Hi Code4Lib, We are in the process of designing new workflows for preservation and access of our digital stuff, and I'd like to get a sense of how people understand digital objects in the preservation space. My gut tells me that it might be useful for future digital archivists/archaeologists to have an object's descriptive metadata closely associated with the object's files in the same directory in a human readable plain text format. So that one directory would contain all of the object's files and descriptive metadata in an easy to read package. Alternatively, descriptive metadata for many objects could be stored in a single external file, say at the root of a preservation accession directory, according to a recognized standard like METS. That requires more work to reconstruct an object, and the linkage between an object's files and descriptive metadata is looser, but it seems more efficient. How do others approach this problem? Are there recognized best practices to adhere to? Thanks, Andy Weidner