Hash function to have sequenced numbers look like "random".

Hello experts,

In short:
How do we take sequenced unique id's to appear as random?
We now have _unique_ id's ABCD1234, ABCD1235, ABCD1236 etc, which should appear more random, without losing the unique-ness. We are thinking of implementing some sort of hash functionality.

More lengthy:
For our custom documentum object type we have a request to provide each custom object with an unique 8-character (a-z0-9) object name. Our current implementation takes 8 unique characters from the object_id generated by documentum, which guarantees uniqueness of the object_name within the docbase. (Which is ok with our client.)

The problem however is that the r_object_id's are more or less sequential, and the client requested to have it look more random.
One way we looked into, is to have a unique random object generated in the doPostCreate by obtaining a lock on a specific object, generate a unique id and test it against the objects in the docbase, then release the lock. We think that this doesn't perform well, as over 50 objects will be created by a single workflow method at each call. That would require a lot of checkin's/checkout's and dql queries.
Our alternative is to find a hash function which takes the current 8 character substring of the r_object_id, and have the sequential id's appear to be random.
In our current situation when we run the workflow method generating not that many objects, we get object names like:
800F43B5
800F43B6
800F43B8
800F43C1

What hash function can we use?

-r-