Using CIDs in the knowledge graph leads to 3 problems:
Content duplication and rank inflation as result. While IPFS itself strives to deduplicate content, CIDs creates a situation in which 1 unique piece of content can identify by infinite variants. This potentially could kill the idea of ranking.
Inefficient memory usage. CIDS are 64 bytes. We can do 32 bytes. While 64 bytes max length is enforced the average length is around 40 bytes. In Cyber GPU memory is the most scarce resource. Switching to Keccak-256 (also used in Ethereum) we can reduce requirements twofold.
Potential censorship for encodings. Currently, all additions to CIDs are prone to censorship. We can move out the information on how to interpret given CID to the knowledge graph itself hence making the approach ultimately flexible and resistant to censorship.
That is. I cannot see now how we can move the Great Web further with CIDs. They are clunky, authoritative, and inefficient for use in the knowledge graphs. I would switch content identification to Particles IDs (pids) - simple KECCAK-256 which is already being used in Ethereum. This decision simplifies so much and make cyberlinks even more powerful and universal construction.
Using CIDs in the knowledge graph leads to 3 problems:
That is. I cannot see now how we can move the Great Web further with CIDs. They are clunky, authoritative, and inefficient for use in the knowledge graphs. I would switch content identification to Particles IDs (pids) - simple KECCAK-256 which is already being used in Ethereum. This decision simplifies so much and make cyberlinks even more powerful and universal construction.