I am currently reading John Craddock and Salley Storey’s book Active Directory Forestry. It is a very good read, and I can recommend it highly.
Anyway, I discovered something interesting. The book mentions some attributes of the RID Set object of Domain Controllers, specifically the rIDNextRID attribute of that object. rIDNextRID holds the value of the RID that will be given to the next object created on that domain controller. On my DC that value was 1130. Just for fun I created a new user and checked the value of rIDNextRID, only to discover that it had jumped two values insted of one, to 1132. This was a lone DC and no other objects had been created at the same time, so that was not the reason. I checked the SID of the newly created user and discovered that it had a RID of 1131. So which object had the 1130 RID?
I searched the forest for that SID and the search returned 0 objects. But then I asked LDP to return deleted objects as well, and then I got a match. The object with RID 1130 had been deleted at nearly exactly the same time that my test user had been created. The deleted object also had been named exactly the same as my test user. What had happened?
The answer was that when I created my test user I specified a password that did not meet the password requirements of my domain. You complete the create user wizard completely before you get that error from the directory service. So what happened was that the user had been created in the directory, but had been immediatley deleted by the system when it was found that the password did not meet the policy.
In the user interface I could simply click the back button and reenter a new password. That however triggered the creation of a new user, identical to the first one, but now with a compliant password.
I found this to be an interesting quirk and a nice insight into how the administative tools work.