Dev Diary #34 - Operations 2.0 💥
What's happening / TLDR: Developer diaries introduce details of Espiocracy - Cold War strategy game in which you play as an intelligence agency. You can catch up with the most important dev diary (The Vision) and find out more on Steam page.
---
This is probably the most frequently asked question about Espiocracy: is the game more about espionage or more about the Cold War?
Although the or is not exclusive, this question often is asked in a less suggestive way, lined with the fear that it's "just a spy game" instead of a sufficiently complex Cold War grand strategy game. My answer - which did evolve over almost two years of development - is usually along the lines of: it's a game about the Cold War from the perspective of espionage. This point of view, instead of subverting gameplay, enriches it in many fantastic ways and is in my opinion critical step in actually capturing the essence of the Cold War ([1], [2], [3]...).
However, deep down in my heart, I'm fine with Espiocracy being (viewed as) just a spy game. The very best craftsmanship is dedicated to espionage mechanics - to make them stand out on their own, to provide strategic and immersive gameplay like no other game. In today's dev diary, we'll return to that part of the game, exploring the core mechanic of espionage: intelligence operations.
(As always, a "2.0" dev diary is not a patch note, you don't have to be familiar with the previous one. Enjoy!)
[h2]The Big Picture[/h2]
In the full UX flowchart of the game, operations occupy large and central area:
There are currently 8 (!) entry points leading into this world - eight distinct ways to launch an operation, from selecting actors to approving recommendations. By design, it follows a hierarchy of attention: the player can designate just a general operation category (eg. "eliminate X, I don't care how"), choose a particular operation type (eg. "murder would cause too much fuss, X should be expelled"), or go into nitty gritty of tactical approaches (eg. "burn agent Y to distance ourselves from the manufactured evidence").
In the spirit of Wittgenstein, operations in the game are defined by shared properties, not only inside categories but also in general. All operations have a target - influential actors, or any human, organization, or object (which means for instance that you can assassinate a witness, and then silence a witness of the previous witness' assassination). Among other universal properties of operations, all of them use prospective outcome which is a 0-10 score summarizing complex factors in one easily understood number. In the previous dev diary, it was tied to probabilities but this is no longer the case (more on that in the next section), as the current prospective outcome is more of a guidance that is both descriptive (takes into account factors that may become important in the simulation) and causal (influences paths taken by the simulation, especially the final attempt at achieving the objective). Here's the combination of targets, scores, categories, and types in the wild:
Once an operation is launched, it can be handled autonomously by operatives but it (again, by the same design principle) oozes with optional player agency. Available decisions include:
As Mark Rosewater, a veteran designer of Magic the Gathering said: "Be more afraid of boring your players than challenging them!".
[h2]Simulations Upon Simulations[/h2]
Hairy dev diary about simulations already spilled the beans about the new shape of operations. To recap in a more literate manner, every operation spawns its own simulation, characteristic for the category and type, which proceeds step-by-step towards the final attempt at achieving the objective, which is usually resolved by at least one other (sub-)simulation. For instance, a recruitment operation progresses day-by-day through preparation, intelligence gathering, preceding meetings, all the way to the final recruitment pitch which is simulated minute-by-minute (beware, radioactive work-in-progress interface):
Practically, the availability and quality of the operational culmination depend on parameters such as tactical intelligence or trust, so operatives increase them through continuous (eg. surveillance) and discrete (eg. breaking in) actions.
That's it. This simple idea, however, contains the entire universe of emergent simulations - hundreds of possible steps and events, actions depending on anything from operative's traits to carried gear, conversation simulation flowing into car chase simulation flowing into shootout simulation, operation A launching operation B, involvement and interactions between many types of participants (journalists, police officers, or even third-party actors)...
[h2]A Game of Information[/h2]
Espionage is about information - who's who, what's happening, why it happens. Espiocracy tries to capture this angle in many mechanics (such as secrets or coup plots) but one of the most, I have to use this word, brazen implementations lies at the heart of operations.
Sticking with the word "participant" introduced in the previous section (unofficial term, that's just how these variables are called in the code), classic operations have three participants: attacking intelligence agency, target, and defending (counterintelligence) agency. However, participants may not know about each other, identity, motivations, or even about particular actions!
Each participant has the parameter of situational intelligence, increased via actions much like other parameters. Zero means no knowledge of the operation - targets and counterintelligence start with 0. First suspicions or rumors slightly increase it, then passive and active probing, acquisition of evidence or even direct contacts raise it further. As with any other parameter, it can be also changed externally in both directions, for instance leaving a false trace may decrease the intelligence of other participants, whereas poor tradecraft or reckless behavior might up it for others.
The strength of the cover and analogous factors determine how much situational intelligence is required to uncover a particular participant's what, who, and how:
Furthermore, every action has an inherent requirement for situational intelligence, with half of this number required to see the existence of (blacked out) action at all, and values below that hiding completely the action.
Now, on to much-needed practical examples. Three primary perspectives:
For the sake of readability, I omitted active covers in the examples up to the last sentence. Active cover is false what-who-how which overrides required levels of situational intelligence. It's an expensive approach to an operation, where a participant (usually an attacking agency) can hide under the cover of another why, another who, or even another what such as the press, with the entourage of false actions, false objectives, and decisions (eg. change objective for a false-cover-agency while pursuing another real objective).
[h2]Behind The Scenes[/h2]
â–º At some point in development, operations were also launchable against any process in the game world but it was a tad too abstract, especially with the activity and what-who-how of participants. The UX option remains ("fund a coup") but mechanically it's always about entities engaged in the process, in this case, funding coup plotters.
â–º Among other tested & shelved ideas, I temporarily implemented the use of situational intelligence even for operations launched by the player. Instead of inherently knowing what our operatives did, they would first have to communicate that to the player - with the ability to manipulate the communication (a.k.a. lie). This was mainly tested in the context of conversations, where contents would be generally hidden from the player, except for the parts revealed by the operative during debriefing. It turned out to be too much of a balancing & debugging pain in comparison to gameplay gain, at least for now - and redundant to better mechanics that handle operatives going rogue.
â–º "Infiltration" is not the best word to describe break-ins or wiretaps but alternatives ("penetration", "intrusion") work even worse as nouns and verbs in the UI...
â–º Operations, in a way, are at the forefront of Espiocracy's game design. They are prone to difficult bugs, degenerate strategies, poor UX, or beautifully collapsing simulations. In some cases they resemble a Rubik's cube - the entire game has to be slightly shifted to solve operational issues. As an example of such a case, early prototypes evoked strong fear of missing out on opportunities to launch interesting operations. This was mitigated not only by the improvement of UX and introducing new features (such as, literally, "opportunity" mechanic) but also by shifting code architecture to allow efficient & automatic & regular calculations for all possible prospective outcomes, both for the human player and for the AI. This in turn, initially created an obvious degenerate strategy (regularly launch all the ops with top scores), so new features and shifts followed here. Then, this strategic complexity proved to be challenging for a standard approach to AI, so I started working on an unusual AI system that can handle such operations. Then... you get the picture.
[h2]Final Remarks[/h2]
Operations will definitely receive "3.0" DD in the future, not only because I'm constantly working on them but also because I still didn't mention quite a few interesting angles (such as operatives traveling on the map, operations with evolving objectives, or post-operational fallout).
If you're not already wishlisting Espiocracy, consider doing it:
https://store.steampowered.com/app/1670650/Espiocracy/
There is also a small community around Espiocracy:
---
"Challenge between intelligent people on both sides" - Aleksander Makowski, a spy who tracked UBL, on what makes intelligence operations thrilling
---
This is probably the most frequently asked question about Espiocracy: is the game more about espionage or more about the Cold War?
Although the or is not exclusive, this question often is asked in a less suggestive way, lined with the fear that it's "just a spy game" instead of a sufficiently complex Cold War grand strategy game. My answer - which did evolve over almost two years of development - is usually along the lines of: it's a game about the Cold War from the perspective of espionage. This point of view, instead of subverting gameplay, enriches it in many fantastic ways and is in my opinion critical step in actually capturing the essence of the Cold War ([1], [2], [3]...).
However, deep down in my heart, I'm fine with Espiocracy being (viewed as) just a spy game. The very best craftsmanship is dedicated to espionage mechanics - to make them stand out on their own, to provide strategic and immersive gameplay like no other game. In today's dev diary, we'll return to that part of the game, exploring the core mechanic of espionage: intelligence operations.
(As always, a "2.0" dev diary is not a patch note, you don't have to be familiar with the previous one. Enjoy!)
[h2]The Big Picture[/h2]
In the full UX flowchart of the game, operations occupy large and central area:
There are currently 8 (!) entry points leading into this world - eight distinct ways to launch an operation, from selecting actors to approving recommendations. By design, it follows a hierarchy of attention: the player can designate just a general operation category (eg. "eliminate X, I don't care how"), choose a particular operation type (eg. "murder would cause too much fuss, X should be expelled"), or go into nitty gritty of tactical approaches (eg. "burn agent Y to distance ourselves from the manufactured evidence").
In the spirit of Wittgenstein, operations in the game are defined by shared properties, not only inside categories but also in general. All operations have a target - influential actors, or any human, organization, or object (which means for instance that you can assassinate a witness, and then silence a witness of the previous witness' assassination). Among other universal properties of operations, all of them use prospective outcome which is a 0-10 score summarizing complex factors in one easily understood number. In the previous dev diary, it was tied to probabilities but this is no longer the case (more on that in the next section), as the current prospective outcome is more of a guidance that is both descriptive (takes into account factors that may become important in the simulation) and causal (influences paths taken by the simulation, especially the final attempt at achieving the objective). Here's the combination of targets, scores, categories, and types in the wild:
Once an operation is launched, it can be handled autonomously by operatives but it (again, by the same design principle) oozes with optional player agency. Available decisions include:
- Meaningful temporal dynamics - operational pause means staying low for some time and losing the tail, whereas aborting the operation may require a daring escape
- Reactions - operatives asking the player to choose one of the few approaches, options, spend resources, or resolve issues
- Calls (example below, note that they aren't fixed and evolve during operations) - modifying the operation, covers, priorities, or choosing A over B in trade-offs
As Mark Rosewater, a veteran designer of Magic the Gathering said: "Be more afraid of boring your players than challenging them!".
[h2]Simulations Upon Simulations[/h2]
Hairy dev diary about simulations already spilled the beans about the new shape of operations. To recap in a more literate manner, every operation spawns its own simulation, characteristic for the category and type, which proceeds step-by-step towards the final attempt at achieving the objective, which is usually resolved by at least one other (sub-)simulation. For instance, a recruitment operation progresses day-by-day through preparation, intelligence gathering, preceding meetings, all the way to the final recruitment pitch which is simulated minute-by-minute (beware, radioactive work-in-progress interface):
Practically, the availability and quality of the operational culmination depend on parameters such as tactical intelligence or trust, so operatives increase them through continuous (eg. surveillance) and discrete (eg. breaking in) actions.
That's it. This simple idea, however, contains the entire universe of emergent simulations - hundreds of possible steps and events, actions depending on anything from operative's traits to carried gear, conversation simulation flowing into car chase simulation flowing into shootout simulation, operation A launching operation B, involvement and interactions between many types of participants (journalists, police officers, or even third-party actors)...
[h2]A Game of Information[/h2]
Espionage is about information - who's who, what's happening, why it happens. Espiocracy tries to capture this angle in many mechanics (such as secrets or coup plots) but one of the most, I have to use this word, brazen implementations lies at the heart of operations.
Sticking with the word "participant" introduced in the previous section (unofficial term, that's just how these variables are called in the code), classic operations have three participants: attacking intelligence agency, target, and defending (counterintelligence) agency. However, participants may not know about each other, identity, motivations, or even about particular actions!
Each participant has the parameter of situational intelligence, increased via actions much like other parameters. Zero means no knowledge of the operation - targets and counterintelligence start with 0. First suspicions or rumors slightly increase it, then passive and active probing, acquisition of evidence or even direct contacts raise it further. As with any other parameter, it can be also changed externally in both directions, for instance leaving a false trace may decrease the intelligence of other participants, whereas poor tradecraft or reckless behavior might up it for others.
The strength of the cover and analogous factors determine how much situational intelligence is required to uncover a particular participant's what, who, and how:
- What - knowledge about the existence of a participant of a particular type
- Who - identification of the participant
- How - methods, objectives, and progress toward them
Furthermore, every action has an inherent requirement for situational intelligence, with half of this number required to see the existence of (blacked out) action at all, and values below that hiding completely the action.
Now, on to much-needed practical examples. Three primary perspectives:
- As an attacking agency (eg. you try to recruit the target), you start with minimal but steadily increasing situational intelligence. Logically, you don't need situational intelligence to know the details of your operation but it will be useful in spotting (and reacting to) the actions of the target. Moreover, as the counterintelligence service starts sniffing, the higher the situational intelligence, the faster you'll know they're on to you. Although the "who" level here is irrelevant (usually a country has a single CI agency), it's critical to get to the "how" level and get a look into their progress (race with time on many levels) and objectives (eg. huge difference between interception that may just lead to silent expulsion vs ambush with possible casualties, arrests, long-term loss of operatives, diplomatic scandal, etc). Moreover, the full what-who-how path applies to other participants, for instance, press or police forces getting involved.
- As a counterintelligence agency (eg. someone tries to recruit an actor in your country), you start with no knowledge of the operation. Usually, in the course of regular surveillance, you get a wind that something is brewing - an operation category, a target, or an agency, depending on calculated required levels of situational intelligence, which is communicated in appropriate notification. Then, counterintelligence operatives pursuing leads increase the situational intelligence to uncover any of the following: target's what (eg. a political leader), who (eg. actor X), and how (their stance and actions); attacking agency's what (eg. from country Y), who (eg. agency X; may be irrelevant), and how (their objectives, methods, progress, actions); other participants. All of these also directly contribute to available decisions and methods, sometimes like dominoes, for instance discovering the involvement of the press allows the player to pressure them into revealing situational intelligence collected by them which in turn reveals details on other participants.
- As a target (eg. someone tries to recruit your operative; the targeted player also controls counterintelligence so you can pull off an operational game with two entities simultaneously), you also start with no knowledge about the operation. Generally, targets remain more or less clueless until the final approach (or until a major slip-up of the attacking agency), during which situational intelligence is immediately boosted above what-who-how levels, with the exception of active covers.
For the sake of readability, I omitted active covers in the examples up to the last sentence. Active cover is false what-who-how which overrides required levels of situational intelligence. It's an expensive approach to an operation, where a participant (usually an attacking agency) can hide under the cover of another why, another who, or even another what such as the press, with the entourage of false actions, false objectives, and decisions (eg. change objective for a false-cover-agency while pursuing another real objective).
[h2]Behind The Scenes[/h2]
â–º At some point in development, operations were also launchable against any process in the game world but it was a tad too abstract, especially with the activity and what-who-how of participants. The UX option remains ("fund a coup") but mechanically it's always about entities engaged in the process, in this case, funding coup plotters.
â–º Among other tested & shelved ideas, I temporarily implemented the use of situational intelligence even for operations launched by the player. Instead of inherently knowing what our operatives did, they would first have to communicate that to the player - with the ability to manipulate the communication (a.k.a. lie). This was mainly tested in the context of conversations, where contents would be generally hidden from the player, except for the parts revealed by the operative during debriefing. It turned out to be too much of a balancing & debugging pain in comparison to gameplay gain, at least for now - and redundant to better mechanics that handle operatives going rogue.
â–º "Infiltration" is not the best word to describe break-ins or wiretaps but alternatives ("penetration", "intrusion") work even worse as nouns and verbs in the UI...
â–º Operations, in a way, are at the forefront of Espiocracy's game design. They are prone to difficult bugs, degenerate strategies, poor UX, or beautifully collapsing simulations. In some cases they resemble a Rubik's cube - the entire game has to be slightly shifted to solve operational issues. As an example of such a case, early prototypes evoked strong fear of missing out on opportunities to launch interesting operations. This was mitigated not only by the improvement of UX and introducing new features (such as, literally, "opportunity" mechanic) but also by shifting code architecture to allow efficient & automatic & regular calculations for all possible prospective outcomes, both for the human player and for the AI. This in turn, initially created an obvious degenerate strategy (regularly launch all the ops with top scores), so new features and shifts followed here. Then, this strategic complexity proved to be challenging for a standard approach to AI, so I started working on an unusual AI system that can handle such operations. Then... you get the picture.
[h2]Final Remarks[/h2]
Operations will definitely receive "3.0" DD in the future, not only because I'm constantly working on them but also because I still didn't mention quite a few interesting angles (such as operatives traveling on the map, operations with evolving objectives, or post-operational fallout).
If you're not already wishlisting Espiocracy, consider doing it:
https://store.steampowered.com/app/1670650/Espiocracy/
There is also a small community around Espiocracy:
---
"Challenge between intelligent people on both sides" - Aleksander Makowski, a spy who tracked UBL, on what makes intelligence operations thrilling