Preserving and providing meaningful access to any traditional media is based on a history of knowledge and practice. Initially through the process of trial and error and then through scientific principles, people have been conserving traditional objects for hundreds of years. This time span allows for models to be created: models for definitive collection building, models for creation and authority, and models for use. However, videogames do not have the luxury of history, their modes of production and use are either in flux or are difficult to formally represent, and there is very little time to “hope for the best,” in terms of preservation. While it is still possible to look at drawings on cave walls from 35,000 years ago, for example, or Roman frescoes, and in some cases we can even look at preliminary or ancillary materials related to those artifacts, there is a significant risk of losing basic access to a digital file as soon as 10 years after its initial creation (Conway, 1999).
Because videogames do have a significant digital component, they tend to rapidly become inaccessible or irretrievably lost. With funding from the NEH and IMLS, scholars in the related field of new media art have produced numerous theoretical and practical tracts with which to work, including the development of a notation framework for new media art (Rinehart, 2004), a systematic review of emulation as a strategy for preservation of a multimedia work (Rothenberg, 2006), and the formulation of agreed-upon theories and methods for the preservation of variable media art (Depocas, Ippolito, & Jones, 2003). Furthermore, there have been multiple studies within the digital curation community focusing on emulation strategies for console games in particular (Guttenbrunner, Becker, & Rauber, 2010; Hedstrom, Lee, Olson, & Lampe, 2006). While these projects have achieved great success in terms of tool and theory development, they focus to one degree or another on an entity produced by an individual or a small team of authoritative creators, which has an end product that is relatively static with easily defined boundaries. This is not the case for modern videogames, particularly MMORPGs, which are often created by huge, geographically diverse development teams and have multiple technological dependencies including hardware, software, and network requirements. Furthermore, MMORPGs construct and depend upon social interactions that are difficult to formally model. Major research projects that have focused specifically on MMORPG preservation include federal grants from the Library of Congress' NDIIP to fund the “Preserving Virtual Worlds” project (McDonough et al., 2010b), which systematically examined preservation and representation challenges inherent in virtual worlds; and an IMLS funded project focused on examining the creation behaviors of videogame producers, developers, and designers (Winget & Murray, 2008).
The primary concern with the longevity of digital documents is the “viewing problem” (Besser, 2000). Unlike analog or physical information, which tends to exist independent of human involvement, digital information needs constant intervention to survive. History has shown that digital documents are problematic by default. Whereas we can actually look at the Sistine Chapel ceiling, painted 500 years ago, or play games such as go that was invented over 1,000 years ago, it is difficult if not impossible to simply view documents on 8-in. floppy disks created in the last 20 years, even if there has been an immediate, proactive role in preserving them. Without concerted effort on the part of archivists and preservationists, digital objects quickly become obsolete or inaccessible due to unforeseen, although anticipated, advances in information technology.
The second method of preservation is emulation, which can be either at the system or the software level. System emulation focuses on developing systems that mimic the hardware used to create or run the original artifact. By writing just a few hardware emulators, dozens of operating systems, thousands of applications, and millions of documents could be accessed (Rothenberg, 1999). Emulation is currently the most widely accepted digital preservation method for many kinds of digital artifacts, including variable media art (Rinehart, 2000, 2002) as well as videogames.
In 2006, Jeff Rothenberg reported his process for renewing Grahame Weinbren and Roberta Friedman's The Erl King (1982–1985), hailed as one of the first works of interactive video art. In terms of preservation, The Erl King was a combination of obsolete hardware, artist-written software, and custom-made components. In his article, Rothenberg (2006) described the technical processes involved in exhibiting the work for the “Seeing Double: Emulation in Theory and Practice” exhibit held at the Guggenheim Museum in 2004. In this exhibit, curators at the Guggenheim displayed seven original interactive and performative artworks created between 1960 and 2004, alongside emulated versions of the same work. Although The Erl King emulation was deemed a “success” in that the emulated version looked and acted like the original, Rothenberg's article leaves no doubt about the theoretical, technical, and philosophical complexities involved in emulating such a complex work. In addition to needing extensive knowledge of artistic intent, a lot of time, and a deep practical skill in low-level programming, the emulation team had to make important theoretical decisions regarding what was “important” about the work in terms of creation, reception, and exhibition.
Within the public sphere, a number of emulation projects are focused on games. The MAME architecture, which supports the emulation of many arcade games, is a prominent example (Salmoria, 2010). Additionally, a number of console emulators are available on the web, but most of these depend on hacked BIOS and thus are illegal. Legitimate platform developers often use emulation to extend the shelf life of their earlier games and to extend their user base. Sony's Playstation 3 has a PSX emulator, as does the Playstation Portable (PSP). Likewise, the Nintendo Wii provides access through its online store to a large selection of emulated versions of earlier console titles (Pinchbeck et al., 2009).
Even though emulation is the most widely recognized preservation solution, traditional emulation is problematic for a number of reasons, including usability and legal concerns, but the most egregious is the fact that an emulator itself can be considered a file format and hence is susceptible to the same preservation challenges that can befall any other kind of digital content.
Modular Emulation and the Universal Virtual Computer
The concept of the Universal Virtual Computer (UVC), developed by Raymond Lorie (2002a, 2002b) provides a viable solution to ensure that emulators will be consistently available in the future. A UVC is a virtual machine that was specifically designed to preserve digital objects held by libraries and archives. The method is based on emulation, although it does not require specialized hardware or full emulation. Instead, the UVC combines elements of migration and emulation in innovative ways: It calls for emulation in that the UVC is a platform-independent logical layer that can sit on top of current and future hardware and software, and it uses migration in that it specifies format conversion to “universal technology-independent formats based on XML-like specifications” (van der Hoeven, Van Diessen, & Van Der Meer, 2005, p. 197). Because both the UVC and the file formats are independent, they will be able to run on current platforms as well as those that have not yet been developed.
The National Library of the Netherlands (Koninklijke Bibliotheek, or the “KB”) has been a leader in the development of the UVC-based preservation method. Following the successful implementation of the UVC on a test set of images (van der Hoeven et al., 2005), the KB has continued to develop their emulation strategy for long-term digital preservation by focusing on modular hardware emulation, and in cooperation with the National Archive of the Netherlands, the KB recently delivered Dioscuri (http://dioscuri.sourceforge.net/), the first modular emulator designed for digital preservation.
Dioscuri is a modular x86 computer hardware emulator written in Java. It is component-based: A software surrogate, called a module, emulates each hardware component. Users can combine several modules, allowing personalized configuration of the resulting system. It also is possible to add new or updated modules to the software library, resulting in a more comprehensive base from which to work.
Dioscuri is still in active development, so it is difficult to evaluate; the most recent release, from September 2010, is Version 0.6.0. In a 2009 report to the National Library of Australia (Long, 2009), Dioscuri Version 0.4.0 was only functional with the MS DOS 6.2 and MS Windows 3.1 operating systems. It was very slow, and “none of the tested media files could be rendered sufficiently well to give a useful performance” (Long, 2009; p. 39). On the other hand, van der Hoeven, Lohman, and Verdegem (2008) stated that “old games like Chess, Ironman and PC versions of Tetris and Prince of Persia all work well” on Dioscuri Version 0.2.0. What constitutes “working well” is unclear, but because of its modularity and platform independence, the UVC-based preservation methodology represents a promising avenue of investigation for the digital preservation community.
One exciting project using the concept of modular emulation is Keep Emulation Environments Portable (KEEP), which was launched in January 2009. This is an international project that will attempt to build a prototype emulation access platform, specifically for games (Pinchbeck et al., 2009). Success for this project would go a long way toward addressing the difficulties of archiving large bodies of related material associated with most games. Opening up an obsolete gaming environment provides coverage for many more kinds of artifacts than simply the game itself. Users could run mods within this emulated environment, or they could choose different modules to access different aspects of the game: Some modules could support game play over operating system functionality whereas other modules could support code examination over graphical considerations (Pinchbeck et al., 2009).
Role of Significant Properties in Digital Preservation
Making copies of materials, as in the case of migration, or mimicking an artifact's original computing environment, as in the case with emulation, inevitably introduces some degree of loss (Yeo, 2010). Migration provides access to content rather than to layout or structure. Emulation focuses on surface reproduction, but changes the artifact's underlying computing environment. Neither of these approaches deals with input/output peripherals or network dependencies. Each of these methods is acceptable under certain circumstances, given an understanding regarding what qualities are important about an artifact or a class of artifacts, what characteristics must be retained, and what characteristics might be acceptably lost. These important qualities are variously called significant properties (Hedstrom& Lee, 2002), “salient features” (Allison, Currall, Moss, & Stuart, 2005), or “essential characteristics” (Hoffman, 2002).
The idea of preserving an artifact's “significant properties” is a central tenet of the archival tradition. It refers to the idea that archivists and other cultural heritage managers should be able to identify a document or collection's most important characteristics to make appropriate appraisal, collection, preservation, and access decisions. However, there are a number of different theories regarding what properties of different artifacts are, in fact, significant. In making a distinction between an artifact's form and its content, some writers have suggested that the most significant properties are those related to content rather than those that are related to appearance (DeRose, Durand, Mylonas, & Renear, 1990). While this work was focused on literary analysis and the development of the Text Encoding Initiative, this attitude also is present in library imaging projects, which attempt to capture “… the informational content of the original … no more, no less” (Kenney & Chapman, 1996, p. 8; as cited in Yeo, 2010). Virtually all of the large digitization projects have emphasized structured content over form (cf. Project Gutenberg, www.gutenberg.org), where the primary artifact is often stripped of all design and formal characteristics not relating directly to content. This focus on structured content also is present in the archival literature: The Effective Records Management Project at the University of Glasgow (Currall, Johnson, Johnston, Moss, & Richmond, 2002) recommended that digital preservation strategies focus on maintaining the structured content of records as opposed to their presentational form. According to Currall et al. (2002), this is in direct opposition to the preservation practices and expectations associated with paper documents (as cited in Yeo, 2010).
This attitude is not true for all archival scholars: In the 1990s, Bearman and Sochats (1996) explored the notion of digital records' “essential properties,” specifically as pertaining to a record's ability to stand up as evidence. Forgoing the language of “significance” or “essential,” they determined that a “complete” record is one that retains its content, structure, and context. Furthermore, they felt that the most important characteristics of a record were not solely related to its content but also in its connection to appropriate metadata, specifically that metadata that can verify the record's comprehensiveness, identity, and authority (Bearman & Sochats, 1996).
The archival literature often refers to a “record,” typically defined as
materials created or received by a person, family, or organization, public or private, in the conduct of their affairs that are preserved because of the enduring value contained in the information they contain or as evidence of the functions and responsibilities of their creator. (Pearce-Moses, 2005; cf. term: “archival records”)
The term implicitly refers to documents rather than to artifacts or published materials, but archival records can be in any format, including text, photographs, and motion pictures. Archival records are often collected and preserved for their evidential, or legal, value. While videogames are not archival records in the sense that they serve an evidential purpose, by collecting and preserving them, archival institutions are implicitly responsible for preserving them in such a way that they retain their comprehensiveness, identity, and authenticity as artifacts. There has been little formal research in the videogame literature to suggest that anyone knows precisely what it means to preserve a videogame's authenticity, although McDonough and colleagues recently won an IMLS National Leadership Award to examine that question (GSLIS, 2010). One would hope that the game's content, structure, and context would be preserved.
Finally, because archives traditionally were narrowly defined as being repositories for evidential records, the idea of significance and authority was limited to those users (e.g., historians or lawyers) who had a narrow set of needs, specifically that a record be “what it purports to be, and is free from tampering or corruption” (Duranti, 2002, p. 21). These are admirable goals for many kinds of artifacts, but not necessarily useful from a videogame perspective. For example, videogames are often “tampered” with, as in the case with different kinds of game modifications, and while “corruption” is a strong term, many games are released as “buggy,” with patches and new releases being the norm. Furthermore, like many digital artifacts, videogames are boundary objects (Star & Griesemer, 1989) that anticipate multiple user groups, which have diverse needs and varied goals. For example, different users accessing the same game might want to play the game and have an “enjoyable” experience whereas other users might want to play the game in its original form; finally, some users might want to access the game code to see the game's inner workings. The idea of authenticity, therefore, must become more expansive, based on user needs rather than archival expectations and traditional practice (Dappert & Farquhar, 2009; Yeo, 2010).
In 2006, Hedstrom et al. (2006) presented an experiment reviewing users' reactions to preservation artifacts. In this article, which dealt with normal text documents as well as an early British videogame (Chuck E. Egg), users interacted with original, migrated, and emulated versions of the materials and were then asked their opinions of each. This research showed that the concept of digital preservation of something like a videogame is more complex than originally was thought, and that some of the assumptions underlying digital preservation research are unfounded. Emulation, for example, did not preserve “look and feel” properties any better than did migration. Furthermore, while users recognized that they were losing some “important” aspects of the original game, such as the original, obsolete keyboard and the processor, they preferred to play migrated and emulated versions of the game over the original. The keyboard was unusual and difficult to navigate, the “slow and cumbersome” processor screen made interaction difficult, and the aesthetic experience was unpleasant. These negative characteristics were mediated somewhat by the processes of migration and emulation.
Although there are no other user-centered experimental research studies that have focused on videogames' significant properties, informal experience suggests that there is a preference for characteristics that might be considered “inauthentic” given traditional archival theory. Because processor speeds continually improve and there are no intuitive markers for “age,” or “rarity” in digital artifacts, it is difficult for modern users to recognize the value of digital originals. Compared to contemporary videogames, games produced only 20 years ago are slow, their graphics are pixilated, and the interactions are both tortuous and simplistic. Furthermore, older games are often much more difficult to play and have different game mechanics than modern players are used to. Part of playing Ultima I (California Pacific Computer Company, 1980), for example, is to get killed over and over again in the first encounter until the player realizes that this particular monster cannot be killed, and the only option is simply to run away. This is anathema to modern gamers who are typically not given challenges they are unable to master (Arnold, 2008). Providing reliable representations of these originals is therefore problematic because interaction with old games primarily evoke either a “so what” or an “argh” response, which overrides the ability of users to recognize the innovative aspects of the game. In the words of one academic blogger, “I can no longer assume the game [Ultima IV] will make its case for greatness all by itself” (Abbott, 2010). What does this mean in terms of videogame collections within research institutions?
Future research must address this reality of varied user needs and expectations, particularly as it pertains to complex digital interactive systems such as videogames. Instead of relying on traditional notions of authenticity and significance, institutions need to develop multiple models, based on extensive user study, which will support utility over static, evidential representations. Archivists understandably want to provide the most authentic, reliable representation of the artifacts under their care. Unfortunately, what constitutes authentic and reliable is no longer fixed, and necessitates the development of new models.
In their experiment to explore emulation strategy for videogame preservation, Guttenbrunner et al. (2010) provided some commentary on this question, albeit tangentially. In their experiment, they created a taxonomy of significant properties for console videogames and tested various games against commercially available emulators to see how well the different methods maintained the significant properties. These properties were defined as those characteristics related to (a) object characteristics such as speed, graphics, audio, and network support; (b) infrastructure, those characteristics related to legality, stability, and the need for peripherals; (c) process characteristics such as usability and configurability; (d) costs related to preserving the gaming environment; and (e) context and data characteristics, including metadata referencing the game system and particular applications. In terms of this literature review, the most interesting finding was that the highest scoring preservation method was to simply film game play. Although interactivity was completely lost, which automatically ruled it out as a viable preservation method because of the setup of their evaluation system, the video option was the highest scoring and most reliable representation of the game as a whole in all other respects. This finding is not particularly surprising, although it is gratifying to see a formal study supporting intuition. For nearly a decade, Henry Lowood (2002) has called for the development of what he termed “game performance archives,” arguing that these materials represent game culture and interactivity in a way that the system alone can not.
As mentioned earlier, the studies referenced here focus on the preservation of relatively static artifacts that have a traditional creation model. Console and arcade videogames and interactive art objects such as The Erl King have clear-cut boundaries, and it is still possible for institutions to identify technical approaches to technical problems. While there are some aspects of digital preservation that are problematic even for these “simple” artifacts, emulation and migration still appear to be reasonable—if somewhat problematic—preservation methods because of their static and stable nature. This is not necessarily the case with the more complex, modern videogames such as World of Warcraft, which have multiple technical and network dependencies and have highly variable boundaries. It is exceedingly difficult to formally define where an MMORPG begins and ends, how it differs from its augmentations and surrogates, and the role of the players in game creation and representation. Emulating the hardware and software will only go so far in truly preserving this kind of work and would never be able to reproduce the entire Internet environment necessary for comprehensive representation. Research on MMORPG preservation is still in its formative stages, where scholars are still attempting to build models of what an MMORPG is, how to formally describe it, and how to represent the interaction of all its parts (McDonough et al., 2010a).