ArticlePDF Available

Abstract

The modal logic was made to be a tool aimed to assess fundamental linguistic concepts, in order to make evident the underlying assumption of language (any language). This tool is also very useful to apply the same depth of analysis that we can find in semiotic and linguistic into ludology and game studies. This essay will cover the foundational of modal logic applied to games, and the general conclusions that it allows to derive. Moving from semantics, we can show that any game inherently define a sheaf of possible game worlds, and because the translation principle isn't guaranteed in some game (and, more specifically, is never given in specific categories of games-like RPG) we are able to built a distinction between "formal games" and "modal games" (that we will call more exactly procedural and behavioural games respectively). Once built, this theory (that we also called "Fictionally Distributed logic" or [fd]logic) runs across surprisingly conjectures and results, allowing to embrace the infinite variety that games possess, all that without losing its solid theoretical foundation.
A Formal Study of Games
Federico Danelli
Abstract
The modal logic was made to be a tool aimed to assess fundamental linguistic concepts,
in order to make evident the underlying assumption of language (any language). This tool
is also very useful to apply the same depth of analysis that we can find in semiotic and
linguistic into ludology and game studies.
This essay will cover the foundational of modal logic applied to games, and the general
conclusions that it allows to derive. Moving from semantics, we can show that any game
inherently define a sheaf of possible game worlds, and because the translation principle
isn’t guaranteed in some game (and, more specifically, is never given in specific
categories of games - like RPG) we are able to built a distinction between “formal games”
and “modal games” (that we will call more exactly procedural and behavioural games
respectively).
Once built, this theory (that we also called “Fictionally Distributed logic” or [fd]logic) runs
across surprisingly conjectures and results, allowing to embrace the infinite variety that
games possess, all that without losing its solid theoretical foundation.
Prefaction to the 4th edition
Originally (2013), this was a little more than a rant (in Italian), towards some game
concepts (specifically about role-playing game). My goals were to show how (thank to the
“flexibility” of games) any theory, be it math or psychology or (case in point) linguistic,
could be used as a cornerstone to assess game .
1
I then produced a second (English) version. Yet the paper resulted too rooted in
philosophy of language and linguistic to be accessible to game designer; and at the same
time yet too generic for an academic journal. This was the reason for third revision, which
included a split that is still the structure of the papers: one (this) specifically related to
linguistic and modal logic, with a foundational goal. Then another one, focused on the
practical applications and examples of said theory (“A Formal Study of Game Design”).
Since then, I submitted only this paper to a dozen of academical journal of philosophy,
linguistic and game studies. Unfortunately, it seems this paper is still either too gamey for
1 Also, do it whilst doing it better than any other supposedly “observation-based” theory.
a philosophy journal, and too philosophical for a game studies journal. That’s why I
worked on a fourth revision, specifically for online self-publishing.
While the theoretical part is almost exactly the same after 6 year, I revised profoundly the  
part related to “A Formal Study of Game Design” to make it (at my best) understandable
and practical to use to someone without any particular background knowledges.
However, remember that what you’re reading now is instead the foundational paper. It’s
not exactly easy to comprehend and apply… but nothing worthed ever is!
Summary
Abstract 1
Prefaction to the 4th edition 1
Summary 2
Introduction 3
Overview & methodology 3
Fundamental assumptions 4
Primordial sets 4
Primordial sets: properties 5
Game statements and game elements 5
Possible worlds and game worlds 6
Circumscribing a (single) game world 7
Setting the equivalency 7
Translation principle and translation function t 8
A practical example of translation function applied to games 9
Rigid designators 9
Looking for coherency 10
The perks of narrative 11
Rules and Story 12
The universal game mechanic 13
Rules vs Story 13
What is a game (for)? 14
Cardinality of games 14
[fd]logic 15
Universe, language and games 16
Future developments 16
Addendum: truths 16
Examples of truths 18
Validation in behavioural games 19
A proposal for truths modelization 19
Addendum: closure notes 20
Bibliography 22
Introduction
Games are between the oldest human activity (with music, alcohol and smoking), but they
have usually suffered from a theoretical underdevelopment. This happens even in fields
that are almost directly derived from games (i.e. game theory and probability calculus).
Hopefully, this paper will foster a discussion about the foundational issue hidden within
any theory of games, and it will help progressing in game studies with a more cohesive
and formal point of view (whatever it happens supporting or disproving what’s proposed
here).
Usually any scholars in game studies approach the matter with a taxonomical perspective
like proposing a definition of “game”, or “rules” or “fiction” or whatever. Alternatively, it is
done starting from assumption taken from other field of knowledge (sociology, or
psychology) in order to use the concept proven in other science and brought them into
game studies. In some cases, “pure” observation of games are used instead of derivative
concept.
Overview & methodology
We personally refuse the foundational approach by axiomatization described above, and
firmly believe it’s wrong. No knowledge is created in a bubble, and whilst to define terms
like “rules” of “fiction” beforehand may seems the proper way to proceed, it’s not for
games. Games possess what are called “emergent” properties, and so they cannot be
explained by an axiomatic approach.
So, if we need some foundation for our theory (and we need it indeed!), we have to resort
to the second strategy and take them from somewhere else . And also we have to
2
demonstrate that those definition will work within our new boundaries and landscape.
2We won’t apply the foundational approach based on empirical observation, because, even if theoretically
interesting and valid, in practice it will require the entirety of neurosciences and sociology studies to explain
the simplest of game.
Fundamental assumptions
The idea here is very simple: to use the tools of modal logic (developed primarily by
Frege, Carnap, Hintikka, Quine, Godel, Lewis... and others) to study games.
The reason why we are allowed to apply modal logic to game, is that any “game” is
inherently a specific subset of either a natural or a formal language. If this is somehow
self-evident in those “story game”; nevertheless formal and modal logic are respectively
used to describe the structure of any language, either extensional or intensional (and
games, as we will see, are either one of those).
So, we have to reformulate modal logic to describe games. We will use the “possible
worlds” concept, that is quite fitting. Even without stretching the concept till the modal    
realism (which is an impegnative ontological position) this idea of possible worlds fits
3
instinctively and perfectly to games (either if you believe that possible worlds are actual
worlds within a multiverse, or if you instead thinks that those are only concepts that we
use in order to manage our inherently uncertain knowledge).
Primordial sets
Now let’s define the initial sets we will use to build our foundations for a structured theory
of games.
Definition Table 1
I. U | ¬U = Ø
II. , x,y | x,y s(x,y) 2
III. G = {g1, g2, g3, … , gN}
IV. W = {w1, w2, w3, … , wN}
First of all, there is U: the set that includes any logically possible object (define by
opposition: its negation is an empty set, as in I.).
Then we define as the general linguistic set (or class): any possible symbol (x) and any
composition of symbols s(x) will belong to (or one of its subset - II.).
Then there is G, which is the set that includes any game (III.) and W(“world”, a subset of
U - IV.). Please note that provisionally (but for good reason) Gand Ware defined by
enumeration.
3 Lewis, 1973
Primordial sets: properties
There is then a number of relationship we may derive from those very simple sets, and
additional sets that we can establish.
Definition Table 2
V. U G
VI. L = {s1, s2, s3, … , sN}
VII. L
VIII. s | s 2L 2
We will come back to V. before the end, but let’s state it and simply leave it right (for
now).
One set particularly useful is L(VI.), which is the entire class of natural languages. Since
we don’t need to face yet the translation problem , we don’t need to differentiate within L  
4
(as total class of all natural languages) its subset like L1,L2,L3(those are specific  
languages like English, French… etcetera).
The VII. lemma states that as power set of signs combination is wider than natural
language (it includes symbolic languages and casual combination of signs).
VIII. defines that any statement smust be always contained within at least one subset of
natural languages (that serves to differentiate between an unarticulated noise and a
proper statement). Those belonging relationship aren’t written as proper subset () but
instead as improper subset because, even if there are strings of symbols (thus belonging
to ) that doesn’t have a correspondence in any language, we cannot exclude at all that
they have a (yet unknown) language that uses them.
Game statements and game elements
Using the elements that we have, we can define the game components (elements those
belongs to the set of games G). Those are splitted in two categories, depending on their
simultaneous belonging - or lack thereof - to both games set Gand natural language set
L.
Definition Table 3
IX. g G
X. gS (g L)
XI. gE (g - L)
4 Kripke, 1979
IX. is about a single game g: this is a proper subset of games in general as G.
X. is what we will call a game statement gS, something that belongs both to a (any)  
language and to a (any) single game.
XI. is instead a game element gE, something that belongs to a (any) game but that’s not  
part of the natural languages (anyway, it will always belong to , of which G is a subset) .
5
We can now define the following additional properties about them.
Definition Table 4
XII. gS gE = Ø
XIII. gS gE = g
XIV. g = ∑gE(1n) ∑gS(1n)
The intersection and relative complement of a same set is always an empty set, so if XIII.
is true (and it is, by definition) thus XII. is proven true too.
XIV. presents an alternative to define a game g, which is exactly the same as XIII. but here
it is explicitly written that gSand gEmay assume any possible value, and that the game    
includes them all.
Those definition may be summarized as follows: in a game we have either game elements
or a game statements, and the sum of them all constitute the game itself.
Possible worlds and game worlds
With our baggage of definition we establish a connection between the concept of
possible world and our emended concept of a game world.
In fact, a game world is indistinguishable from a possible world: they share the same
construction, simply with a little twist on their initial foundation .
6
Definition Table 5
XV. x x
XVI. P, gS | gS gS Φg
5For example, a game element is a a chess piece: something that has formalized properties (that we could     
describe logically). Or, for example a game element may be some specific subset of the code within a video
game.
6We are here using exactly the same solution Quine [Quine, 1960] applies to solve the referential opaque
statement problem: giving them an implicit assertion of existence.
XVII. Φg = ΣnW
XVIII. W | P W Φg
XV. is the assertion of existence as Quine makes it. This, in the possible world logic,
implies that everything logically possible is then possible within at least one possible
world: everything possible is then true (real) at least in one possible world.
Therefore, we can in XVI. redefine existence () using the concept of belonging (). We
can use XVI. as: “Player Pbelieves that a game statement gSis compatible with any  
possible world W belonging to Φg”.
To be precise, the strictly logical definition (in XVII. and XVIII.) may be paraphrased as:
“There is a player Pand a statement gSso that, given any gS,gSbelong to Φg. Φgis      
defined as the sum of npossible worlds W, so that to any Wthere is a Pthat attributes W    
to Φg”.
Speaking about game, we can define Φg: that’s the sheaf that define the relationship
between a single game g and all the related possible game world (Wg) that belongs to it.
Circumscribing a (single) game world
Since they share exactly the same construction (in the definition above, substitute “game
world” with “possible world” and you’ll have the classic definition of them), we can
assume that game worlds are “constructed, and not discovered” [Kripke, 1979].
Now, we have to face another problem stated by David Lewis [Lewis, 1973]: any
counterfactual will never be an exhaustive description of a (singular) possible world. That
description will always embrace an unlimited number (sheaf) of game worlds compatible
with the property (or properties) chosen.
The conclusion is that we create our own game worlds depending on how we want to use
them, and in doing so we always create a sheaf ΦGof multiple game worlds that are all
compatible with our requirements.
Setting the equivalency
It seems natural to establish a connection between W(possible worlds) and WG(game  
worlds). As stated above, and since their definition are equivalent, it makes perfectly
sense that, given any description of WG, we can create a correspondent W.
Given this fact, maybe should we state that possible world and game world are equal by
definition?
Definition Table 6
XIX. W WG
That’s exactly what’s written in XIX. This make perfectly sense (even if it is a little
counterintuitive): there are sounded reasons to believe that we may “play” with (within)
any set W, either that is defined by a game or by something else.
Translation principle and translation function
There are other implication in linguistic that descend by what has been written until now,
specifically those related to the so-called “not-extensional statement”. We can now use
them in our game study (motivated by the application of the possible world semantic as
game world semantic).
The inapplicability of the translation principle [Kripke, 1979] is the most relevant for our
purposes. Given its importance, it maybe worth having a brief overview about what it
meant.
If in classical logic (or logic of first order) we can substitute any lemma and statement with
another one with the same truth value, and the composed truth value doesn’t change. So,
if A=B and B=C then A=C (transitive property) .
7
Intensional statement, instead, does not necessarily possess this property. Any
substitution of a game statement with another one does not necessarily save the truth
value, even if we substitute a true lemma with another one, true by itself.
Then, we can try to define the properties of this translation function t(x) in games. It’s
domain and codomain will be of course g1 and g2 (two different game).
Definition Table 7
XX. t | x g1 t(x) g2
XXI. x,y g1 | t(x) = t(y) x ≠ y
XXII. y g2 | x g1 y = f(x)
Their properties are simply listed and not explicitly demonstrated due to reason of length:
it’s neither an injective function (XXI.) or a surjective one (XXII.)
If you’re interested in the related proof, here’s the general proceedings: to demonstrate
that at least once a single element of g2is equivalent to more than one element of g1(two  
g1elements has the same translation in g2); and to demonstrate that at least in a single
case a function will not exhaust g2(or: there are specific elements in g2that remain  
untouched by the translation from g1). Because we need a single case in which properties
7This is an incredibly important point, specifically in game-design, and thus will be also covered by “A
Formal Study of Game Design”
do not apply to deny it, the translation function is games benefit from neither of those     
properties.
As we will see, the translation principle will become central later when we will find out that
its one of the distinctive trait that we can use to classify games.
A practical example of translation function applied to games
We’ll use a simple, practical experiment strictly connected to the translation principle (or,
to be precise, lack thereof) that we described above to show how we can “feel” the
difference that a behavioural game possesses in relationship with a procedural one.
1. Imagine a situation in any roleplaying game you know (works with any game, but
it’s easier with roleplaying games). Ideally, think about something that requires “a
test” or any other kind of interaction between player and application of rules and
mechanics. The only requirement, is that it need need to be something not self
evident, like any situation with an undetermined outcome. Now write that situation
down (in natural language, with your own words, without using the specific idiolets
of the game).
2. Pick up the game rules, and assign the numbers / descriptors / bonus-malus /    
mechanics (whatever it is) to the situation you written down above, in order to to
solve it (either or a numeric resolution, a poll, or whatever the game use). This
simply means to translate it into games mechanics. Wrote now down the
procedure that “the game” requires.
3. Choose another person. Provide her/him only with the description in natural  
language from step 1., and ask her/him to execute the bulletpoint 2.
Are the bullet point 2. and 3. the same? If those are, then the game is (pretty certain) a
procedural one. If the two differ, and none made explicit mistake in interpreting or
applying the rules, then the game is a behavioural one.
Those description does not correspond to a single Playing Characters, but to unlimited
different playing character that share the same description. Yet, they’re each one different
from another. To say it as Kripke will have said that: there are infinite valid translation in
natural language (or fiction) of the same mechanical descriptive element. We never have a
single character, we always have a sheaf of characters compatible one with each other.
Rigid designators
Another important piece of the puzzle was developed by Saul Kripke [Kripke, 1979] as is
called “rigid designator” principle. This is an assumption needed in order to assign a truth
value to any counterfactual. This represent (also) an answer to problem of translation
principle described above.
The translation problem by itself implies that any counterfactual broke the identity
principle, and thus destroy the chain of significance. Literally: once we accepted a single
counterfactual (“this things isn’t itself but it is something else”), we cannot assign any
truth value anymore (because, if something is different, while only that
should be    
different? Why something else cannot be? And then, why something else again? And
again? And again…
In order to solve this problem, which is embedded in language and logic itself, Kripke
defines a “rigid designator” by an intuitive way, which however allows to create a very
precise definition. The principle is: “a proper name is always a rigid designator”.
To paraphrase that: if we say “Mark”, we always mean the real Mark here and now, even
if we use him in a counterfactual statement, like: “what if…Mark was 15” taller”. In this
case, Mark is exactly the Mark we know, from every point of view, except his height
,    
because that’s what our counterfactual is about.
So, in order to conceive that a counterfactual can be really discussed , we have to  
assume that its truth value may change only on what’s the counterfactual is about. The
conclusion of Kripke is straightforward: if something isn’t explicitly stated as different,
than it’s the same.
Note that only proper names are rigid designator: terms that identifies something by a
definition that can change with the pèossible world (like “my friend” or “the 43rd President
of United States”) are called non-rigid designator
, of flaccid designator
. Differently from    
the rigid designator, their truth value may change with the counterfactual, not only by the
explicitly cited exception.
Looking for coherency
Now, we have to find out some kind of principle to organize truth values. In modal logic,
this is possible thanks to Kripke’s rigid designator: assuming that an element xof a    
possible world WAmaintains all its standard properties p(x); except for those listed in the
Aset that we can explicitly define as “the set of all the differences between WAand W®”.  
Note that W® is the “real” world, the one we’re actually in.
Defining it in this way is no more than a simple notation that allows to point out that ®is   
the reference systems we use to organize any possible world: in other words, the
reference system is the only world we can experience directly.
Here below a single counterfactual is written as a.
Definition Table 8
XXIII. A = {a1, a2, a3, … , aN}
XXIV. a WA a W®
The same identical definitions will apply to our game logic. Games in general are not a
freeform babbling and doing stuff, but possesses some kind of inner logical structure.
We can safely assume that simply observing the fact that players are able to
communicate with each other about the game world, even when they discuss about
something (like the effects a wizard can create by her spell) that has not referent in real
world, thus is a counterfactual without a strict meaning.
At this point we need to understand what is (if there is) the difference between a
performative use of language in game compared to the natural language.
In a natural language we have a criteria (the coherence of the truth value) to lock the
significance chain: we have a way to assign truth value to non-extensional statement (by
their similarities with our reference, ®).
In game language, we have to assume that the players create forecast and expectations
about what’s about to happen, and that is exactly as formulating a belief statement
(regulated by modal logic). Then, we need a decisional logic to assign truth value.
It’s interesting to note that the properties of this internal logic has never been explicitly
8
stated in any game. Intuitively, someone call it “narrative coherence” and connect it with
the suspension of disbelief we have in literature. Other relate it to an (implicit or explicit)
agreement between player calling it “creative agenda”. Someone else talk about features
derived from the game world, and use the term “consistency” or “shared imaginary
space” taking for granted that there is a specific reference criteria ® within any game.
None of these ideas is entirely correct. However, there is an underlying feature that those
ideas share. They state the existence of an “ontologic layer” in games, that provide a set
of decisional criteria whenever we cannot directly apply the “standard” criteria we use for
counterfactuals (adherence to the real world).
Thus, a simple disproof of their validity may be the Occam’s razor principle (we should
avoid anywhere’s possible any multiplication of ontological objects to solve a causation
or - or signification - chain).
The perks of narrative
Now, let’s make a small parentesys to face the fact that there are self-evident differences
between various forms of games, and how they applies to the differences between
extensional and intensional statement in games.
9 10
8 The one we will call “Fictionally Distributed Logic” or [fd]logic
9 Obeying classic logic
10 Obeying modal logic
In a chess game we have either intensional statement (“the white player believe he’s   
gonna lose
”) and extensional statements (“a pawn is in E2
”). In games, we can determine    
if an extensional statement is true or false by applying the game rules and mechanics,
and we have to apply modal logic in order to solve any intensional statements related to
the game. As anticipated, we will call them procedural games.
The most interesting feature of those games is that “what’s matter” into the game are
only the extensional statement. In no procedural games there are any legit intensional
game that can be solved or determined by the rule of the game itself.
There are, conversely, behavioural games. Those, as proven separately , doesn’t have a  
11
clear and fixed distinction between extensional and intensional statements, it’s possible
that indistinctly we may need classic logic and game elements, or modal logic and game
statements to assess the truth value of part of the game. In those behavioural games the
truth is “locked” inside a different level than the rules (or to say it in a more accessible
language, “the truth we are interested about regard the story, not the application of
rules”).
This happens only in this category of games: in any other game are the rules themselves
(and the result of their combination) are enough to create a validation structure to
determine any truth value (which should be clear, right now, that is an equivalent for
“playing the game”).
So, if we define the “purpose” of the game as the calculation of truth values for the
game-related statements, there are games that require modal logic (and a specific
behavioural attitude) in order to be used. Those, as said above, are the game where truths
are located within the narrative, into the game world, and not into the game.
From a general standpoint: games are a incredible fast growing and wide ecosystems.
There are literally more games in the world than how many any human being can
experience in a full lifetime. However, we don’t need to list all existent games to provide a
definition and a differentiation strategy to categorize them: it’s enough to observe if the
the game itself includes intensional statements and the procedures necessary to manage
them. If not, it’s a extensional game (and doesn’t matter what are its components, shape,
medium or focus).
Rules and story
We can now give the following logical definition for what’s usually called “story” (XXV.
),  
juxtaposed to “rules” (XXVI.
). A “game” generically intended (XXVII.
) is the sum of the two.
11 This is a focal point of “A formal study of game design”. For further explanations and example reference    
to that.
Definition Table 9
XXV. Sg = ΣgS g | Ե(gS) =
XXVI. Rg = ΣgE g | Ե(gE) =
XXVII. g Φg Sg Rg = Σ(gS gE)
In this definition table we begin to use Եas a univocal symbol for the specific function f(x)  
that calculates the truth value of the element x, giving an output which is binary: either    
(true) or (false).
Written not in formal language, XXV. predict that the story is the sum of game statements
that belongs to the game, and are true. The rules, similarly, are the sum set of all the true
game elements (XXVI.) A game itself is the union set of those two sets.
Thanks to all the principles enunciated until now we know that we can rewrite a truth
value into a game as a statement of existence, and this statement of existence may be
rewritten as a statement of belonging to a possible world, which really is a sheaf of
compatible possible world differentiated one from another for any counterfactual small as
we want.
The universal game mechanic
We now can create have a more powerful definition that describe what a game (any
game) does, and specifically what a “game mechanics” means.
Definition Table 10
XXVIII. g G | g Ե(g) [, ]
XXIX. |Ե(g)| = 2
A game mechanics is then any truth value calculation over gin Gthat can be effectively
solved, which values is either “true” or “false” (its codomain has cardinality=2.
Rules vs Story
We have an interesting implication from Definition Table 9 and 10: are either the “story” or
the “rules” enough, by themself, to create a game? The answer, as should be
self-evident, is a sounded “yes”.
There are two ways to prove it: one is that, logically, we can substitute gS=Ø in the
relevant definitions. The other, already cited before, is that procedural games (like chess)
does not have any gS that they legitly can process by their own rules.
What is a game (for)?
Now we have enough elements to provide a preliminary definition of what is a game. Here
is written in natural language (the logical formulation is straightforward from the definitions
used until now).
Definition Table 11
XXX. A game gis a function to assign truth value Եrelated to a specifically
constructed possible world W, called game world Wg.
Using the tool of modal logic that we have created, we can expand from XXX. including
the two different subsets of games that we cited until now.
Definition Table 12
XXXI. A game assign truth values by a specific sub-set of criteria.
XXXII. Those criteria may be explicit (procedural games) or implicitly involved by
a gaming attitude (behavioural game).
Procedural games can be also identified by the greek letter rho (ρ). Behavioural games    
will be indicated instead by the greek letter beta
(β).
Cardinality of games
From the description above we can immediately quantify the size of the different sets and
subsets by their cardinality:
Definition Table 13
XXXIII. |Σs| = g
XXXIV. g Ø g ρ
XXXV. g > Ø g β
XXXVI. ρ > β
Given the cardinality of the sum of all possible statements sdefined on gas in XXXIII., a    
game is procedural if this cardinality is at maximum equal to aleph zero (Ø). If it’s greater,
than it’s a behavioural game.
Please note that (at least for now) the use of Øas measurement of cardinality is purely
arbitrary, and those not implies that cardinality here are equivalent as defined in math.
However, there are sounded reasons to believe that a game like chess - a good example
of procedural game - “contain” exactly the same cardinality of natural numbers (N set),
given the fact that we can explicitly create a chess game that is made of infinite “integer
and positive” moves (the natural number set, which cardinality is exactly Ø).
An interesting and still open question is: does behavioural games posses the same
cardinality of the real numbers (R) set? Or, instead, do they have the same cardinality of
the integer (Z) set?
12
Regardless of the exact relationship between the different , definition XXXVI. is the      
pivotal point here: the cardinality of behavioural (β) games is always greater (and never
equal to) than the cardinality of procedural (ρ) games.
[fd]logic
Finally, we are able to start using our Fictionally Distributed logic. The name, now, should
make a lot more sense: it’s the definition of the fact that a game logic cannot be neither
classic logic or modal logic, but have to use them both, respectively in the two different
category of games, and provide with the tool in order to distinguish when should be use
one or the other.
At this point, we can propose some general principles because we have (until now)
created a formal foundation and a definition of applications for this logic.
1. In order to function, [fd]logic presume that what’s written until now is correct. So,
for example, that game may be defined as tool to assess the truth values.
2. The fictional distributed logic is branch of logic applied to games in general, that
allow the description in logical term of the activity commonly defined as “game”.
3. In case of a procedural games, [fd]logic collapse into formal logic; when applied to
behavioural games it collapse in modal logic (with the definition “game world”
instead of “possible world”).
4. [fd]logic allows to identify precisely the component of a game, thus giving
differentiated and aimed procedures applicable to different typology of games .
13
5. [fd]logic is capable of explaining why, for example, procedural game may be
played through any language barrier, by people that only shares the rule of game
themselves. This cannot be true in a behavioural game instead requires a
completely different level of interaction and common basis between players: you
can’t play a narrative game if, at least, all the players doesn’t share a common
language.
12 One of the future development of this study is the idea that is possible to create a Cantor-like diagonal
proof, that can connects the cardinality of a procedural games to those of behavioural games.
13 Even if this point will be better managed into “A Formal Study of Game Design”, it’s worth noting that,
actually, almost all current game design relies only on classical logic, even in behavioural games.
Universe, language and games
We can finally address a simple but profound consequences of what’s written until now. If
it is true that a gaming attitude is necessarily and sufficiently to have a “game”; if also it is
true that the “game world” may be as “big” as the material universe: then the relationship
between U, and G (at V. - VIII.), need to be updated as follows.
Definition Table 14
XXXVII. U G
We have to admit that there is actually a terminal possibility (in case of =) that the
universe itself is a symbolic systems, and may, in its entirety,considered a game .
14
Future developments
Those are a number of general conclusions, guidelines, suggestion and topic of further
investigation that derives from what’s written until now. This is only an overview: a
number of them will be discussed and analyzed deeper in “A Formal Study of Games
Design”.
A diagonal demonstration applied to games with a strong math component (like
chess or domino) should, in theory, be able to prove or disprove the hypothesis
made here about cardinality g = 0.
There are two kind of games. Some are made up by of procedures, and those
game may be played by a Turing Machine. There are games intrinsically made from
a behavioural attitude (like role playing game). Even if someday, we will be able to
create perfect natural-language syntax machine there are reasonable doubts that
this machine will be able to play a behavioural game. That means that a revised
version of the Turing Test for AI should be the ability to comprehend (thus to play)
a behavioural game. Since behavioural game requires the use of modal logic, this
will be inextricably connected with the recent development in deep learning,
machine learning applied to natural language and the classic Turing test . Our
15
idea, is that a “proper” Turing test for an IA is not based on the capability of a
machine to “travesty” as human, but by their capability of playing a behavioural
game.
Caillois’ taxonomy of games that defines games by the “experiences” (agon, alea,
16
mimesis, ilinx) also includes the idea that games dispose themself on an axis of
17
between ludus
and paideia
. Caillois thought the two were related to “complexity”  
14 This is surprisingly consistent with the holographic theory of the universe.
15 Case in point: today we have machines able to pass a classic Turing test. Yet, we “feel” that they still are
not “true” artificial intelligence.
16 Caillois, 1967
17 There isn’t a direct equivalent of those four area in this paper.
of game rules: an intuitive, but not formalized concept. We believe that those two
extreme correspond to what we discover as procedural (ludus) and behavioural
(paideia) games. Differently from Caillois, however, those are two completely
separated set and there is a precise boundary, not a gradual transition from one to
another .
18
Playing a game is making expectations about a game world, in order to test them.
Playing means find out what’s right and what’s wrong about our game world. As a
necessary corollary, playing games means improving in both the skill required to
19
validate statements, and knowledge of the prior criteria chosen.
A β-game is, by definition, a subset of ρ-game. Also, it is without precise
boundaries. There are infinite possible truths that can be verified by the game. A
procedural game, instead, could (theoretically) be exhausted, given enough time
and resources (a.k.a. a Turing-machine can solve it).
We can define “flaccid-procedures” for behavioural games in order to calculate any
truth values that we need (see “Addendum” below). Those procedures rely on the
game world, not on rules composition. Those procedures are necessarily
incomplete (the game world is incomplete by definition: it’s always a sheaf of game
worlds, exactly like there are infinite possible counterfactual).
The “practical” game is no sure indication about the ρ or β structure. A game may
seem behavioural because it involves storytelling and offer a great degree of
options to players to contribute to creation of the game world… but being entirely
procedural from the standpoint of validation .
20
We can define a gamelet: it is the specific subset of game that any different player
uses. Those gamelets may or may not overlap, and may be the results of either
casual of systematic factors .
21
We can define “story” and “rules” logically, but they do not exist in practice as
separate entities. In a ρ-game, the “story” is equivalent to the rules application,
whilst everything outside that does not belong to the game (do not have a truth
value). In a β-game, instead, the “story” is a rule by itself, since it provide criteria to
assign truth value.
If games are function applied to sets, it will be interesting to study if those function
are continuous, discrete, neither or both (in the latter cases, it means that different
games can be discrete or continuous). This is a particularly complex thing to prove
or disprove. Actually, we propend towards the idea that the game function operate
18 Intuitively, we can explain the apparent gradual transition due to the fact that many games “seems” more
procedural or behavioural depending on their practical rules. Our distinction is instead theoretical.
19 Either “play many different games” or “play a lot with the same game”
20 A similar game is quite easy to write: “the player to the left determine the truth value of the statements       
pronounced by the player on the right”. This is also a “flaccid procedure” as defined above
21 More on them in “A Formal Study of Game Design”
in the continuous, but can be practically managed only in its discrete
approximation.
Addendum: truths
Exactly like there is a number of different truth values and definition in linguistic , so there
22
should be in games. This analysis, even if not needed for foundational goals, allow us to
understand how different statement about the game world (and the game itself) can (or
cannot) be validated... and by what procedures.
This part of the paper will be less formal the the one above, but it’s extremely important if
you want to apply also the notion contained within the other paper “A Formal Study of
Game Design”.
Definition Table 15
XXXVIII. Necessary [NE ] truth includes all those statement that are either true or  
false in all possible game world. They are the defining elements of the game, and
can be further divide into:
a. Prior
[Pr ] truth is something self-explaining as true/false. It’s usually  
a logical tautology like x=x, or a property that we use to wrap the game
world around it. Usually those are the very cornerstone of the game.
b. Analytical
[An ] truth it’s something true by how it’s presented (by  
structure). An analytical truth start from another truth (whatever it is) and
then, demonstrating that there is a connection with the required truth
value. Bug/fallacies within the game belong to this category, as general
statement like “someone will win”.
XXXIX. Contingent [CO ] truth includes those statement either true or false in  
some game worlds, while not in others. Any contingent truth need a specific
procedure to determine if it’s true or false for the current game world. Those are
the defining element of the play that we’re having, and not of the game and game
worlds themselves.
a. Posterior
[Ps ] truth is valid in the current game world. It’s true only  
because “it happened”. In the case of a narrative game, we can use it as:
“something true because it happened in the game world”. In the case of
games, the most evident truth belonging to this set are those caused by
dice or random chance. It’s a truth valid in our current play, not in the
game in general.
b. Potential
[Pt ] truth, also called as undetermined, is a truth that is  
valid in some game world, but not -yet, or eventually never - in the one
22 Lewis, 1973
we’re in. Those are, also, all the question that the game may (and will)
answer while the players are playing.
There are two broad categories that exist in language and we can use directly into games.
Some have “constitutional” truth values, embedded within the entire sheaf of possible
world compatible with the game: we literally cannot change them without changing the
game itself.
The others are truth values specific of a subset of the game worlds: their truth value may
(and often will) change with further explorations of the game world. Their variations,
however, still keep us within “the same” game.
Examples of truths
But let’s see some examples of different truths in case of games. Since we yearn to
create a general game theory, we’ll show how this applies to his original context (a literary
work), to a classical game (chess) and to a video-game (Fallout 3).
Just remember, that in order to analyze the truth value of a statement, we don’t need if to
be necessarily “true”. “False” is a lecit outcome too. A problem will arise if (or when) we
cannot properly assign neither.
Example 1
Truth
“Lord of the Ring”
“Chess”
“Fallout 3”
NE
Pr
Gandalf is Gandalf
Board size is 8x8
Year is 2277
NE
An
Gandalf is an Astari
A single Pawn deploy in D2
It’s been 200 years from war
CO
Ps
Sauron has been defeated
White lose his Queen
Luck stat is < 7
CO
Pt
Sauron wins the war
White win the game
Vault 101 never opened
How this different truth values may be calculated in different scenarios of games isn’t of
capital importance here. The important things is that we can use some of them: we will
call them “validation procedures” or simply “checks”. Also, we can synthetically
demonstrate the difference between procedural (XL.) and behavioural (XLI.) game,
because we have now a way to describe the different possible condition of validation for
both of them.
Definition Table 16
XL. (gS1, gS2) | Ե(gS1) PS, Ե(gS2) AN Ե(gS1) = Ե(gS2)
XLI. (gS1, gS2) | Ե(gS1) PS, Ե(gS2) AN Ե(gS1) ≠ Ե(gS2)
This definition differs for the presence of either = or in the end. It means that we can
pick any statement with a contingent truth value, and analytically derived another
statement from it. Then we have a “procedural” game if the truth value of second
statement changes accordingly with the truth value of original one. If, instead, the derived
statement changes truth value by its own terms, (regardless of the value of the first one),
then we have a behavioural game.
Not surprisingly, this demonstration still operate under the frame of the translation
principle. Indeed, the presence or absence of a translation principle equivalent applicable
to a game is also a necessary and sufficient condition to classify it as either procedural or
behavioural.
Validation in behavioural games
Now we have different notions of truths, and possible procedures to validate them. Let’s
see the most interesting case: how can validate a posterior truth Pt in a behavioural
games.
Definition Table 17
XLII. We can validate the potential truth by abstraction/deduction from other
truth we already know. For example: we can prove that either Pt or ¬Pt is a
tautology (then it’s a Pr truth); we can prove that from a Ps,An or PT) truth we will
derive the Pt (then it’s an An truth); or we can prove that it is part of the definition of  
the current Wg (then it is a Ps truth).
XLIII. We can create a validation chain: creating additional Ps or An truth as long
as we are able to connect our Pt truth with one of those, and then apply one of
the procedures above .
23
Note that you can use those strategies also with a procedural game: in this case,
however, any procedure you pick the result will be the same (because the formal logic
that applies to procedural game state that different composition of truth values doesn’t
change if you change the process of calculating it, or translation principle).
Instead in modal logic (and behavioural game) any different procedures we pick up to
calculate the truth value affect the final result.
23 Here basically we develop further characterizations of the game world until we have what we need. We      
can do that, since the game world isn’t ever fully and self-sufficiently defined.
There is however a counterintuitive consequence to what we have described right now. It      
may seems that any statement may be either true or false depending from how we
describe it, rather than “what it is”.
It seems strange, right? Common sense told us that there is a thing called truth, and this
can’t be both true and false… right? No: that’s wrong. Let’s see why it is.
Addendum: A proposal for truths modelization
The relationship between different truths may also be represented graphically. This kind
of description is a double utility: first, it allow to spatially represent the different validations
procedures we have. Secondly, we can use all to standard tool of cartesian space (for
example standard functions definitions and analysis) to games.
The idea is to draw a cartesian plane. Any point into it may be described as usual as a
couple of real number [x,y]. We use + and - to identify the different quadrants.
There are four possible iteration: [+x,+y], [+x,-y], [-x+y], [-x-y]. Different truth value are
consistently within a single quadrant:
Analytical truth are [+,+];
Prior truth [+,-];
Potential truth [-,-]
Posterior truths [-,+].
In this representation, necessary truth have a positive component on the x axis [+x], while
contingent truth a negative one [-x].
An interesting hypothesis to make is that the x axis somehow represent the “logical
consistency” of the proposition, while the y axis is connected with a specific moment in
time (to be precise, a certain point in the exploration of the fictional world).
This representation is particularly apt to identify kinds of truth, but we suppose (and save
it for further studies) that it is possible to define functions map the transaction between
different truth defined in the totality of the Wg.
This is also interesting because it shows how games are dynamical system of
relationships, kept together by a description of a game world.
Example 2
It’s still unclear and deserve
further study if a game is an
integer of the boundary
function that discriminate
between True () and False
(); or if it is instead only the function by itself which result as and anything outside it is
.
Bibliography
Bacharach, M. and Gerard-Varet, L.A. and Mongin, P. and Shin, H.S. 2012. Epistemic       
Logic and the Theory of Games and Decisions.
Springer
Baker, G.P. and Hacker, P.M.S. 1980. Understanding and Meaning: Vol 1 of an Analytical    
Commentary on the Philosophical Investigations.
Blackwell.
Boghossian, P. 1997. “Analyticity” in A Companion to the Philosophy of Language
.    
Blackwell.
Caillois, R. 1967. Les jeux et les hommes: le masque et le vertige
. Gallimard.
Cantor, G. 1891. Ueber eine elementare Frage der Mannigfaltigkeitslehre. Jahresbericht   
der Deutschen Mathematiker-Vereinigung 1890–1891
. English translation Ewald, W.B. 
1996. From Immanuel Kant to David Hilbert: A Source Book in the Foundations of  
Mathematics, Volume 2
(pp. 920–922). Oxford University Press.
Carnap, R. 1937. The Logical Syntax of Language
. Routledge.
Carnap, R. 1942. Introduction to Semantic.
University of Chicago Press.
Carnap, R. 1943. Formalization of Logic.
University of Chicago Press.
Carnap, R. 1947. Meaning and Necessity: A Study in Semantic and Modal Logic.   
University of Chicago Press.
Chomsky, N. 1994. Language and Thoughts
. Moyer Bell.
Chomsky, N. 2000. New Horizons in the Study of Language and Mind
. Cambridge      
University Press.
Davidson, D. 2001. Subjective, Intersubjective, Objective.
Oxford University Press.
Deutsch, D. 1997. The Fabric of Reality
. Viking Press.
Dummett, M. 1973 Frege. Philosophy of Language.
Duckworth.
Dummett, M. 1978. Truth and Other Enigmas.
Harvard University Press.
Dummett, M. 1991. The Logical Basis of Metaphysics.
Harvard University Press.
Wittgenstein, L. 1953. Philosophische Untersuchungen.
Blackwell.
Frege, G. 1892. “Über Sinn and Bedeutung” in Zeitschrift für Philosophie und    
philosophische Kritik 100
(pp 25-50). Haake.
Gil, F. 2000. La Conviction.
Flammarion.
Grice, P. 1989. Studies in the Way of Words
. Harvard University Press
Hintikka, J. 1962. Knowledge and Belief - An Introduction to the Logic of the Two     
Notions.
Cornwell University Press.
Hintikka, J. 1997. Lingua Universalis Vs. Calculus Ratiocinator: An Ultimate   
Presupposition of Twentieth-Century Philosophy.
Springer.
Hintikka, J. 1998. Language, truth, and logic in mathematics.
Springer.
Hintikka, J. 1998. Paradigms for language theory and other essays.
Springer.
Kripke, S. 1979. A puzzle about belief
. Springer.
Kripke, S. 1980. Naming and Necessity
. Blackwell.
Lewis, D.K. 1973. Counterfactuals
. Blackwell.
Putnam, H. 1978. “Meaning, Reference and Stereotypes” in Meaning and Translation
(pp   
61-81). Duckworth.
Quine, W.V.O. 1951. “Two Dogmas of Empiricism” in Philosophical Review, 60 (pp    
20-43). Harvard University Press.
Quine, W.V.O. 1956. “Quantifiers and Propositional Attitudes” in Journal of Philosophy,    
53
(pp 177-187). Journal of Philosophy Inc.
Quine, W.V.O. 1960. Word and Object
. John Wiley and Sons.
Russell, B. 1919. Introduction to Mathematical Philosophy.
Allen and Undwin.
Randolph, A. 1999. “Homo Ordinator” in Board Games in Academia III. An     
interdisciplinary approach.
Niek Neuwahl.
Tolkien, J.R.R. 1954. The Lord of the Ring
. Houghton Mifflin Harcourt
Vv. Aa. 1966. L’analyse structurale du recit.
Communications, 8. Seuil.
ResearchGate has not been able to resolve any citations for this publication.
Book
Introduction. Origin of the Essays. 1. The Games of Logic and the Games of Inquiry. 2. No Scope for Scope. 3. Informational Independence as a Semantical Phenomenon (with Gabriel Sandu.) 4. 'Is', Semantical Games and Semantical Relativity. 5. Logical Form and Linguistic Theory. 6. On the Any-Thesis and the Methodology of Linguistics. 7. Paradigms for Language Theory. 8. (with Gabriel Sandu.) The Fallacies of the New Theory of Reference. 9. Perspectival Identification, Demonstratives and 'Small Worlds'. 10. Game-theoretical Semantics as a Synthesis of Verificationist and Truth-conditional Meaning Theories. 11. (with Gabriel Sandu.) Metaphor and Other Kinds of Nonliteral Meaning.
Chapter
I hope that some people see some connection between the two topics in the title. If not, anyway, such connections will be developed in the course of these talks. Furthermore, because of the use of tools involving reference and necessity in analytic philosophy today, our views on these topics really have wide-ranging implications for other problems in philosophy that traditionally might be thought far-removed, like arguments over the mind-body problem or the so-called ‘identity thesis’. Materialism, in this form, often now gets involved in very intricate ways in questions about what is necessary or contingent in identity of properties — questions like that. So, it is really very important to philosophers who may want to work in many domains to get clear about these concepts. Maybe I will say something about the mind-body problem in the course of these talks. I want to talk also at some point (I don’t know if I can get it in) about substances and natural kinds.
Book
1. What Is Elementary Logic? Independence-Friendly Logic as the True Core Area of Logic. 2. A Revolution in Logic? 3. A Revolution in the Foundations of Mathematics? 4. Is There Completeness in Mathematics After Godel? 5. Hilbert Vindicated? 6. Standard vs. Nonstandard Logic: A Watershed in the Foundations of Mathematics. 7. Standard vs. Nonstandard Logic: Higher-Order, Modal and First-Order Logics. 8. (with Gabriel Sandu.) The Skeleton in Frege's Cupboard: The Standard vs. Nonstandard Distinction. 9. (with Arto Mutanen.) An Alternative Concept of Computability. 10. (with Gabriel Sandu.) What is the Logic of Parallel Processing? 11. Model Minimization - An Alternative to Circumscription. 12. New Foundations for Mathematical Theories.
Article
Modern empiricism has been conditioned in large part by two dogmas. One is a belief in some fundamental cleavage between truths which are analytic, or grounded in meanings independently of matters of fact, and truths which aresynthetic, or grounded in fact. The other dogma is reductionism: the belief that each meaningful statement is equivalent to some logical construct upon terms which refer to immediate experience. Both dogmas, I shall argue, are ill-founded. One effect of abandoning them is, as we shall see, a blurring of the supposed boundary between speculative metaphysics and natural science. Another effect is a shift toward pragmatism.