You are here

Gameplay concept

[Gameplay concept] Introduction

AHasvers's picture
Submitted by AHasvers on Sun, 01/04/2015 - 23:48

The topic of this series of devlogs is the very reason for Lilavati's existence: an idea for dialogue as the main course of our gameplay, rather than a mere digestive (or, at best, a delicious entremet). The basic concept - showing the dialogue tree instead of hiding it, and turning it into an actual strategy game - is simple enough and has been approached from various directions by various other games, yet none seem to have gone all the way. But hey, that means we have a niche, which we are glad to share if you feel inspired.


Y'see, we like stories about clever people, from Paul Atreides to Hannibal Lecter, who always come out on top by the sheer silverness of their tongue. The grand orators, the cold readers, the Trickster Gods, you get the idea.

We also like stories about human relationships evolving through complex conversational interplay, from Tom Stoppard to My Dinner with Andre, The Sunset Limited and all those European movies where people just. won't. stop. talking.

It makes sense that we would like to play those sorts of scenes or stories. There is so much going on - rivalry and empathy, misdirection and seduction, careful risk-taking and planning, glorious victories of logic over emotion or the opposite - that the possibilities seem at least as vast as, say, military strategy, city building, dungeon looting or cow clicking.

Unfortunately, almost all that we have now, even in the very best-written and most engaging of games, is multiple choice tests and skill checks. For all that I love Planescape: Torment and Arcanum, there is something unsatisfying in the sheer gap of player agency and abilities between fighting situations and diplomacy.

Which is not to say that nothing has been tried. Most of the games I will mention below are very good in their own right. (Most of them.) They are simply not scratching this specific itch.

[Gameplay concept] First Problem

AHasvers's picture
Submitted by AHasvers on Tue, 01/06/2015 - 03:49

First problem: the guessing game

In some cases, dialogue trees can be really complex (indeed, more conditional graphs than trees) and exploring them becomes a real puzzle, see e.g. Galatea by Emily Short.

In other cases, as in Deus Ex: Human Revolution, you can see some trappings of a dialog-specific gameplay. But it is still largely a mix of skill checks and guess-the-right-answer with a magnificent paint job.

What you lack in all those cases is simple: the ability to plan. To take the lead, to act rather than react, and make decisions not for their immediate reward, but in preparation for other options later on. A rule system that allows you to learn patterns and evaluate what the effect of your choice will be, not just vaguely infer it from situaton-specific hints put in by the writers.

In the very best case, you approximate this by replaying the same conversation again and again until you get the gist of it, but that's level memorization, not skill acquisition.

(Disclaimer: this is something our own contest entry Exeunt Omnes was entirely guilty of, for lack of time. I am not proud.)

These games are more akin to puzzles in a point'n'click adventure than to battle systems, or strategy and simulation games. Which is fine, but not quite what we are looking for here.

Other examples: the insult swordfighting in Monkey Island, any conversation in Interactive Fiction, the Phoenix Wright series and the entertaining Socrates Jones: Pro Philosopher, or Goblin Noir (despite its appearance of having a battle system, the battles are really puzzles).

[Gameplay concept] Second Problem

AHasvers's picture
Submitted by AHasvers on Tue, 01/06/2015 - 03:50

Second problem: the lack of semantics

Some games take the opposite route: they have gameplay sequences that represent conversation, but that gameplay has no conversational content, or the connection is tenuous at best. You play the role of someone who talks, but no actual exchange that you can read comes out of it. (In other words: ludonarrative dissonance)

This can be, for example, a typical RPG battle system where the actions are given discourse-related names, although what they do is the usual attack/heal/defend, as in The Logomancer. In the best cases, the author's intent may have been to capture only some aspect of the conversation which lends itself better to such abstraction, like tone in Last Word.

There is also the persuasion minigame in Oblivion, and QTEs in the middle of cutscenes in various games, where you have semantic content interspersed with some sort of action, but the action itself - what the player ends up doing - is really disconnected from the content.

Other examples: Super FreudBot, Dangerous High School Girls in Trouble (social interactions as parlour games).

A related minority case: games that would ideally be conversations with an AI, which will be very exciting the day that AIs have something interesting to say. Facade falls more on this end of the spectrum, although it lies somewhat in between the two classes.

A stronger example is Argument Champion: it has some semantics, which is quite impressive as it is not at all pre-written but based on semantic analysis (MIT's ConceptNet). It remains a little silly - trying to argue that my audience should like navigation because they like Africa does not feel quite like being Cicero. But it is a step in the right direction, and probably the best that can achieved without a human writer.

[Gameplay concept] Where to look for a solution

AHasvers's picture
Submitted by AHasvers on Tue, 01/06/2015 - 03:51

Where to look for a solution

The first problem is simply one of building a system rather than designing a series of puzzles and clues. The second is a problem of gameplay-narrative divide. We want there to be an actual content to the conversation, and that requires it to be written down in advance, because AI just isn't good enough. The trick is to find rules that fit the content.

Thankfully, all the real work has been done for us. By rhetorics, conversation analysis, politeness theory and various incarnations of interactionism. These people have recorded, studied, classified the already highly game-like interaction that is argumentation and conversation.

As it turns out, it has a clear set of mechanics:

  • a few modes of action, such as the classical ethos (authority), pathos (emotion) and logos (logic) [Aristotle on the SEP], or in a more, er, free-spirited interpretation, Revelation, Imagination and Explanation.
  • a few distinct resources: the positive and negative face [Brown and Levinson], the mental and affective state of the speakers and listeners, their knowledge of each other's beliefs,
  • codified rules for the interaction: politeness codes, turn-taking [Hirsch], conversational maxims [Grice].

Sure, we will have to leave out a lot of the meat of these theories as we turn them into game rules. But at least, we have a battle plan: write a map containing many possible ideas and arguments relating to a central topic, which will be the territory on which the conversation is "fought", and then implement mechanics for all three aspects above.

In brief, the map will be the semantic content, and what we do on it will have rules, enabling control and strategy.

[Gameplay concept] Logos

AHasvers's picture
Submitted by AHasvers on Wed, 01/28/2015 - 15:21

Logos is the core of the gameplay: it is the web of topics and arguments on which the conversation is played.

Topics are nodes, connected by links called Arguments. Here, characters in the game differ primarily by: which topics and arguments they know at the start of the conversation, and what are their opinions on various topics.

Topics, Biases and Arguments

Each topic is a node on which characters can have a positive, negative or neutral Opinion. This opinion can represent a judgement of

  • truth ("We should invade Bhutan": true or false),

  • subjective value ("Roses are beautiful": agree or disagree),

  • or relevance to the conversation ("Quine argues against Kant": we care or not really)

Arguments are links connecting an opinion on one topic to an opinion on another: if "Peace is better than war" is true, then "We should invade Bhutan" is false.

Any character's opinion is the product of the topics and arguments they know, and a few initial biases on certain key topics. Without these biases, everyone would have the same neutral opinion on everything.

The crucial idea here is limiting biases to a few topics, and making all other opinions the result of cascading logical links spreading out from these initial biases. This allows to change someone's opinion of almost anything, by identifying the roots of their belief and giving counter-arguments surgically where they have the most impact.

It also means that, provided you know someone's biases, you can generally guess their opinion on a new topic, which allows for strategic planning.

Link genres and fallacies

So far, we have assumed that all characters agree on arguments (link between two topics), and the only differences lie in their initial biases and knowledge of those links.

It seems that the web of arguments is absolute and static. But in real conversation, a lot of effort is devoted to contesting not the premises of an argument, but the argument itself. Some links may be erroneous: they could represent fallacious reasoning such as overgeneralization, ad hominem, and so on. We may even want the player and opponnents to be able to make up some fallacies on the spot, to convince their unsuspecting audience of something they do not believe themselves.

The problem of fallacious arguments is complex: we don't want people to be able to refute "real" logical links, but if we allow refuting only fallacious links, then it is obvious to the player which is which. Even if their character is fooled, the player will know exactly where to attack an opponent's argumentation.

The solution requires revising what we think of as logic: logical rules are not self-evident, they are learned. You cannot use the rules of logic to prove themselves, they are not a justifiable knowledge. They are simply a cognitive know-how - for instance, you learn an example of syllogism, and practice rearranging other arguments to fit this form, until the rule is so ingrained that you believe them to be true because logical [H.M. Collins Changing Order (1985), see also McFarlane]. Therefore, there is no such thing as absolutely true logical links just awaiting to be discovered: the web will be different depending on the reasoning rules that you have learned.

In gameplay terms:

  • Arguments have genres: Syllogism, Analogy, Excluded middle, Bayesian inference...
  • These genres are learned, most characters know most of them but no character knows them all.
  • Links of a genre your character does not know are shown with "Unknown genre" and you can dispute them.
  • Fallacious links made up on the spot have "Unknown genre" for everyone (including their creator!)
  • Links successfully revealed to be fallacious cause their author to lose a huge amount of Face* (and trust from everyone present), but a successful defense under contestation causes the attacker to lose Face instead.

Therefore, if you see a link of an unknown genre, you can try to dispute them, but you are taking a risk: is it a fallacy, or simply a valid argument model that you do not know?

* See Ethos below.

[Gameplay concept] Ethos

AHasvers's picture
Submitted by AHasvers on Wed, 01/28/2015 - 15:21

Ethos is originally meant to represent the trustworthiness of the orator - how willing the audience is to believe them on principle. It is the core of arguments from authority - which are considered a fallacy by modern cognitive psychology, but are an absolute necessity in real life: you cannot go on checking every single fact by yourself before accepting what someone else says. You will generally trust that this quantum chemist has an informed opinion on atomic orbitals. The fallacy is placing argument by authority above experience... and then again, not always - there are a lot of experiences you just cannot achieve by yourself without training.

Here we mix it up with modern Face negotiation theory [Toomey, Brown and Levinson] to elaborate the representation of each character's standing in the conversation. This is the resource management part of the gameplay.

Each character is given a set of variables:

  • Face: authority, reputation (as in losing and saving face), the "mask"
  • Territory: confidence, emotional stability, cognitive and material resources
  • Affinity: for each other character, how much they trust them due to shared history. This is generally not symmetrical: A can like and trust B much more than the reverse.

They intervene in multiple aspects of the gameplay:

  • Certain topics or arguments may come with a bonus or cost in any of these resources:
    • a subtle argument will increase your Face,
    • a criticism will decrease the target's Face,
    • revealing an intimate secret will cost you Territory.
    • anything that causes someone to lose some Face or Territory will also decrease their Affinity for the speaker.
  • When any resource is too low, a character may be driven to leave the conversation
    • This is generally to be avoided: even if you have a good argument, refrain from using it if it may hurt your audience's feelings.
    • This can be a goal in a debate with an audience; lowering the Face of the opponent untl they leave in humiliation.


Characters also have some actions at their disposal that directly affect those resources: this is generally known in scientific literature as politeness, facework and face reparation.

When using a topic or argument would cause the loss of some resource (I will call this a "threatening act"), this loss can be compensated by additional actions:

  • Apologies sacrifice a little of the speaker's Face to restore some of the listener's
  • Self-abasement also sacrifices Face to draw the audience's empathy and Affinity
  • Promises and contracts sacrifice some Territory for someone else's benefit
  • Euphemisms remove the edge of an attack but take longer to enounce

By using these actions at the same time as the original threatening act, the player can avoid any additional effect (such as a decrease in Affinity) that the losses would have triggered.

Ethos positions

Finally, "ethos positions" have additional effects depending on the relative Face of two characters. A good orator knows when to stand above their audience, treat them as peers, or play a subservient or subversive role.

  • Dominant face: A speaker with significantly higher Face will tend to be believed on principle on topics on which the listeners have no prior opinion (instead of them starting in a neutral opinion).
  • Equal face: When two characters have approximately the same standing, anything that affects one affects the other a little - any bonus or malus is shared in a small fraction, which encourages collaboration and causes equals to unite when one of them is attacked.
  • Dominated face: A speaker can use an inferior position for their own benefit, as inferiors' defects are more easily forgiven and a successful attack from an inferior is devastating (aka the Diogenes method). In other words: when one has significantly lower Face, all resource costs to self are decreased, and all damages on opponents' resources are increased.

[Gameplay concept] Pathos

AHasvers's picture
Submitted by AHasvers on Wed, 01/28/2015 - 15:22

Pathos is the use of affect to persuade your audience to take a certain stance. It can work by appealing

  • to emotion ("Think of the children!")
  • to ego ("Any intelligent person knows that the result in 47." or in implication "Obviously, the result is 47. ")

As such, it may seem antithetical to Logos - it seems to be there to override reason and create arbitrary biases in people (even if you deem like me, perhaps after reading the previous posts, that logic is not a God-given web of intrinsically true relationships).

This creates multiple implementation problems:

  • If we decide that pathos is some power that allows a speaker to shift belief in a given claim, then we need some mechanism to shift it back, lest these effects be permanent.
  • Furthermore, it is very aggressive: contrary to our maxim that "dialogue is both cooperative and competitive", pathos seems to exist only as a way of manipulating people into false beliefs.
  • Furthermore, people in real life use pathos on themselves all the time (that's the very essence of self-help) which would ha veno benefit here.
  • Finally, we need to add a mechanism for resisting pathos: how does a character keep their cool and see through this aggressive.

All these problems are solved by a simple choice: pathos does not affect your belief, it creates an incentive for believing.

What is affect? It is investing emotional resources - i.e. Territory - into a given position. You are rewarded if you feel justified in that position, and punished by cognitive dissonance if you feel opposition to it, which pushes you to ignore arguments against your side.

In gameplay terms:

  • Pathos power is collected by claiming certain topics that have emotional resonance (e.g. "Many children are dying of hunger")
  • This power can be used to induce the speaker, the audience, or both into investing Territory in a given opinion on another topic that is connected to the first one (e.g. "We should produce more food").
  • Invested Territory is a double-edged sword:
    • when the opinion is corroborated by the general state of the conversation (when you and your audience agree that it is true), your Territory reserve is increased by the invested amount. You feel justified and confident.
    • when the opinion is opposed (when at least one of the characters disagrees), your Territory reserve is depleted by the invested amount. You feel wronged and on shaky ground.

Thus, even if you see logical arguments against that position, you will be driven to ignore them because acknowledging them would cost you. Resistance to pathos comes simply from having a large enough starting reserve of Territory, so that losing or gaining a little more is not important - which fits with the notion that Territory represents emotional stability and therefore resistance to manipulation.

Finally, by collecting pathos and sending it to other topics along fallacious links - arguments that you create on the spot - you can manipulate your audience into believing essentially anything, but your entire strategy may crumble if they see through you.