DH09 Tuesday, session 1: Preserving Virtual Worlds: Models & Community
[Again, live blogging with all its pitfalls and disclaimers. I almost certainly won't get most or all of the live discussion, in particular; if you remember the Q&As, please put those in comments.]
This panel is put on by members of the Preserving Virtual Worlds project, a multi-institutional collaboration “funded by the Preserving Creative America initiative under the National Digital Information Infrastructure Preservation Program (NDIIPP) administered by the Library of Congress.” (quoted from the PVW site)
First up, Matt Kirschenbaum, “Twisty Littel Passages Almost All Alike: Applying the FRBR Model to a Classic Computer Game.” The panelists, working as a group, chose interactive fiction because they thought it’d be easy: Adventure has a few hundred words, how hard could it be? Ha.
Adventure is a classic text adventure. “You’re standing at the end of a road before a small brick building. Around you is a forest. A small stream flows out of the building and down a gully.” From there, you can explore (go north, enter cave, inventory to see what you’re carrying) and act (talk to Grue, hit Grue with sword). (The grue is actually from another piece of interactive fiction, Zork.)
It’s a virtual world in a formal sense: a program that has a state; there are objects that can be possessed or not by characters in the program, there are spaces and places, it’s got a bunch of complexity!
Adventure has been versioned and ported numerous times (no, seriously, hundreds); the scoring system has changed over time, and is an indicator of the game’s history. It was adapted for the Atari 2600 and even for the iPhone!
1973: Will Crowther creates Adventure, which is, by the way, based on a real place (the Mammoth Cave). 1977: Don Woods gets a hold of Adventure and expands it. It’s actually Woods’ version that became popularized. The original Crowther version was lost for a number of years, but two years ago Dennis Jerz recovered the original game from a backup tape of Don Woods’ student account at Stanford! Hello, digital forensics. Jerz’ article “Somerwhere Nearby is Collosal Cave: Examining Will Crowther’s Original ‘Adventure’ in Code and in Kentucky” is in Digital Humanities Quarterly 1.2 (2007).
Jerome McDonough is next, talking about the group’s investigation of just how well recent descriptive practices adapt to describing games (in this case, interactive fiction). Do problems arise when we apply the FRBR model to Adventure?
[will get Jerry's FRBR slide, which summarizes the complexity of this system fantastically well]
A game doesn’t have a single component; even in Adventure you have at least a source file and a data file. More modern games — audio, video, etc. All of these components have different authors, were created at different times — very difficult to come up with a bibliographic description. Plus, where does the game end? It’s not just a source code file and a data file. You need to compile source file and put it into a data file, which is put into storage, on hardware with memory and an operating system and an executable program. So if we want to preserve games and not files, we have a big mess.
The game (say, the source file) has a working relationship with a working copy of the game (binary file). The distinction between those two entities is vital, and you must both distinguish between them in FRBR and indicate that they’re related (FRBR says that’s a translation).
When do you have a new work? That varies depending on your point of view. Is a port into a Windows version a new work? If you’re a programmer, this is definitely a new version, a new expression, a new work — significant intellectual effort went into it (this is what FRBR wants, for it to qualify as a new work). If you’re a gamer, the ported game is identical to the “original.” We need bibliographic systems that support both points of view depending on what the user wants.
FRBR does allow you to indicate various work-to-work relationships, so you could say that every new implementation is a new work. Or — and here Jerry doesn’t want to offend FRBR purists — how about a superwork? This carries its own difficulties: if there’s a superwork, and you’ve changed Adventure by adding spells and monsters and different gameplay (and creating Zork), why isn’t Zork part of the superwork? Maybe it should be; this is the sort of thing that drives catalogers insane.
The solutions? There’s a workload problem. We need systems that support incremental extension and refinement of description. We also need systems that support user generated metadata. Librarians alone can’t do this. There’s also the perspective problem: privileged perspective (game developers) vs. situated knowledge (gamers). Libraries and classification systems don’t support different points of view, but perhaps they should!
Next up, Henry Lowood, played by Jerry McDonough. “The Open Archival Information System Reference Model vs. the BFG 9000: Issues of Context and Representation in Game Software Preservation.” This talk uses Doom as its example. (It’s a shooter; you shoot everything in sight.)
Doom and its like are complex artifacts both in a technical and in a cultural sense. We’re not just preserving data files; we’re trying to preserve the ability to understand the artifacts both technically and culturally. Doom is a DOS program written in C; about 75 files including both C files and a bunch of header files. A single data file known as a WAD (where’s all the Data) file. Later the WAD file was abandoned in further instances of Doom.
Doom as a cultural production: remarkable that shareware distributed over the internet in the mid-90s would have such a huge cultural impact. But there’s a comic based on it; other versions of Doom have been created; other productions (Simpsons, Family Guy) have referenced it; there’s a major motion picture of Doom; a board game of Doom; Doom plush toys are created by rabid fans. And so on.
If we’re going to preserve this, how do we go about it? The development of the OAI System Reference Model is one of the most significant developments in digital preservation. It’s an information model, among other things: what do we expect to see in the preserved object. Two kinds of information: content information, and representation information. (Data and metadata.) Content is further broken down into structural and semantic information. Representation is broken down into, among other things, provenance and context information.
One very important concept in the OAI SRM is designated communit[y|ies]. If you’re a programmer, you don’t need much information to understand the C code; if you aren’t, you need all the background information on programming, and C in particular, to understand the game. Everyone, including humanists and librarians, has different levels of knowledge. What does this mean for representation and context information?
In the room, currently, one professional C developer. But about half the room would like to study Doom, even on the technical level. This isn’t a metadata problem, but a far trickier issue — a knowledge preservation problem.
Different types of context information for Doom: versioning (Doom 1, 2, 3), reuse (Doom characters in Leggo), use (actual gameplay), interpretation (secondary literature, for example).
Possible solutions: preservation scaffolding. We don’t know exactly what info every community is going to need. Many communities already generate this information; they just don’t put it anywhere safe. The Internet Archive has created a sub-archive for preserving video games. This is a space for anyone to contribute data, and IA will [attempt to] preserve it.
Games exist in a variety of different states that are of interest of different people. A lot of what we’re seeing above applies to other software applications.
Next: Kari Kraus and Rachel Donahue, talking about some of the social aspects that they’ve discovered on the PVW project. “Game Change: The Role of Professional and Amateure Cultures in Preserving Virtual Worlds.”
Community curation can be fantastically positive. Archives are usually maintained by professional archivists “to preserve and provide access to the records of government, industry and culture that have lasting value.” So what’s the gaming industry doing to create its own institutional archives? …Nothing, actually.
Game industry pointed to things like version control and configuration management policies. But version control repositories, for example, get emptied out once software is released.
What about archives in the game community? The [positive-sense] amateurs are doing a lot to preserve the games that they love, complete with metadata not unlike that in the OCLC. They provide context: “screenshots; digital surogates of manuals/boxes/media/feelies, etc; walkthroughs and reviews; speedplays, machinima, and other player-created video.”
A couple of projects that capture the neat things about amateur preservation. First up, C64 Preservation Project. Run by volunteers who aim to “archive pristine versions of original Commodore 64 software, including copy protection.”
There’s also releasing the unreleased. Game community has many collectors, and collectors LOVE rare items. And what’s more rare than a game that was never actually released? “When discovered, and with the proper equipment, these games can be released on native media.” Nifty.
In addition to preservation works, there are online forums with a ton of information on repair, system refurbishing, lists of game vendors, etc.
So preservation is happening, just not in the expected places.
That was Rachel speaking. Now, Kari’s turn. IP rights and donor agreements inform the somewhat dysfunctional relationship among the game industry, game players and game archives. Industry: MINE! Players: *sneak*. (Industry: *smite*.) Archives: look, we’re preserving YOUR history! (Industry: um, no.)
“Game industry and archivists rely on players to track the transfer of game assets. Developers rely on pirated copies and abandonware sites to access their content once it is no longer supported or maintained by the game industry. (Example: Robert Pinsky’s Mindwheel.)
“Academic institutions are beginning to maintain and preserve abandonware sites for vintage computer games, which often lack the resources to sustain their archives. (See Home of the Underdogs.)
“Players often augment and extend the cultural relevance of computer games by modifying them… These adaptations in turn increase the likelihood that professional archivists will eventually judge the games worthy of digital preservation. (Example: Adventure.)”
Archivists vs Players (paraphrased from slide):
- Comparatively low vs. high tolerance for variation;
- Digital preservation = reproduction vs. adaptation;
- Authenticity through stabilization vs. adaptive reuse;
- Authenticating function invested in the author vs. the user;
- Emulation and migration vs. emulation but also reprogramming, remixing, adaptation, re-creation, reimplementation.
Kari describes in detail the Mystery House Taken Over project. Too much detail for me to summarize; do explore the site, which is fascinating. But the point of this: context information, usually external, has been ingested into the game itself, and the game curates/exhibits information about its context.
There follows a video demo of Stanford folks preserving artifacts using the open-source Sirikata virtual world platform. An alpha version should be available later this summer. If the movie is available online, I’ll include a link to it later.