URL structure is pretty straight forward. If you want to access the discussion there is a singular URL called /thread/XXXX which will load the discussion starting at the root message.
Every reply to that thread is given a unique Message ID and corresponding URL.
In essence, there are 2 logical URLs to load the rootMessage in a discussion (/thread/XXX and /message/XXX)
There is no predicting the URL as the ID is basically a sequence and has not bearing on the contributor.
Does that help?