Private servers or "shards" have been a staple of game preservation efforts for years, and they've expanded the life of games beyond developer support for many gamers. Even for currently-maintained games, many servers are created to mimic games as they existed in prior versions, such as Everquest in its original form, or Ultima Online. However, these efforts are often closed and protected, seemingly as a competitive, capitalistic effort between server communities. Like the officially supported services before them, these servers collapse as people lose interest or they fail to keep backups of their work, as we saw with the schtserv server for Phantasy Star Online.
In light of these failures, it dawns upon me that even as people continue to keep these games alive, they risk being lost again because people do not document or publicize their efforts. As far as we know, schtserv's code is gone forever, and even though Phantasy Star Online has many projects to implement servers across all the versions, many of them are poorly documented and there's no central resource for anyone wanting to continue developing them.
At the same time, some games have had their source code leaked and this has allowed them to be kept alive long after their official services close. An example of this is GunZ: The Duel by the now-shuttered MAIET Entertainment. Despite the widespread availability of this leak, many third party servers keep their modifications of this leak private in attempt to compete with other servers for players, adding new features and game maps that aren't available elsewhere. This is of dubious legality and also ethically questionable, because the improvements that people now enjoy are bound to be lost in the same way that schtserv was.
I propose a unified effort to preserve these games by maintaining documentation of reverse engineered protocols and formats, supporting open source projects to keep the games alive, and setting standards for archival and preservation. Sadly, even with the new Digital Millennium Copyright Act exemptions for preserving games with online connectivity, this is still a violation. But I firmly believe that it is a cultural responsibility for us to preserve these works in any way we can, even if lack of revenue forces them to shut down, and even if new games make older ones obsolete.
This is a tricky minefield of issues to deal with. Many games that are still running have reverse engineering efforts; this is not without legal risk, as seen by the bnetd developers 17 years ago. That may have happened shortly after Battle.net's public launch, but even still it has a chilling effect on today. It is ethically questionable to develop for these games, and for a site seeking to be an authority on online game preservation, it may be necessary to reject efforts for those games, to an extent that protects the rest of the site.
As a method of ensuring that prospective server providers play nicely with preservation efforts, and also in the event that a DMCA notice forced a project to shutter, projects maintained by the site would likely be required to be licensed under the GNU Affero General Public License. Under this license, services hosted on AGPL code are required to provide the source, including any changes, to users who connect to the service. This is a marked difference from regular GPL which can still be used to host proprietary services, even with modified code. The viral nature protects the longevity of the source, so anyone interested in picking it back up after the original maintainers leave it can easily do so. Requirements may be loosened for adopted projects with longer histories on another license.
Leaked sources of games may be accepted under very specific circumstances, such as when the game is decidedly abandonware and no clear owner of the property exists.
GunZ is an example of this; since MAIET no longer exists and it is unclear that anyone still has the rights to that specific game, it might be okay to hold on to those leaks as long as licensed middleware is taken out. Updated: This appears to not be the case, as MasangSoft owns the GunZ property for distribution of GunZ 2.
Again, this is merely a proposal. I would love to see this come to life, but it would take a significant undertaking by multiple people to get it started. Perhaps this post will stir up some interest in a project.