Content uploaded by Janet Anderson
Author content
All content in this area was uploaded by Janet Anderson on Oct 17, 2014
Content may be subject to copyright.
Breaking New Ground: Innovation in Games, Play, Practice and Theory. Proceedings of DiGRA 2009
© 2009Authors & Digital Games Research Association (DiGRA). Personal and educational classroom use of this paper is allowed,
commercial use requires specific permission from the author.
Emulation as a strategy for the preservation of games: the
KEEP project
Dan Pinchbeck, David Anderson, Janet Delve,
Getaneh Otemu, Antonio Ciuffreda
University of Portsmouth
School of Creative Technologies, Middle Street,
Portsmouth, PO1 2DJ
dan.pinchbeck@port.ac.uk
Andreas Lange
ComputerSpieleMuseum
Marchlewskistr. 27 D- 10243 Berlin
lange@computerspielemuseum.de
ABSTRACT
Game preservation is a critical issue for game studies.
Access to historic materials forms a vital core to research
and this field is no different. However, there are serious
challenges to overcome for preservationists in terms of
developing a strategic and inclusion programme to retain
access to obsolete games. Emulation, as a strategy already
applied by major developers and the gaming community, is
introduced and the KEEP project, designed to create an
open emulation access platform is described.
Author Keywords
Games, preservation, emulation, archiving
WHY GAME PRESERVATION MATTERS TO GAME
STUDIES
The preservation of digital games is of vital importance to
game studies. As with any other field, a record of the
historical development of the medium and access to specific
artifacts within this should not be underestimated. This
extends beyond classic or important titles, to the vast
numbers of less well known or critically lauded games
released over the last thirty years.
Preservation of games tends to be piecemeal. National
libraries and archives do not currently have systematic
strategies for collection and certainly not for preserving
access and runtime functionality. Private collectors and
fansites have played a major role in providing access but
once again, these are not systematic and access remains a
problem. This paper splits the issue into two major
components: the necessity for creating archives of digital
games; and the barriers to successfully preserving access to
these.
Software development is rarely carried out from the ground
up, that is, without the re-use and adaptation of existing
technologies and techniques, and this is particularly evident
in the games industry. Dynasties of build engines,
middleware and plug-ins form an essential map of the
history of the medium. As an example of this, consider the
Source engine, used for titles such as Half Life 2 [24].
Source arose from the GoldSrc Engine, itself derived from
the original Quake engine [10]. It is thus linked into an
engine dynasty with the entire Quake series, dating back to
1996. Quake 4, however, was built using the idTech 4
engine, part of another dynasty which originates in the
original Doom engine [11]. Adoption of both engines has
been limited in comparison to a third dynasty, the Unreal
engine, whose most recent ‘children’ include Bioshock [1],
Unreal Tournament 3 [5], Gears of War [6] and Turok.
What makes one engine more frequently adopted than
another should be an interesting question for scholars
interested in the development of games from a variety of
perspectives from development processes, economics and
licensing to design, functionality and specific approaches
graphics, audio, multiplayer options and artificial
intelligence.
To extend this example, now consider an approach a
scholar may take to addressing this question. On one hand,
a film researcher interested in the influence of A Trip to the
Moon [17] could rely on secondary data: interviews, pieces
written by other scholars or journalists or critics, examples
of derivative material. But it is a highly questionable
approach to not actually engage with the primary data itself:
to not examine A Trip to the Moon as a piece of media.
Likewise, an understanding of the Unreal engine is always
going to be limited if the actual objects built using the
technology are not examined. Recent publications have
argued that the playing of games constitutes an essential
aspect of game studies [2], and beyond this there is the
under-represented but equally vital process of reverse
engineering build tools and component parts to understand
how they function. A robust understanding of why and how
the Unreal 1.5 engine was selected, adapted and applied,
and what impact this process had upon the final object that
is Deus Ex [12] becomes extremely difficult without access
to this primary data. On one level, verifying secondary data
is critical to avoid error, on the other, we cannot assume a
comprehensive exposition of the object will be available –
indeed, this is a completely unrealistic assumption to make.
The requirement to preserve access to both games
themselves and their build tools (and raw data) extends far
beyond the technology driven research suggested by this. It
has been argued, for example, that any cultural
interpretation of game elements should rest within the ludic
and technological constraints and circumstances that
underpin the objects themselves – that there is room for
structuralism as well as culturalism in game studies,
regardless of the dubious outcomes of the so-called
ludology/narratology debates. For example, the
visualization of agents in games is impacted by the
difficulties in real-time graphical representation. Hard
armor is less processor intensive than soft, flowing cloth; a
leather catsuit is much easier to deal with than Jedi robes.
Games are, after all, products of economy, whether the
constraints are monetary or technological. When asking
why games look as they do, then, understanding what they
are capable of, in system terms, is at least as important as
any cultural reading.
Running alongside this is the content-driven evolution of a
medium, including archetypal mechanisms of play as well
as symbolic and semantic representations of game elements.
Searching for an explanation of why Birth of a Nation was
shot in black and white and not engaging with the brute fact
that only black and white film was available is self-
evidently ridiculous; a similar process of understanding
historical constraints upon design is critical for game
studies. Content and construction are fundamentally
interwoven in this medium and in order to properly
understand historical objects, we must preserve access to
both the primary data and build tools. This is therefore in
addition to the power of using historical games as tools in
game education. Assigning students the task of developing
according to the constraints of an older, simpler system and
giving them access to the objects developed in this period to
understand how solutions were reached has clear pedagogic
power. This should be especially relevant now, with the rise
of mobile gaming suddenly undercutting the graphics arms-
race and returning us to games that have more in common
with Manic Miner [4] than Assassin’s Creed [23]. Likewise,
in a medium where the shoulders of giants form the basic
building blocks of development, understanding how
evolution has occurred by assessing historical artifacts
allows us to not only trace clear problem/solution pathways,
but to question assumed methodologies by identifying their
origins and processes of adoption. In short, to move
forwards with any aspect of game studies without keeping
the preservation of, and access to, historical artifacts is
near-sighted, self-defeating and, considered against the core
activities of other fields of media, art and cultural studies,
palpably absurd.
SPECIFIC ISSUES IN GAME PRESERVATION
Having said that, games are particularly difficult to preserve
and it is perhaps no surprise that so little has been achieved
when the full complexities of what it means to archive a
game are considered, let alone the technical problems with
retaining runtime functionality. In this section, we will offer
an illustration of just how complex capturing a full game
actually is, and offer some indication of the quantity of
supplementary data that may be of interest surrounding an
archived game, before moving on to how the KEEP project,
in particular, is tackling the second problem.
Assuming for a moment, that the core technical issue of
platform obsolescence can be bypassed, what exactly does
it mean to archive a game such as S.T.A.L.K.E.R.: Shadow
of Chernobyl [7]? The game was notoriously buggy when
released and six separate patches were subsequently
released to fix most major issues and add additional
functionality that was omitted in the initial release. On top
of this, the game could be purchased in disc format, or
digitally downloaded, each of which required separate
patches. Localisation meant translation of the substantial
text-based dialogue trees. A Collector’s Edition box
contained not only the standard game manual but additional
print such as a Zone Map and Survival Guide, and a DVD
containing supplementary images, text and video files. The
multiplayer option meant the establishment of servers, both
official and unofficial, which contain data about the history
of the online aspect of the game, not to mention potential
information about how these online games have been
played and whether they fit any generalisable pattern of,
say, deathmatch behavior (itself an understudied and
important question in game studies). Alongside all of this
official data, S.T.A.L.K.E.R. has been adapted and altered
by the modding community, adding new assets, tweaking
and fixing code and, in the case of Kanyhalos’ Oblivion
Lost [13], subject to major revision. If all this wasn’t
enough, the community of gamers has also added reviews,
discussions, walkthroughs, forum arguments, cheats and
hacks, not to mention that the proprietary XRay engine
developed for the game has evolved along with the sequels,
Clear Sky [8] and the forthcoming Call of Pripyat [9]. This
is not only a vast body of data surrounding a single object
(which, in itself requires 10Gb of hard disc space to store),
but it raises profound questions of what should be
prioritized in terms of preservation. For games ported to
several platforms the problem increases for each variation
on the game. It is necessary, for example, to preserve all
versions of the release – digital and DVD-ROM based – in
the original form, or final, patched, version? If the patches
are deemed important, as they presumably should be for
any scholar interested in the shift in development practice
towards releasing clearly unfinished games, then how are
these to be stored and what relationship should they have to
the artifact itself? Archivists are faced with a stark choice:
collect and archive everything, including multiple versions
of the same object; or make decisions about excluding
material strategically to make the process more feasible,
2
and risk consigning what may one day be important data to
the rubbish bin. Finally, there is the question of the XRay
engine itself. Unlike many FPS games, which fall into
dynasties of engines, GSC Gameworld created a proprietary
engine for the game, including features that do not exist in
these others, such as the dynamic A-Life engine. The issues
with intellectual property will be covered in a later section,
but for the moment, it is worth noting that alongside final
products and supplementary data, the tools and build data
for games is of equal value to future researchers.
Understanding, for example, how the X-Ray engine
functions; issues and advantages in developing using it
compared to other build engines; its use of middleware and
plug-ins; its handling of AI and rendering; the relationship
between scripted sequences, sandbox design and diegetic
and gameplay construction, all have a profound value to
scholars as well as future educators and developers. Put
another way, being able to access pre-compiled game data
enables a far deeper, greater understanding of the game as a
media artifact than simply playing or studying the final
build. Thus, game preservation should not only aim to
capture games and their surrounding data, but, wherever
possible, the tools and assets used to create them.
Technically, the problems do not get any easier.
S.T.A.L.K.E.R., like any other game, is reliant upon an
operating system with the correct system specifications to
run. The issue here is self-evident: operating systems, along
with hardware components, are superseded and become
obsolete. This is as true for consoles as home computers.
Obsolescence is a major issue in the preservation of any
digital artifact. Historically, the primary solution has been
to migrate such artifacts to current platforms to enable
continuing access to them. Migration effectively means
altering the code of an object to enable it to be rendered on
a non-obsolete platform [25]. However, migration
inevitably accelerates the process of bit-rot, or data
degradation, meaning that the life-span of migrated objects
is generally reduced. Further, migration is highly
inefficient, as every time it is required, every object must be
migrated individually [21, 22]. Equally, the process of
migrating an artifact such as a game is substantially more
complicated than a simpler object such as an electronic
document, or even audiovisual data. Put simply, migration
is of limited value in digital preservation generally, and of
extremely limited value in the preservation of games.
EMULATION AND GAMES
The alternative to migration is emulation, meaning that the
environment used to run the game in its original format is
recreated virtually on a contemporary platform. The focus
therefore is on the platform or enabling technologies
required the run the object, not on the object itself This is
left untouched, which has advantages in terms of not
contributing to data degradation. It also means that dealing
with obsolescence becomes an issue of creating new
emulators for platforms as they become obsolete, so large
numbers of objects can be served by more generic
emulators, streamlining the process of preserving access.
A number of issues have been put forward as arguments
against emulation as a core strategy for preservation. For
example, Phelps & Watry [18] have contended that a major
block is that emulation prevents searching within
documents. Whilst this is a concern, it demonstrates the
slant in digital preservation towards textual material and has
far less problematic implications for games. The complexity
of emulation systems (particularly for users, where lack of
technical knowledge may be prohibitive) is a bigger
problem. Do if the fact that the overwhelming majority of
current emulators rely upon specific platforms which are
equally vulnerable to obsolescence, and potential issues
with stacking emulators to reach an object. In other words,
emulating an environment from which to emulate another
environment within which to run an object is theoretically
possible but remains largely untested. Finally, as Bearman
[3] notes, many of the target environments for emulation
are locked within copyrights even after obsolescence. This
is particularly true for games console emulators. It is clear
that emulation is certainly not a ‘magic bullet’ solution, as
Bearman caricatures Rothman’s approach. However, the
notion of migrating games, even without considering the
general problems with migration, is clearly not feasible.
Even the most cursory conversation with developers about
the technicalities of porting games would make that
explicit.
It is not surprising, therefore, that emulation has been
adopted by the games preservation community more widely
than for other digital objects, with a large number of
solutions already existing in the public domain. The
MAME architecture, which enables the emulation of large
numbers of arcade games, is well established and well-
known [16]. There are a substantial number of console
emulators also in existence, although the majority of these
rely on hacked BIOS to function and therefore infringe
copyright law. Emulation is also used actively by legitimate
platform developers: Sony’s PlayStation 3 contains a PSX
emulator, as does the PSP, and the Nintendo Wii Store
offers access to a wide selection of previous console titles
via emulation and the purchase of a bespoke hardware
controller. Ensuring backwards compatibility through
emulation makes sound financial sense as it extends the
shelf-life of intellectual property and it seems likely that
this policy will continue.
The biggest problem, however, with all current emulators is
their own obsolescence, as each is built for a specific
platform and thus vulnerable to this being superseded. It is
this issue that the KEEP project aims to tackle directly.
THE KEEP ARCHITECTURE
In January 2009, the KEEP (Keep Emulation Environments
Portable) project was launched. Funded through the
European Commission’s Framework 7 program, KEEP is
being developed by an international consortium: the
3
national libraries of France, Germany and the Netherlands;
Tessella (UK/NL) and Joguin SAS (FR), software
developers specializing in preservation; project consultants
CrossCzech (CZ); the ComputerspieleMuseum (DK); the
University of Portsmouth (UK) and the European Game
Developers Federation. The first phase of the project,
lasting until 2012, aims to develop a prototype of an
emulation access platform to enhance the preservation of
digital objects, with a particular focus on digital games.
Unlike current emulator systems, KEEP is not built upon a
specific platform technology, but a virtual machine. This
follows the conceptualization of such a system by
Rothenberg [22] and Lorie [14]. The OLONYS system,
developed by Joguin SAS, is a series of virtual machines
stacked in order of complexity that will interface with, and
support a modular emulation framework [15]. Thus, at root,
KEEP is far more future proof than current solutions. It also
has the advantage of offering multiple emulation solutions
to any given artifact within a single, user-friendly interface,
allowing both bespoke manual configuration of the
emulation process and a more automated and simple means
of accessing artifacts for users not requiring this. On one
hand then, KEEP benefits from less reliance upon any given
platform and a modular architecture that enables
independently developed emulators to function within its
framework (meaning that existing components can be
integrated). On the other, it enables archivists and
researchers to bypass the traditionally complicated process
of installing and running emulation software.
Alongside the emulation access platform, KEEP will also
develop a transfer tools framework to enable new objects to
be integrated with the system. The existing archives of the
Bibliothèque nationale de France, Koninklijke Bibliotheek,
Deutsche Nationalbibliothek and Computerspiele Museum
will form the initial core of the available KEEP archive, but
this framework will establish a means for new objects to be
added. Further, research is currently being carried out to
supplement current metadata standards for archiving with
emulation metadata to ensure high compatibility with
existing international archives. Part of this process is
evaluating and enhancing existing metadata systems to
ensure maximum compatibility with games.
KEEP’s focus is on retention on existing emulation work
and enabling archives and users to transfer objects to the
KEEP architecture as seamlessly as is possible. Whilst
conceptually and technologically advanced, it is, at root, a
deeply pragmatic solution to an extremely difficult
problem.
If successful, the impact of KEEP upon game studies will
be profound. Although initially limited to localized access
in three European countries, a second phase of the project
(once proof of concept is established) may be to roll the
architecture out to other archives internationally, and to
explore potential public release of the system so individual
users can transfer their obsolete media into KKEP and
retain access. This is one means of circumventing the
normal copyright problems, as KEEP therefore functions as
an enabling technology for ‘home archiving’ to supplement
access to archived artifacts held in national storage.
In terms of both research and education, a modular and
open emulation platform goes some way to addressing the
difficulties of archiving the large bodies of material
surrounding most games. For example, the Dioscuri
emulator, which emulates x86 hardware and can run
Windows98 [26], has the potential to enable game patches
or build engines to be run within this native environment. In
other words, emulating hardware to provide native access to
the functionalities of obsolete operating systems provides
access to many of the other tools and data surrounding a
game, rather than simply providing access to the final
object itself. Equally, a modular architecture means that a
variety of emulators, each with particular strengths and
functionalities may be selected to access a particular object.
So a user seeking to simply emulate S.T.A.L.K.E.R. to
access the game may opt for an emulator that sacrifices
additional O/S functionality for increased performance,
whereas another looking specifically at the codebase behind
the game may choose to dispose of advanced graphics
emulation in favour of alternate functionality.
There is a caveat in all of this, of course. The emulators
themselves need to be written, and the KEEP consortium is
working closely with the existing emulator community to
try and maintain a high level of adaptability with the
emulation access platform and current emulators. The
metadata extensions require a careful balancing act between
what is both pragmatic and compatible for archives to
implement, and the high-level information required for the
modular emulation architecture to function intelligently.
There are outstanding issues with copyright protection in
regard to games that a legal study is exploring. Nor does
KEEP directly solve the issue of the large potential body of
supplementary information surrounding each game.
However, it is the first systematic, large-scale attempt to
solve the technical problems with access retention for this
medium and, as such, presents a major potential benefit to
the games research and education community.
CONCLUSION
Game studies requires systematic archiving of historical
titles. Otherwise it runs a serious risk of data loss. Personal
collections, fan archives such as Abandonia or Home of the
Underdogs and the rerelease of old IP through digital
distribution such as Playstation Store cannot and should not
be counted on to ensure access is protected for obsolete
titles. The preservation of games is a very difficult issue,
partially due to the large bodies of secondary artifacts
surrounding each release; partially due to big fixes and
patches, and partially due to the technological challenges of
ensuring run-time functionality. Emulation is the only real
solution to this challenge, but an open access architecture
based upon a virtual machine is the only means of future
4
5
proofing these emulators from the same cycle of
obsolescence as faces the original media. Hardware
emulation and a modular framework not only enables
original titles to be run, but offers access to codebases,
build engines, middleware and game assets, all of which, it
has been argued, are of potential equal value to future
scholars.
The KEEP project, like emulation in general, is not a
‘magic bullet’ solution. It cannot ease the secondary artifact
burden, nor can it resolve the problems with the
continuation of copyright post-obsolescence. What is does
offer, however, is the best current solution to retaining
access to obsolete games into the future. For this reason, it
should be of interest to any researchers interested in the
past, or the future, of our medium and field.
REFERENCES
1. 2K Boston + 2K Australia. Bioshock. 2K Games: PC.
2007
2. Aarseth, E. Playing Research: Methodological
approaches to game analysis, Melbourne DAC - the 5th
International Digital Arts and Culture Conference (2003).
3. Bearman, D. Reality and Chimeras in the Preservation of
Electronic Records. In D-Lib Magazine vol 5 no 4, (April
1999)
4. Bug-Byte. Manic Miner. Bug-Byte: ZX Spectrum, 1983
5. Epic Games. Unreal Tournament 3. Midway: PC, 2007
6. Epic Games. Gears of War. Microsoft Game Studios:
Xbox360, 2006
7. GSC Game World. S.T.A.L.K.E.R.: Shadow of
Chernobyl. THQ/GSC: PC, 2007
8. GSC Game World. S.T.A.L.K.E.R.: Clear Sky. Deep
Silver: PC, 2008
9. GSC Game World. ST.A.L.K.E.R.: Call of Pripyat. PC,
Q42009
10. id Software. Quake. id Software: PC, 1996
11. id Software. Doom 3. Activision: PC, 2003
12. Ion Storm. Deus Ex. Eidos Interactive: PC, 2000
13. Kanyhalos. Oblivion Lost. Uncommercial modification
for S.T.A.L.K.E.R.: Shadow of Chernobyl. PC, 2007.
Available from www.oblivion-lost.net
14. Lorie, R.A. Long-term archiving of digital information.
IBM Research report, IBM Almaden Research Center, San
Jose, Almaden, 2000
15. Joguin, V. Emulating emulators for long-term digital
objects preservation: the need for a universal machine.
Emulation Expert Meeting 2006, (The Hague, The
Netherlands).
16. MAME: Multiple Arcade Machine Emulator (1997-
2009). Information available at http://mamedev.org
.
Retreived 21/7/2009.
17. Méliès, G. A Trip to the Moon. 1902
18. Phelps, T.A., & Watry, P. A no-compromises
architecture for digital document preservation. ECDL,
2005: 266-277.
19. Propaganda Games. Turok. Touchstone: PC, 2008
20. Raven Software. Quake 4. Activision: PC, 2004
21. Rothenberg, J. Avoiding Technological Quicksand:
Finding a Viable Technical Foundation for Digital
Preservation. A Report to the Council on Library and
Information Resources. Washington: Council on Library
and Information Resources, 1999
22. Rothenberg, J. Using Emulation to Preserve Digital
Documents. The Netherlands: Koninklijke Bibliotheek,
2000
23. Ubisoft Montreal. Assassin’s Creed. Ubisoft: PC, 2008
24. Valve. Half Life 2. VU Games: PC, 2004
25. van der Hoeven, J., vanWijngaarden, H. Modular
emulation as a long-term preservation strategy for digital
objects. In: Proceedings of the 5
th
International Web
Archiving Workshop (IWAW05) held in conjunction with
the 8th European Conference on Research and Advanced
Technologies for Digital Libraries (Vienna, Austria,
September 2005
26. van der Hoeven, J.; Lohman, B.; and Verdegem, R.
Emulation for digital preservation in practice: The
results. International Journal of Digital Curation vol.2, no.
2 (2007), pp123– 132.
