Which is better, rs(0) or rs("column_name")?

It is marginally faster to perform recordset element lookups by ordinal position (e.g. rs(0) or rs(0).value) vs name (e.g. rs("column_name")). 
 
Now, it might be difficult to keep track of name->ordinal pairs when dealing with larger resultsets. There is definitely a trade-off for maintainability... but as long as you control the order of the elements (by avoiding SELECT *, which you shouldn't be using anyway—see Article #2096), you should be able to stay on top of it. To make the transition easier, you should insert a little key for yourself when generating a resultset, such as the following: 
 

<%      set conn = CreateObject("ADODB.Connection")      conn.open "<connection string>"      set rs = conn.execute("EXEC dbo.GetAuthors")      if not rs.eof then          do while not rs.eof                id = rs(0) ' SELECT ID,              fname = rs(1) ' fname,              lname = rs(2) ' lname FROM table                ' ... process results using local vars                rs.movenext          loop      end if      rs.close      set rs = nothing      conn.close      set conn = nothing  %>

 
The only danger here is when somebody changes the order of columns in the SELECT list for the SQL statement or stored procedure. 
 
Now, an even better way is to use GetRows() or GetString() (see Article #2467). But speaking from experience, I know it is tough to teach an old dog new tricks, and you can't just suddenly convert all your code to use those methods. I believe that GetRows() / GetString() are much faster than either option above, but I find the development cycles are much longer because I haven't yet bothered to memorize the syntax.