This paper presents a streamlined metadata record format designed to support the permanence of network discoverable objects. It starts with the Dublin Core consensus and distills out a subset of four semantic buckets − a metadata kernel − that balances the needs for adequate identification of persistent objects and for low cost metadata generation. To minimize the burden of creating, understanding, and manipulating data in those buckets, a very simple record format has been designed, called an Electronic Resource Citation (ERC). The basic ERC can be parsed by two lines of Perl code. Beyond permanence support, the ERC design suggests quite a new path for the ongoing development of simple metadata; readers familiar with the current evolutionary challenges may find the ERC to be simpler, and yet more complete, compact, extensible, and international than the Dublin Core.

This document specifies GrabIt, an HTTP-based protocol for systematically transferring lists of digital packages. A package is an arbitrary sequence of octets (e.g., a single TAR archive) that can come with an associated content type indicating that extra steps (not central to GrabIt) may need to be taken to complete and validate a package before successful receipt can be reported. The sender issues an HTTP "post" with a list URLs corresponding to packages to "grab". In response, the receiver returns a job URL in an HTTP "Location" header that can be probed periodically to retrieve transfer status reports.

The ARK (Archival Resource Key) is the only persistent naming scheme for internet information objects that squarely faces the service issues underlying practical electronic permanence. The ARK is a specially constructed, actionable URL encapsulating a globally unique identity that is independent of the current service provider. Each ARK is by definition bound to three things: (a) object access, (b) object metadata, and (c) a faceted commitment statement. These service bindings provide the best way for service providers to communicate their varied commitments to objects. It is for these reasons that the California Digital Library is pursuing the use of ARKs.

The Bibliothèque nationale de France (BnF) looks back at lessons learnt over a decade of implementing persistent identifiers (ARKs). Those lessons give an insight into what it means to maintain persistent identifiers in the mid-term. A systematic gap analysis between what is and what should be, especially in a Semantic Web context, lead to open questions about best practices and standards compliance.

The Digital Preservation Program of the California Digital Library (CDL) is engaged in a process of reinvention involving significant transformations of its outlook, effort, and infrastructure. This includes a re-articulation of its mission in terms of digital curation, rather than preservation; encouraging a programmatic, rather than a project-oriented approach to curation activities; and a renewed emphasis on services, rather than systems. This last shift was motivated by a desire to deprecate the centrality of the repository as place. Having the repository as the locus for curation activity has resulted in the deployment of a somewhat cumbersome monolithic system that falls short of desired goals for responsiveness to rapidly changing user needs and operational and administrative sustainability. The Program is pursuing a path towards a new curation environment based on the principle of devolving curation function to a set of small, simple, loosely-coupled services. In considering this new infrastructure the Program is relying upon a highly deliberative process starting from first principles drawn from library and archival science. This is followed by a stagewise progressesion of identifying core preservable values, devising strategies promoting those values, defining abstract services embodying those strategies, and, finally, developing systems that instantiate those services. This paper presents a snapshot of the Program’s transformative efforts in its early phase.

The WARC (Web ARChive) file format offers a convention for concatenating multiple resource records (data objects), each consisting of a set of simple text headers and an arbitrary data block into one long file.