One argument against this is that many usernames / identifiers in other areas do not require numeric components. Most likely, users will be better off remembering the user IDs they used elsewhere, and this is more likely if they should not include numerical values.
In addition, depending on the system, user IDs may work well by default when connected to external systems (ssh behaves this way on unix-like systems). In this case, it is obviously useful to have one identifier that is shared between systems.
, . , - ( , ) . (, , , , , , , , , .)
: , , , , .