...
- Supports billions of identifiers with very fast resolution
- Robust architecture and implementation - a highly available system
- Need to avoid unbounded generation of surrogate persistent identifiers
- Should support multiple delivery formats for an identifier
- Must support splitting collections (what does this mean?)
- Need a lightweight understanding of identifier equivalence
- Need a way to integrate outside PIDs with Cornell (what does this mean? examples?)
- The identifiers and the associated content should be easily discoverable by Google
- The overall system should integrate well with the "web architecture"
- Should have a PID corresponding to every Cornell NetID and potentially other non-digital resources, not necessarily at Cornell?