The CClip ID would be the content address of the CDF. In the CDF there is (are) the content address of the referenced BLOBS.
Which naming scheme is used depends on the settings of storage strategy, the M++ settings and the size of the object.
GM would be the naming scheme for the CClip ID.
The M returns a 128-bit CA which is the MD5 hash of the object. GM returns a 256-bit CA containing the MD5 of the object, a random bit-string, timestamp, counter and some header bits. M++ is also a 256-bit CA containing the MD5 and a truncated SHA-256 bit of the object.
If I remember correctly, GM is used for the CDF's and Blobs smaller than the storage strategy full performance threshold. M and M++ are used for the BLOBS larger than the threshold and depending on the M++ setting.
I don't have all the docs to really figure this out in detail for you but hope this helps as a starting point.
Clips and blobs are named based on the md5 hash of the objects being written.
As Holger stated, the actual name depends on the storage strategy defined however the storage strategy appends additional information to the md5 name such as a GUID etc. The base is always around the md5 hash though. This hash is 26 characters. Depending on the storage strategy etc, additional information can be appended to this to create a 53 (26 character md5 hash + additional such as GUID) character name.