Adding the Dimension of Time to HTTP

While the web is distributed, most web archives are centralized silos that do not cooperate with each other. This is partially because the technology that is necessary to replay the archived content and keep it from being influenced by material on the live web also makes it difficult for web archives to cooperate. The Memento Protocol (which we played a central role in defining) addresses this problem by defining an extension to the Hypertext Transfer Protocol (HTTP) that allows for standardized, machine-readable integration of both the past web and the present web. The Memento Protocol extends the concept of HTTP content negotiation to include not only well-known dimensions such as Multipurpose Internet Mail Extensions (MIME) types (e.g., JPEG vs. PNG) and file encodings (e.g., gzip vs. compress), but also the dimension of Coordinated Universal Time (UTC) as a universal versioning system. The protocol can be supported by all systems that hold temporal resource versions, including conventional web archives, as well as resource versioning systems such as wikis. The Memento Protocol introduces some standard terminology with which to discuss web archiving, the most fundamental of which are: original resource (the resource on the live web), Memento (an archived version of an Original Resource, frozen in time), TimeGate (a resource capable of datetime content negotiation to discover a temporally appropriate Memento), and TimeMap (a machine-readable list of all Mementos for an Original Resource). Furthermore, the Memento Protocol is the first web archiving API, enabling aggregation of access to disparate web archives. Web archiving has been dominated by the Internet Archive’s Wayback Machine, but via the Memento Protocol it is possible to leverage the more than a dozen publicly accessible web archives throughout the world for increased completeness, consistency, verifiability, resilience, and availability. 14