We've had terrible problems with bugs caused by developers using select * with bind_cols. Frequently the problem can not be reproduced on our development system. Table changes affect the order and/or number of columns, so the hard-coded variable list in bind_cols is a bug waiting to happen.
Selecting explicit columns fixes the problem, but then the column list is redundant with the bound variable list -- bad (short, weird abbreviations, etc.) variable names can be used because it's "obvious" what the variable holds.
I like fetchrow_hashref because it's safe to use with select *, self-documents use of database columns in the code, and does not have annoying redundancy between variable names and column names. Hakkr was probably thinking of these advantages.