As a matter of policy, the DNS technical specifications do not mandate a particular tree structure or rules for selecting labels; its goal is to be as general as possible, so that it can be used to build arbitrary applications. In particular, the system was designed so that the name space did not have to be organized along the lines of network boundaries, name servers, etc. The rationale for this is not that the name space should have no implied semantics, but rather that the choice of implied semantics should be left open to be used for the problem at hand, and that different parts of the tree can have different implied semantics. For example, the IN-ADDR.ARPA domain is organized and distributed by network and host address because its role is to translate from network or host numbers to names; NetBIOS domains [RFC-1001, RFC- 1002] are flat because that is appropriate for that application.
However, there are some guidelines that apply to the "normal" parts of the name space used for hosts, mailboxes, etc., that will make the name space more uniform, provide for growth, and minimize problems as software is converted from the older host table. The political decisions about the top levels of the tree originated in RFC-920. Current policy for the top levels is discussed in [RFC-1032]. MILNET conversion issues are covered in [RFC-1031].
Lower domains which will eventually be broken into multiple zones should provide branching at the top of the domain so that the eventual decomposition can be done without renaming. Node labels which use special characters, leading digits, etc., are likely to break older software which depends on more restrictive choices.