Town Blue”. It includes all of the character classes and settings from the base game. It adds:
-
a higher difficulty setting - all characters are some factor (greater than 1x and less than or equal to 1.25x) faster than in the base game
-
a new location map, “Downtown”, with a different layout and environmental features than the base game. It otherwise makes the same functional assumptions as the original Environment, including the Peril area.
-
a new character class that populates the Downtown location map
The new non-playable character class “Cop” is based on the Hero character class from the base game. However, it is a new character class made just for this DLC, so it also borrows/reinterprets features from the Thief class and adds new custom features.​
Cop character icon
Cop Free Enters Peril and Force Exits Peril, to account for a moral incentive to save as many Civilians as possible. The model counts every instance a Cop and a Civilian interact.
When a Civilian In Peril and a Cop interact, a heads-waited coin is flipped.
-
If the coin lands on heads, the Civilian Force Exits Peril. This accounts for the Cop continuing rescue efforts throughout Peril even after saving a given Civilian.
-
If the coin lands on tails, there is no change.
When a Civilian Not In Peril and a Cop interact, a heads-waited coin is flipped.
-
If the coin lands on heads, there is no change.
-
If the coin lands on tails, the Civilian Force Enters Peril. This accounts for the Cop failing to rescue the Civilian, escalating the situation, or otherwise not saving the Civilian.
Ratio variable “Off the Cuff” tracks and assigns a positive ability weight factor to the number of the Cop’s Civilian Force Exits Peril during active play, to account for a positive reputation as Cops save more Civilians. “Off the Cuff” also tracks and assigns a negative weight factor to the number of the Cop’s Civilian Force Exits Peril during active play, to account for a negative reputation as Cops put more Civilians in Peril. In practice, this would reflect in-game as an interval rational variable that can increase to a maximum of 1 and decrease to a minimum of -1 in real time.
.png)
An example UI dial element tracking
“Off the Cuff”
Cops can Battle the Hero, the Villain, and the Thief. The standard Battle rules apply to the Cop for each character class if the Cop loses.
-
If the Cop and the Hero Battle and the Cop wins, standard penalties apply, except the Hero cannot move in the Environment for ten (10) seconds, and Power of Good is halved.
-
If the Cop and the Villain Battle and the Cop wins, standard penalties apply, except the Villain cannot move in the Environment for twenty (20) seconds, and Power of Evil is halved.
-
If the Cop and the Thief Battle and the Cop wins, standard penalties apply, except the Thief cannot move in the Environment for thirty (30) seconds and the Thief’s Sneakiness doubles, to account for the Thief having to hide from the cops. The more the Thief interacts with other characters, the lower their Sneakiness is because they keep getting caught.
Cops have the class objective “Protect and Serve” to Force Exit as many Civilians in Peril as possible during active play (with a maximum of every Civilian in the Environment).
-
A Cop also achieves Protect and Serve if all Civilians are Not In Peril.
-
If a Cop achieves Protect and Serve, the Cop wins and active play ends.
Cops also have a dormant ability “Right to Remain Silent” that tracks Safety, Power of Good, Takeover, Power of Evil, Sneakiness, Loot and Off the Cuff.
-
If a Hero, Villain or Thief achieves their class objective before the Cop, the Right to Remain Silent activates. If the Cop’s Off the Cuff is higher than the winning character’s weight factor, the Cop wins and the other player character loses. This accounts for institutional power to sway public opinion.
When active play starts, the computer rolls 1d6 to determine the population of Cops in the Environment. This means there are a minimum of 1 cop and a maximum of 6 cops in the Environment.
​
How does “Paint the Town Blue” impact the Thief Problem?
“Paint the Town Blue” assumes Conclusions 4, 5 and 6 are not true. We can solve the Thief Problem by giving a Thief-style character a moral alignment that isn’t chaotic neutral. This character class has a goal that aligns with the Hero’s class objective. The Cop always works towards removing all Civilians from Peril, and the Cop can win if the Hero succeeds. It’s more difficult for the Cop to win if the Villain succeeds, but it’s not impossible.
In the context of the model, there are a variety of ways to achieve this. One simple method is increasing either of the Thief’s Pickpocket outcomes by some additional weight so it’s not a 50-50 chance. One detailed method is the game features in the DLC premise.
This does not mean Cops are good guys. If anything, they’re lawful evil. The Cop has to deal with the triage elements of the Hero (who, from Conclusion 1, can’t save all Civilians all the time) but lose the perks of staying under the radar (from Conclusion 6).
Counterpoint: With all other things staying roughly equal, the baggage from Vigilante Cooperation applies even stronger in this DLC.
It can get difficult for your audience to keep rooting for your hero if they have abundant resources to spend at any problem they come across. That’s not just infinite time, attention or money but also infinite allies - or at least enough allies that the Hero will always have them. So a police force that is in un-nuanced favor of the Hero has to start inventing reasons why the Hero is rescuing Civilians when the Cops can do that - or you start making serial escalations that only the Hero can solve.
This is why Vigilante Cooperation was a solution to the Thief Problem in the first place. The city is overrun with criminals because the Villain has caused chaos. When the Hero goes off to fight the Villain, there is still chaos that someone needs to help mitigate - because the Thief is taking advantage of the chaos to accomplish Loot. Unambiguously pro-Hero Cops are not guaranteed to stop the Villain from farming enough Power of Evil to stay threatening or from achieving Takeover because all Civilians are in Peril.
But if the Cops aren’t in the un-nuanced favor of the Hero, Cops can function the same way a Thief does - just with slightly more prejudice against the Villain and Thief than the Hero. So Cops might reinforce the Hero’s efforts toward “Safety”, or they could just as likely sabotage “Safety”.
​
Counterpoint: Writers typically don’t want Cops to win because this removes the Hero’s agency and choices to act toward Safety. Again, if the Cops can solve all of the plot drama by themselves, why does a Hero need to take action?
​
Counterpoint: It’s intentional that “Off the Cuff” and “Power of Good” are similar weight factors. It’s also intentional that “Safety” and “Protect and Serve” are similar objectives. Even if the Hero is close to winning or even accomplishes Safety, the Right to Remain Silent can sabotage the Hero’s win at any point in active play. The Hero and the Cops are allies towards similar goals. They are not teammates.
e.g. Lieutenant Gordon is an ally of Batman, but Gordon isn’t part of the Bat Family.
Their efforts towards their respective win conditions look inclusive at first glance. But the win conditions are mutually exclusive.