Video Game Addiction and Gambling

Characteristics in Video Games

Janik Czempin
Technical University of Munich
Hauptstraße 10b
85386 Eching
Bavaria, Germany
+498930668208
[Link]@[Link]

ABSTRACT
The concept of video game addiction is still controversial. The current state of research lacks
a clear foundation, especially concerning the diagnosis and symptoms of this proposed dis-
order. Instead of focussing on this uncertainty, this paper relates video games to gambling
addiction, which is much more clearly understood. Not all video games are the same in how
they relate to gambling and addiction. This paper categorizes video games based on how
close the random mechanisms within them are to gambling. The Audiovisual representation
of random mechanisms in games also affects the extent to which gambling characteristics
can occur within them. For each known characteristic found in gambling, this paper de-
scribes which types of games commonly exhibit the characteristic. Examples and proposals
are provided, for how gambling characteristics can be avoided or made to occur. Avoiding
gambling characteristics in games is critical for video game developers who strive to act
ethically and do not want their game’s players to become addicted. Players knowing how
gambling characteristics can be brought about in games is important so that these possibly
dangerous mechanics can be noticed and evaded.

INTRODUCTION
Video Game Addiction is rarely discussed among video game developers and students of
Games Engineering. Perhaps this is because it makes them uncomfortable. It is difficult
to think of the software one develops as something that might be harmful. It could also be
that Video Game Addiction is poorly understood. The topic could be awkward to talk about
because it is difficult to make definitive statements. Either way, Video Game Addiction is
not taught in many courses, so a lot of people do not have a strong foundation of knowledge
on it.

Learning about Video Game Addiction can have many benefits. A Video Game Developer
trying to behave ethically might consider trying to prevent addiction by refraining from in-
troducing game mechanics that are highly addictive. One who primarily cares about the
financial profit of games could, on the other hand, exploit knowledge of video game addic-
tion to make a particularly addictive game.

But this knowledge is not only interesting to game developers. Games journalists and play-
ers of games can discuss the issue productively only if there is a basis of understanding.
Doctors and parents want to know what steps to take and when to prevent or reduce the
negative effects of video gaming.

This paper aims to give a solid foundation of knowledge: Where the state of science is on
Video Game Addiction. It will provide an overview of the topic, try to define video game
addiction, how it can be diagnosed, and how it compares with gambling addiction. The
paper then goes into detail on how gambling characteristics occur in video games.

DEFINITION AND DIAGNOSIS OF VIDEO GAME ADDICTION

This chapter tries to find a definition of video game addiction and how it is diagnosed. The
chapter also outlines the general positive and negative effects of video game play.

Unfortunately, there is no clear consensus on whether ”video game addiction” exists nor if
the term ”addiction” is appropriate. Here are some terms that have been used by researchers
to describe what might be the same condition: ”video game addiction”, ”problematic video
game play” ”problematic online game use”, ”online gaming addiction”, ”Internet gaming
addiction”, and ”compulsive Internet use”. (Mark D Griffiths et al. 2012) Some of these
terms have been controversial due to a lack of clear boundaries of categorization. In par-
ticular, the focus on online gaming and the exclusion of non-online video gaming has been
criticized as having no clear reasoning behind it. (Anthony M. Bean et al. 2017)

Even definitions proposed by very reputable organizations, such as the American Psychi-
atric Association (2013) and the World Health Organization (2018), have come under harsh
criticism. In 2013, the APA’s Diagnostic and Statistical Manual of Mental Disorders was
updated to its fifth edition, including new mention of a so-called ”Internet gaming disorder”.
However, due to the controversy around this definition, it was not included in the main body
of mental diagnoses, but in the additional chapter ”Conditions for Further Study”. (Amer-
ican Psychiatric Association 2013) The WHO was not spared from criticism either, when
”Gaming disorder” was included next to ”Gambling disorder” in the eleventh revision of
the International Classification of Diseases, released in 2018. (World Health Organization
2018) Specifically, critics lamented that the description of symptoms is too vague and, as
such, not applicable to real-life scenarios. Further, regarding these categories, it has been
suggested that political pressure drove some of the decision-making. This was admitted by
parts of the WHO team running field trials for the gaming disorder categories.(Anthony M.
Bean et al. 2017)

It is difficult to make a definitive statement on whether or not video game addiction ex-
ists as a unique disorder. Some studies have instead indicated that video game addiction
is a symptom of other underlying conditions, not an independent diagnosis. Others have
found that, although ”Internet gaming disorder” appears to not relate to other mental health
concerns, symptoms were not associated with social or psychological impairment, contrary
to the DSM-5 (American Psychiatric Association 2013) and ICD-11 (World Health Orga-
nization 2018) proposals. Excessive video game use is associated with multiple comorbid
disorders. These include attention deficit hyperactivity disorder (ADHD) and symptoms of
generalized anxiety disorder, panic disorder, and depression. (Mark D Griffiths et al. 2012)
Still, some researchers persistently assert that their data indicates the existence of a unique
video gaming disorder. (Anthony M. Bean et al. 2017)

–2–
Whether or not harmful video game play can be classified as an addiction, there are many
indicators that excessive playing of video games may lead to negative psychological conse-
quences for a minority of affected individuals. These include sacrificing work, education,
and sleep, increased stress, loneliness, poorer social skills, and suicidal ideation. Negative
medical consequences such as repetitive strain injuries, obesity, and seizures have also been
reported. (Mark D Griffiths et al. 2012)

It has been argued that their social aspects are what make video games addictive. However,
social aspects of video games have also been shown to be helpful for positive self-esteem de-
velopment, and appropriate social interaction engagements. Video games have also helped
increase social aspects of individuals with Autism, a clinical population known for their dif-
ficulties with social engagement, but increased video game playing tendencies. (Anthony
M. Bean et al. 2017)

Multiple instruments attempting to measure the phenomenon of video game addiction have
been produced. But the symptoms used in many of these surveys are often based on the crite-
ria from the DSM-5. The tautological relationship between survey instruments and diagnos-
tic criteria call the validity of the DSM-5 and ICD-11 criteria into question. Most diagnostic
surveys reformulate criteria from substance abuse disorders and gambling abuse disorders,
assuming these criteria will easily transfer to video games. Words such as ”heroin” are
simply replaced with ”video games”, hardly regarding the validity of rewording questions
without foundational research. (Anthony M. Bean et al. 2017)

Because there is little consensus on measures to diagnose video game addiction, prevalence
rates vary wildly between different studies: Some researchers estimate 45% of their sample
to meet addiction criteria, others suggested problematic gaming to be closer to 1% of the
population. This raises the issue of considerable amounts of false-positive cases. (Anthony
M. Bean et al. 2017)

Great care should be taken with these issues, since a diagnosis lacking basic clarity may
ultimately do more harm than good. (Anthony M. Bean et al. 2017) Unfortunately, it must be
concluded that there is currently no agreed-upon definition or measure of diagnosis among
researchers of video game addiction.

GAMBLING IN VIDEO GAMES

This chapter will compare video game addiction with gambling addiction and discusses if
random reward mechanisms in games (”Loot Boxes”) can be considered gambling.

Behavioral addiction involves a compulsion to engage in a rewarding behavior not related to

drugs. Potential nonsubstance addictive-related behaviors include gambling, video gaming,
Internet use, shopping, exercise, excessive eating, and sex. However, gambling is currently
the only nonsubstance behavior that is officially recognized as addictive by the American
Psychiatric Association. (Anthony M. Bean et al. 2017)

How does Video Game Addiction Relate to Gambling Addiction?

Gambling is defined as accepting, recording, or registering bets, or carrying on a policy game
or any other lottery, or playing any game of chance, for money or other thing of value. (Title

–3–
18 of the United States Code § 1081) Gambling can happen in many contexts, including on
the Internet and in video games. The important defining feature of gambling is the possibility
of getting something of value, such as money, out of it. Buying and playing a video game
often involves only a single monetary interaction, and usually, no money or other thing
of value can be gained from playing video games. So, playing these video games cannot
be considered gambling and excessive or problematic use of these video games cannot be
considered gambling addiction.

However, some video games, including those played via a browser over the Internet, do
involve registering bets and playing a game of chance, for money. These games, such as
Online Poker or Online Blackjack, follow the definition of gambling precisely. So, if some-
one plays these games excessively, they might be considered as suffering from gambling
addiction, as defined by the diagnostic criteria outlined in the DSM-5. Because they are
playing a video game excessively, might they also be considered as suffering from video
game addiction? Compared to gambling addiction, diagnostic criteria and treatments for
video game addiction are poorly understood. Not including video game addiction alongside
gambling addiction in the DSM-5 diagnostic criteria demonstrates this. The symptoms and
treatments for someone gambling in a video game are already sufficiently covered by gam-
bling addiction. So it is not necessary to diagnose them with video game addiction, which
is so poorly understood.

RANDOM REWARD MECHANISMS IN VIDEO GAMES

So-called ”Loot Boxes” became the subject of public political debates at the end of 2017
and resulted in legal action in some European countries. The term ”Loot Box” refers to a
specific visual representation of mechanisms that reward the players of some video games
with random virtual objects if a certain objective has been achieved. (Rune K. L. Nielsen
and Paweł Grabarczyk 2019) Because they do not consider the term ”Loot Box” and the
phenomena it covers as sufficiently precise for academic use, Rune K. L. Nielsen and
Paweł Grabarczyk (2019) instead introduce the notion of ”random reward mechanisms”
(RRMs for short).

To cover many different implementations of RRMs, their structure is described in a very

general form. A RRM consists of three components: An eligibility condition, followed by
a random procedure, resulting in a reward. The eligibility condition is the requirement the
player must meet to trigger the random procedure. This can be the death of a virtual mon-
ster, achieving a certain number of experience points, or a payment of real money. The
random procedure can be achieved by any such method used in programming, the details of
which are not pertinent to the discussion. The nature of the reward can be any element of the
game that can be awarded to the player, such as virtual characters or costumes or in-game
currency. Defined as such, RRMs are not a new phenomenon in games. For example, in
Diablo (Blizzard North 1997) and RuneScape (Jagex 2001), killing a certain enemy (eligi-
bility condition) triggers an event (random procedure) that rewards the player with a new
virtual item (reward). (Rune K. L. Nielsen and Paweł Grabarczyk 2019)

One reason why RRMs have only recently become widely debated might be their audiovi-
sual representation. Players of older games, even those that used randomness extensively,
might not have even realized that there was a random procedure involved. Even though

–4–
the randomness of the procedure was not hidden from the player (for example, it was often
included in the marketing of these games), the games did not themselves indicate random
behavior via their audiovisual representation. Contrary to this, many modern implementa-
tions of RRMs accentuate randomness by using easily recognizable audiovisual tropes, such
as spinning wheels, dice shaking sounds, card shuffling, and so on. Initially these imple-
mentations were dominant only in the mobile market. Only in recent years have large game
publishers introduced similar systems in games developed for home consoles and PCs, such
as Overwatch (Blizzard Entertainment 2016), Call of Duty: Advanced Warfare (Activision
and Sledgehammer Games 2014), and Star Wars: Battlefront II (Electronic Arts and DICE
2017). (Rune K. L. Nielsen and Paweł Grabarczyk 2019)

Another cause of significant public outcry in the gaming community has been the introduc-
tion of real currency into RRMs of mainstream games. To clarify, games may have any
number of virtual currencies or ”tokens”. Only some of these tokens have a relation to other
objects of value embedded in the everyday economy. This makes it possible to establish the
value of these tokens in different real currencies. For example, if the tokens can be bought
or sold with an existing accepted currency, it is said to be ”embedded” in the economy.
Contrary to this, tokens that have no established relation to any other objects of value out-
side of the game are said to be ”isolated” from the economy. Games with multiple virtual
currencies commonly include one embedded token, the rest being isolated tokens. What
this means in practice is that only one of the currencies can be purchased with an existing
accepted currency and that the relation between the embedded and isolated currencies is
restricted. (Rune K. L. Nielsen and Paweł Grabarczyk 2019)

Note that a currency that can be freely exchanged for an embedded currency must also be
considered an embedded currency. This is because such a currency does have an established
relation to an existing accepted currency, even if there is a ”middleman” in-between: the
other embedded currency. The use of embedded currencies has been common in mobile
games for a long time. But its more recent introduction into the RRMs of high-profile PC
and console games such as Star Wars: Battlefront II (Electronic Arts and DICE 2017) is
likely a cause of the outcry in the gaming community, spawning articles with titles like ”This
game is a Star Wars-themed online casino designed to lure kids into spending money” and
”Why EA is Wrong to Say That ‘Star Wars Battlefront II’ Loot Crates Aren’t Gambling”.
(Rune K. L. Nielsen and Paweł Grabarczyk 2019)

Categorizing Games According to their Random Reward Mechanisms

Which types of games include which gambling mechanisms and to what extent? First,
games need to be differentiated based on the relation of RRMs to embedded currencies,
since chance and things of value are central to the definition of gambling. It is not enough
to talk about whether embedded currencies play a part in a game’s RRMs or not. It should
also be specified if the involvement concerns only the eligibility condition, only the reward,
or both. (Rune K. L. Nielsen and Paweł Grabarczyk 2019) Taking into account the dis-
tinction between embedded and isolated value, and the distinction between the value of the
eligibility condition and the value of the reward, Rune K. L. Nielsen and Paweł Grabarczyk
(2019) classify games containing RRMs into four categories: ”I-I”, ”I-E”, ”E-I”, and ”E-E”
(see Table 1).

–5–
 Requirements
Type (for achieving the Reward Example
eligibility condition)
Isolated Isolated Horizon Zero Dawn,
I-I
(non-purchasable) (non-sellable) Diablo 1, Diablo 2
Isolated Embedded (virtual Diablo 3 (with
I-E
(non-purchasable) sellable object) auction house)
Embedded (real Isolated Overwatch, Star
E-I
money purchase) (non-sellable) Wars: Battlefront II
Embedded (real Embedded (virtual PUBG, Team
E-E
money purchase) sellable object) Fortress 2, CS:GO

Table 1: Different implementations of RRMs (Nielsen and Grabarczyk 2019)

For example, an RRM implementation where an embedded resource is required for achiev-
ing the eligibility condition, and the reward is isolated, is considered an ”E-I RRM”. A game
containing such an RRM may be considered an ”E-I game”. A game may contain multiple
types of RRMs.

For example, PUBG (PUBG Corporation 2017) contains multiple types of virtual ”crates”.
One such crate can be purchased with ”Battle Points”, which are only earned by taking vir-
tual actions within the game such as winning matches or killing players, an isolated resource
required for achieving the eligibility condition. Another crate can only be opened using a
virtual ”key”, which can only be purchased with real currency, an embedded resource. Both
of these crate types trigger a random event when opened, which results in a reward con-
sisting of virtual cosmetic items such as clothes. These items can be sold for real currency,
making them an embedded reward.

When categorizing a game with multiple types of RRMs, only the ”most significant” type,
i.e. the type of RRM which is most similar to gambling, is considered. E-E RRMs are
functionally identical to gambling. E-I RRMs are less similar to gambling than E-E RRMs,
because E-I RRMs, unlike gambling and E-E RRMs, do not result in embedded rewards. E-I
RRMs are more significant than I-E RRMs because the monetary loss in gambling (possible
in E-I RRMs, but not I-E RRMs) is by far the more significant aspect of gambling compared
with the monetary gain (possible in I-E RRMs, but not E-I RRMs), especially when con-
sidering gambling addiction and its negative consequences. I-I RRMs are the furthest from
gambling since they do not include any real monetary aspect. Taking this hierarchy into
account, PUBG (PUBG Corporation 2017), for example, is considered an E-E game, not an
I-E game, since the E-E RRMs the game contains are more significant than the I-E RRMs
the game contains.

Audiovisual Representation of RRMs in Video Games

Another difference between games that contain RRMs is their audiovisual representation.
Some games use explicit iconography to inform (or misinform) the player about the game’s
random elements.

For example, a virtual die being thrown indicates not only the outcome probabilities (e.g.

–6–
a one in six chance of ”rolling” a ”six”, resulting in the most valuable reward) but also the
independence of the probabilities. A virtual die-roll indicates that, like a real die-roll, the
probabilities do not change at all between rolls. Conversely, representing randomness via a
virtual deck of cards would indicate that for every time an undesirable card is drawn from
the deck, the probability of drawing a desirable card increases (the probabilities of each
draw are dependent on the previous draws). The audiovisual representation has a big effect
on player psychology, as will be discussed in more detail later.

This representation can also be used to trick or fool the player. Unlike a real die roll or card
draw, the actual random procedure used does not have to relate to its physical or virtual rep-
resentation. For example, a virtual six-sided die could be rolled, even though the probability
of one of the sides is larger than another. Or a draw from a deck of cards could be shown
to the player even though the outcome probabilities are completely independent of previous
draws.

Some of these tricks might be easily found out by the players through testing. For example,
players could record the results of all their die rolls and notice that out of 6000 rolls, one of
the sides of the six-sided die was rolled only 500 times. However, the audiovisual represen-
tation and random procedure could also be designed to make any trickery almost impossible
to find out. For example, there could be no audiovisual representation of randomness at all,
just a check of the eligibility conditions followed by a presentation of the rewards. Having
no way to easily guess at the random procedure, players will come up with their own as-
sumptions, which may or may not be correct. Still, with a large enough amount of players
using the RRM and recording their results, patterns will emerge and the probabilities can be
found out through testing with a large sample size.

A developer seeking to prevent this to continue their deception could go further. Reducing
the number of times the same RRM is used in a given time frame makes it very difficult
to come by a large-enough sample size to make definitive statements about the random
procedure that is used. The random procedure could be varied between players, for example
by turning each user’s unique identifier into a number and using that number in the random
procedure’s algorithm. This would mean that any information about the random procedure
found out by some players would not be very useful for others. Certain user data could even
be used maliciously in the algorithm of the random procedure. For example, the majority
of users, who do not spend very much real money, would experience a random procedure
that is favorable to them. Meanwhile, users spending the highest amounts of real money
are assigned to a different algorithm, which makes them lose much more often, resulting in
larger profits for the developers of the game. Looking at the results of other players, these
high-spenders might continue to play, believing that they are simply on a run of bad luck.
Also, high-profile players, such as content creators who share their experience of the game
via video streaming services such as ”Twitch” or ”YouTube”, could be picked out by hand
and given a favorable algorithm. This could result in many players believing the RRM’s
random procedure is more favorable to them than it is because all they are seeing is the
”luck” these creators are having and sharing.

This paragraph contains the personal, subjective opinions of the author of this paper. I
strongly condemn these malicious practices, which can financially ruin affected players.

–7–
But, I believe it is important to discuss and explain these mechanics so that they can be
identified and avoided. I believe any software which combines real money with a random
procedure (I-E, E-I, and E-E games) should be required by law to include the exact details
of the random procedure within the software, and the software should have to be submit-
ted periodically for auditing so that this information can be verified. In the following, the
use of the word ”malicious” is meant to indicate the behavior of game developers that is
purposefully antithetical to the financial interests of the players, behavior which I condemn.

GAMBLING CHARACTERISTICS FOUND IN VIDEO GAMES

Returning to our original question, which types of games include which gambling mecha-
nisms and to what extent? This section will discuss some characteristics of gambling which
are believed to explain why people gamble and why they can suffer from gambling addic-
tion.

The Gambler's Fallacy

This bias occurs when ”the expectation that the probability of winning increases with the
length of an ongoing run of losses”. (Rune K. L. Nielsen and Paweł Grabarczyk 2019) Put
simply, a player might believe they have an exorbitant chance of winning because they have
not been very lucky so far. In the context of games, the similarity is straightforward: a player
easily starts to overestimate the chances of receiving the ”loot” they want after the last few
”loot boxes” they opened resulted only in undesirable rewards. (Rune K. L. Nielsen and
Paweł Grabarczyk 2019)

Developers can design audiovisual representation and random algorithms in a variety of

ways to prevent or exacerbate the gambler’s fallacy. A virtual die roll can be used to inform
the player that each time the random procedure is executed, the probabilities are indepen-
dent of previous executions, i.e. each loss does not increase the chances of a win thereafter.
The die metaphor can be helpful because the die is a universal symbol of independent ran-
domness since each side of the most commonly known dice appears equally as frequent as
any other side, independent of previous rolls. This will not eliminate the gambler’s fallacy
entirely, since even players of real-life dice gambling games (such as ”Craps”) can experi-
ence this bias. Other metaphors that can be used to indicate independent probabilities are
roulette wheels and coin flips.

On the other hand, a random procedure could be used which does increase the chance of a
win after several losses. In this case, the gambler’s fallacy would not apply, since the prob-
ability of different events is truly dependent on the previous outcomes. Here, developers
may consider using a virtual deck of cards to accurately represent the dependency of prob-
abilities on previous draws. The opening of a virtual card pack can also be used to indicate
dependence, though this is a more obscure metaphor.

To clarify, real collectible card packs (e.g. card packs from the collectible trading card game
”Magic: The Gathering”) contain real cards which must be printed and distributed. Since
no particular card or set of cards is infinite in reality, opening a pack and not finding a par-
ticular card increases the chances of finding that card in the next card pack. For example,
imagine all ”Magic” cards of type ”Rare” (a particular type of card which, as the name sug-
gests, are printed in lower amounts) in existence which are still contained within unopened

–8–
card packs (say, one ”Rare” card per card pack). Opening a card pack and receiving the
”Rare” card inside would be functionally the same as drawing one card from an enormous
deck of all existing unopened ”Rare” cards. This is functionally identical to the dependent
probabilities of draws from card decks which have already been explained. Depending on
the number of cards printed the effect of one draw on the outcome of future draws might
be absurdly small, though still existent. In practice, virtual card pack openings are meant
to simulate real card pack openings in terms of physicality and ”feel”, but do not simulate
the probability dependence described above. For example, opening a card pack in the video
game Magic: The Gathering Arena (Wizards of the Coast 2019) does not reduce the prob-
ability of finding one of the cards found in that pack in a later pack. This is an example of
an inaccurate application of a metaphor, though in this case, it is not malicious, but simply
easier to program.

Dependence can also be applied not based on the outcomes of previous draws, but simply
on their number. For example, in Magic: The Gathering Arena (Wizards of the Coast 2019)
a ”Rare Wildcard” (a particularly desirable card) is awarded to the player in addition to their
normal rewards, after every sixth card pack they open. However, this does not dispel the
gambler’s fallacy since past losses do not increase the winning chances more than past wins
and the player might still believe that past losses do increase their winning chances more
than past wins.

A malicious developer can use a metaphor, such as drawing from a deck, to make the player
believe their past losses are increasing their winning chances, even though the random pro-
cedure does not work that way.

Reverse Gambler's Fallacy

Parallel to gambler’s fallacy, reverse gambler’s fallacy occurs when the player believes,
after a run of wins, that their winning chances are higher than they are. This is based on
the intuitive and spontaneous application of Bayes’ theorem. Bayes’ theorem describes the
probability of an event based on prior outcomes.

Put simply, if one flips a coin 1000 times and it comes out heads 475 times, one would
predict the future possibility of a heads outcome from one coin throw as 47.5%. If one
flipped another coin 1000 times and it came out heads 100 times, rather than assuming a run
of bad luck, one might argue that the coin, which exists as a real object, might be imperfect or
”loaded” differently than would be expected of a ”normal” coin. Rather than assuming that
a coin comes up heads with the same probability as coming up tails, knowledge of previous
outcomes would be used to infer that this coin has a probability of coming up heads around
10% of the time.

This can become a fallacy when this inference is made with a sample size that is too small.
For example, a ”Craps” player might win three times in a row and be convinced that the
dice are loaded in their favor (through some impurity in their production), even though it
was just a run of good fortune. In a video game, a reasonable player would not believe the
virtual dice are ”loaded”, but they might believe that the random algorithm is purposefully
weighted in their favor (for example because the developer uses one of the tactics described
above to use different random algorithms for different players) or they might believe that

–9–
there is a flaw or ”bug” in the code of the algorithm.

Either way, the outcome is similar to the gambler’s fallacy, and similar strategies can be used
to avoid or exacerbate it (see above). To prevent reverse gambler’s fallacy, a developer might
consider publishing the source code, at least of the random procedure, so that players who
can comprehend the code can convince themselves and other players that there is no ”bug”
or purposeful ”loading of the dice”. A malicious developer could induce reverse gambler’s
fallacy by developing an algorithm to purposefully create winning streaks as long as the
player is only inputting small amounts of money, only to make them lose at the first big bet.

Near Misses

Some slot machines are designed to artificially produce situations where the player experi-
ences nearly winning by, for example, showing the winning symbols just above or below the
losing ones that the player actually got. (Rune K. L. Nielsen and Paweł Grabarczyk 2019)
This is supposed to encourage the player to try again and can make them overestimate their
chances of winning.

This same mechanism is arguably also at play in games like Star Wars: Battlefront II (Elec-
tronic Arts and DICE 2017), where players first get to see how rare the rewards they are
about to receive are, before it is revealed what the actual rewards are and if the players
already own those rewards (and thus will not benefit as much from them). (Rune K. L.
Nielsen and Paweł Grabarczyk 2019) To reduce this effect, developers can use metaphors
that do not emphasize what was missed. For example, drawing from a virtual deck of cards
only shows the rewards actually received by the player. Virtual dice can be represented in
a variety of ways.

Slice & Dice (tann and a3um 2021), which is an I-I game, uses three-dimensional models
of dice: cubes, which are given a random trajectory and rotation when they are rolled. This
means that many of their sides are visible while they are ”in the air”, rotating around. And,
since the fixed viewpoint of the player is at a slight angle to the virtual game board, some of
the sides of the dice which have not been rolled are visible even as the dice are stationary.
This can bring about the near-miss effect, for example, if the result of your roll consists of a
lot of ”blank” sides (which do nothing), while you are shown many die sides like ”attack” or
”healing” (which would have helped you win the battle) which you did not roll. One could
also develop a representation of die rolling which is algorithmically scripted to always face
the best side (which was not rolled) towards the player’s viewpoint, so that it can always be
seen, even going so far as to swap the faces of the die around so that a side, which would
normally face the ”ground” when a certain result is rolled, can still face the player.

Conversely, Dicey Dungeons (Terry Cavanagh 2019), a similar game, avoids the near-miss
effect by only playing a sound effect of rolling dice and then showing the result of each die
as a two-dimensional image, meant to represent the die as viewed from the top, obscuring
all the sides which were not rolled, which might have been preferable outcomes.

A particularly malicious developer might represent the random procedure in such a way
where results that are impossible to achieve with this particular random procedure are shown
next to the actual result achieved by the player.

–10–
Losses Disguised as Wins

”Multi-line slot machines” introduced a subtle yet radical innovation. By allowing players
to control the number of lines they are betting on, along with the amount they bet, players
experience winning more, even though they may be steadily losing. The fact of losing is
masked by a new kind of ”quasi-winning” or ”losing disguised as winning”. In the context
of video games, purchasing an RRM and receiving a ”Common” reward might be viewed
as winning (they did ”win” a reward), when it is in fact quasi-winning (and actually losing).
(Rune K. L. Nielsen and Paweł Grabarczyk 2019)

For example, opening a ”Case” in Counter-Strike: Global Offensive (CS:GO) (Valve 2012)
will always result in some kind of reward. But if the reward is a ”Common” item, the player
could have received the same item and kept more real money if they had simply sold the
”Case” and ”Key” (required to open the case) and bought the item directly (CS:GO is an
E-E game where nearly every item, including RRM eligibility requirements and rewards,
can be bought and sold for real money).

Quasi-winning is possible in every type of RRM-containing game, though it is most easily

spotted in E-E games since both the resources required to start the random procedure and the
reward received from the random procedure can be converted into real money and compared:
A simple loss would be a requirement of some positive amount of money and no reward
at all, while quasi-winning would occur whenever the amount of money the requirement
represents is higher than the amount of money that can be converted from the reward.

To avoid losses disguised as winning, developers can display the difference of value between
requirement and reward when the reward is received, so that it is always obvious to the
player when they have won and when they have lost. Conversely, a malicious developer
could obscure the comparison as far as possible. For example, an E-E game might use
different virtual currencies for requirements and rewards, which are not easy to convert to
real money or to each other at a glance (e.g. $0.754 equals 1000 ”Red Rings”, which can
only be used to purchase objects required for RRMs; $3.14 equals 1000 ”Scarlet Spheres”,
which can only be received by selling RRM rewards).

Cognitive entrapment

Sometimes referred to as the ”sunk cost” fallacy, this bias describes a decision-making
heuristic where a player escalates their commitment to a previously unsuccessful course
of action to justify these past investments. If a player spends $5 on RRMs and does not
receive the reward they were hoping for, they face the choice of either accepting the loss
(”admitting their mistake”, which is psychologically difficult) or spending an additional $5,
$10, $100, etc. to try to recuperate the initial loss. Once a player has started down this path
it can be difficult to stop. (Rune K. L. Nielsen and Paweł Grabarczyk 2019)

This trap is especially dangerous in E-I and E-E games, since real money is being spent,
perhaps a lot of it in a short amount of time. And since the initial investment can be directly
interpreted as money, it can be a lot harder to accept the initial loss. I-I and I-E games
are less dangerous in this respect. The isolated nature of the eligibility requirement is less
tangible, so it might not be able to result in such a large loss. For example, whereas money

–11–
can be loaned from friends or received by selling valuables, isolated requirements like time
spent in the game or skill-based achievement such as virtual ”kills” cannot be easily spent
in a short time. Likewise, not being able to assign a monetary value to the loss the player
experienced can make that loss easier to accept.

Random procedures which are dependent on previous outcomes or previous investment can
be especially liable to cause cognitive entrapment. For example, imagine a virtual deck of
cards consisting of one high-value card and 20 worthless cards. Drawing a card costs $1
and if you draw the high-value card, you keep it forever, but every 24 hours the deck is
reset and re-shuffled (randomized). If a player has already drawn 5 cards, they would be
inclined to keep drawing, because if they stop now, their investment is lost (if they wait
until their next paycheck, they will have to spend $5 again just to get to where they were).
A malicious developer could even hide the number of worthless cards in the deck, so that
the player never knows what their chances are of receiving a desirable reward, only that if
they stop now, they lose their entire investment.

A developer that wants to prevent cognitive entrapment might instead choose a random pro-
cedure that is independent of previous outcomes, like a die roll. Or they can be even more
proactive in fighting cognitive entrapment by reducing the value of the reward each time the
player uses the RRM, so that big cost sinks are discouraged. Magic: The Gathering Arena
(MTGA) (Wizards of the Coast 2019) is a good example of this. Players have multiple
”quests” (objectives tracked across multiple matches of MTGA, e.g. ”Attack with 30 crea-
tures.”) and two ”win counters” (tracking match wins in a single week and in a single day,
respectively). Every day, each player gets a new list of quests and their daily win counter is
reset to zero. At the start of each week, their weekly win counter is reset to zero. Quests are
relatively simple and can be completed in just a few matches. After completion, each quest
gives a reward, and then cannot be completed again that day. Win counters give rewards
for certain numbers of wins per day/week, and these rewards become smaller and smaller
as more wins are accumulated in a day/week until there is no reward at all for winning a
match. This means that on any given day a player might be inclined to play a few matches
to complete their quests and receive the largest win rewards, but is then disinclined to con-
tinue. The risk of cognitive entrapment is greatly reduced since continuing to play the game
becomes less and less rewarding.

Illusion of control
Research shows that even in games of pure chance, like lotteries, people are more likely to
overestimate their chances of winning if, for example, they are allowed to pick their lottery
number themselves. (Rune K. L. Nielsen and Paweł Grabarczyk 2019) This tendency is
seen even more clearly in sports betting, horse betting, and the like where people have been
shown to falsely believe that they have a better than random chance at predicting winners
and losers. (Rune K. L. Nielsen and Paweł Grabarczyk 2019)

In video games, a similar effect might arguably be observed when the player is offered the
choice between different ”loot boxes”, ”crates”, or ”packs”. (Rune K. L. Nielsen and Paweł
Grabarczyk 2019) For example, MTGA (Wizards of the Coast 2019) has multiple different
”card packs” available, each of which can have different cards. Spending some time thinking
about which cards the player wants and which pack is most likely to give them those cards

–12–
can make them believe that they are in total control even when the rewards in the packs are
still random and the likelihood of actually receiving exactly the cards you desire is quite
low.

CONCLUSION

There are still many problems with the concept of video game addiction. Unfortunately,
the current state of research is inconsistent and lacks basic foundational work, especially
concerning categories for diagnosis. The proposal of this new disorder might encourage
research that is aimed not at actually exploring this new concept, but at confirming previous
proposals, which are theoretically unfounded, while the concept itself remains inconclu-
sive. Viewing video gaming through the myopic lens of addiction without solid foundational
knowledge is premature at best and actively harmful at worst.

Instead, this paper has looked at how video games relate to gambling addiction, which as a
disorder is much more clearly understood. Video games should not all be conflated, espe-
cially as they relate to addiction and gambling. This is why clear categories were used to
differentiate video games and how the random mechanisms within interact with the broader
economy. While some video games can be considered structurally identical to gambling,
others are nowhere close.

The main part of this paper has discussed in detail the exact characteristics of gambling
and how these relate to video games. For each characteristic, it has been explained which
types of video games exhibit this characteristic and in which types of video games the most
dangerous effects of these characteristics lie. The perspective of video game developers
has been outlined, how a video game developer might avoid, bring about, use, and abuse,
each characteristic. From the perspective of a player, it is especially important to know how
certain games created by malicious developers can exploit the player, and how to notice
these mechanics. For each characteristic, some practical examples of existing video games,
where that characteristic occurs or is specifically avoided, have been included, if possible.
Hopefully, this information can be useful to video game players and developers.

BIBLIOGRAPHY
Activision and Sledgehammer Games. 2014. Call of Duty: Advanced Warfare [PC].
American Psychiatric Association. 2013. Diagnostic and Statistical Manual of Mental Dis-
orders, Fifth Edition.
Anthony M. Bean, Rune K. L. Nielsen, Antonius J. van Rooij, and Christopher J. Ferguson.
2017. “Video Game Addiction: The Push To Pathologize Video Games.” Professional
Psychology: Research and Practice.
Blizzard Entertainment. 2016. Overwatch [PC].
Blizzard North. 1997. Diablo [PC].
Electronic Arts and DICE. 2017. Star Wars: Battlefront II [PC].
Jagex. 2001. RuneScape [PC].

–13–
Mark D Griffiths, Daria J. Kuss, and Daniel L. King. 2012. “Video Game Addiction: Past,
Present and Future.” Current Psychiatry Reviews.
PUBG Corporation. 2017. PUBG: Battlegrounds (PUBG) [PC].
Rune K. L. Nielsen and Paweł Grabarczyk. 2019. “Are Loot Boxes Gambling? Random
Reward Mechanisms in Video Games.” Transactions of the Digital Games Research
Association.
tann and a3um. 2021. Slice & Dice [PC].
Terry Cavanagh. 2019. Dicey Dungeons [PC].
Valve. 2012. Counter-Strike: Global Offensive (CS:GO) [PC].
Wizards of the Coast. 2019. Magic: The Gathering Arena (MTGA) [PC].
World Health Organization. 2018. International Classification of Diseases, Eleventh Revi-
sion.

–14–