tripleC 9(2): 702-714, 2011 711
CC: Creative Commons License, 2011.
ing the most important game decisions), it is possible to keep the gaming activities under control
(Kabus et al., 2005). In other words, gaming activities on the server are trusted.
Other architectures such as peer-to-peer offer the developers no control over gaming activities
that are spread all over the network of peer computers. In other words, they are untrusted. All the
information manipulated by the game clients implies in fact a degree of risk, in terms of cheating. It
is easier to manipulate the code executed on the clients machines or on the peer nodes. In this
regard the client-server architecture offers more trust than the peer-to-peer, because of the control
that can be exercised by game companies. The client-server architecture is a trust relation that,
implies a terrritorialization in which there is an attempt to reduce the range of possible cheating
actions by exercising control over the execution of the game code: it is quite difficult to, for exam-
ple, modify and exploit the code of the software stored, executed, and manipulated on the server,
whereas by contrast, it is relatively easy to exploit the code stored and executed on the client. This
observation is valid in general: all the game information and code controlled and executed on the
client machines - including files, memory, drivers, services and so on - can in principle be manipu-
lated illegally (Pritchard, 2000).
Therefore, the client-server architecture possesses a specific spatial territorialization capacity in
which the game code is executed on the servers in centralized spaces, rather than being deterri-
torialized onto the player’s machines. In technical terms this is often referred to as 'centralization',
the idea that access, resources, and data security are controlled almost exclusively via the server.
Centralization places all trust in the server.
However, the spatial territorialization of the architecture (the centralization) can never be total.
Indeed, for performance reasons, all the game states cannot reside on the server. For example,
Hoglund and McGraw (2008, p. 142), describe the organization of the data structure
7
of a
MMORPG’s character, and state that “Clearly these data must be stored on the game server, but
sometimes the client program controls the values directly”. If the client controls some of the values
of the character’s data structure, then an expert programmer could easily manipulate these values
to obtain an unfair advantage (for example increasing a specific skill of that character, by manipu-
lating the values that refer to the skill). Another example comes from online Real-Time Strategy
games where sometimes it is possible to illegally manipulate the client information that controls the
"unexplored areas" of the map (Pritchard, 2000). This can bring an unfair advantage by allowing a
player to know the location of enemy units. In conclusion, we always have some degrees of deterri-
torialization of information in the client-server architecture, an information that becomes less trusted
therefore.
5.2. Deterritorializing Trust
One way that game companies can enhance the trust relation between their servers and the player
clients is by using technologies known as anti-cheating tools. Anti-cheating tools are software de-
vices that automatically enforce the terms of legal documents of a game: here we can investigate
an instance of the enforcement of statements/rules as way to build trust, as described before.
To understand the functioning of anti-cheating tools, we briefly describe the operation of a well
known anti-cheating tool known as 'the Warden' (Blizzard, 2005), used in the game World of War-
craft (WoW). Basically, when the player connects the client to the WoW server, the Warden is
downloaded from Blizzard servers onto the user’s client machine. The Warden is composed of
small portions of code that are dynamically assembled at each download. This means that each
Warden is different from one another and therefore it is difficult to create (cheating) code that can
circumvent it. Indeed, if a cheater 'captures' a Warden and creates a software countermeasure
then this measure will not be effective because the next Warden(s) downloaded onto the users’
machines will be different from the captured one (Hoglund and McGraw, 2008).
7
The data structure is a way of storing and organizing computer data.