Idea is to identify pieces of data by the hash-id of the data.
This can be implemented very easily on a key-value store. To store data
hash(x) and store in key-value store using
hash(x) as key.
A bit more work is required to do this securely.
Content on the Web is usually location-addressed. That is the identifier of the data is a location of a host serving the content. This causes problems with content Availability.
The web of names, hashes and UUIDs
An article by Joe Armstrong describing how content-addressed storage is part of his plan of "reversing entropy".
Principles of Content Addressing
A content-addressable storage. Does not do any securing of content.