BaerWald vs VPI setup protractors


Which is more accurate? Recently I decided to check my setup on a Scout using both the VPI gauge and a Baerwald protractor. Using the Baerwald the overhang is dead on in both locations, using the VPI the stylus misses the mark forward by about half a millimeter. Can this small amount of variance have a sonic impact? Has anyone else found this differene and what was your solution? 
128x128gillatgh
lewm, You may enjoy using an expensive over-engineered device, but if you think that with your $800 machine you get a more precise alignment than I with my Cart-a-lign, you're deluding yourself.  Eventually you, like I, have to set the cantilever in a mirror.  With 2 points of tangency the setting is verified.

The point-to-the arm devices are made as precise as necessary by extending the pointing line.  Lately btw I've been using my carpenter's level to throw a laser line to set the direction to the arm--using the edge of the laser beam, it could hardly be more precise.  But I was doing fine before this.
melm, I give up.  Re-read my last 3 posts. Or don't. 
Invictus, Lofgren A = Baerwald.  So, what is your opinion now?
I would never trust a printed on paper protractor on a high-end turntable. I’m an architect, and deal with skilled accurate drawings every day.  Even with good quality printers on good heavy gauge bond paper things can be ever so slightly off. First paper shrinks and expands based on humility. Second printers don’t necessarily print perfectly, depends on how they print and to how acutely the paper mechanism rolls the paper as print, thirdly and most importantly the hole needs to be man made and when fractions of a mm count that’s too much risk. Don’t get me wrong if all you want is a generally OK alignment then maybe paper is OK but I would never trust that on the high-end turntable. Factions of a mm matter. 
@lewm My opinion is that Baerwald is a fraud. Lofgren A is the correct alignment, so why do people insist on calling it Baerwald? 
Thanks all, I appreciate everyone's input and gather from it that there are many ways to obtain desired results. There seem to be equal numbers of respondents in all camps. No one is wrong using their preferred protractor although I'm  intrigued with the mirror protractor.  May have to try that. I'm perfectly content with my set up and it sounds fabulous. Unfortunately no one answered my question.