relational databases are created the way they are for a reason, read a little ot the history on them if you want to know those reasons. I will make an example of this particular case
Here is the data layout
method 1 (two tables)
user_info
userid
name
username
password
phone_nos
userid
phone_number
designation
method 2 (one table)
user_info
userid
name
username
password
phone_numbers
Let's say I want to get someone's work number.
Under method 1:
I construct a sql statement that gets the person's work number and relates it to their username using the two tables and their userid based on the designation field of the phone_numbers table, then I echo that number
Under method 2:
I construct a sql query that gets all the person's phone numbers. Then I write the code to parse that information into arrays or classes. Then I write the code to search those arrays or classes for a work number. Then I display the person's work number.
I think I'll stick with normalization rules the way they are cause people a lot smarter then me thunk them up.