General rules, even ones that are flawed, facilitate behaving rightly, because they allow a low-resolution look at morality with very minimal effort.
Let's pretend moral propositions are dots on a 2-D graph, and rules are lines (infinitely long). To keep it simple, rules only run horizontally or vertically. And the way to analyse the morality of a proposition (a point) is to figure out the closest rules in each direction (above, below, left, right) and which side of each of the rules the point falls on.
So if we wanted to analyse the moral proposition (3,6) we'd just go in each of the four cardinal directions from (3,6) and figure out which 4 moral rules we ran into, and which side of them we're on.
But moral rules aren't perfect! Some are even highly inaccurate. So how can these rules be a good idea? Well the point is if we're analysing (3,6) and we're checking for vertical lines and find them at X=1 and X=66, we know even if the X=66 line was so flawed as to be accurate within plus or minus 40, we'd still be left of it. On the other hand, we can see we're very close to the X=1 line, so even if it's highly accurate, we still need to use a more accurate technique to check the morality of that issue.
Many propositions fall significantly distant from all lines (moral rules), and thus can be analysed purely from a quick, low-resolution look that's quite accurate even with faulty rules.