string method is generally not reliable. Park FAG2 is just an overpriced string. None of the frame alignment "standards" people quoted is really correct, because there are two standards. Any good framebuilder will have two alignment tolerances, one for how far off of the centerline line a part of the frame can be (but parallel to the centerline), and one for how twised it can be relative to the centerline. The twist is the tigher tolerance. The important thing is that when you look behind the frame - are the wheels parallel (and perpendicular to the seat mast / BB) or do they make an "X"? The frame can be out of alignment quite a bit but as long as everything is parallel to the centerline it will handle fine. When the front and rear wheels make and "X", then you start have problems. The issue with the string method is that it measures down the center of the "X", which can read perfect but your wheels may possibly point in completely different directions. The true test is, take your hands off the bars - does the bike pull in one direction? Now turn hard left and right, does is want to dive into the turn going one way, but fights you going the other way? Then you have an alignment issue that can only be corrected by measuring and setting on a precision table. If it handles fine, then forget about it and don't mess with your wheel dish either.