Prisoner 6: All this commotion led me to do some measurements. My 92FS slide has about .015" side-to-side slop at the muzzle end, and about .010" up down. It is definitely looser than my CZ or HI-Power. It is way looser than either of my 1911's (my STI is so tight it's not measureable). The point of the article was that there are lots of well established ways (and parts) to accurize a 1911, but doing it to an M9 was really tough. And, I recall, the M9 slides were tightened up using some kind of inserts installed that could be machined down to fit.
As far as the barrel locking system: if you want to see it's accuracy limits, take the recoil spring out and reinstall everything (gun has to be held upside down so the locking blocks will drop in). When "locked" in battery, grab the rear end of the barrel: mine wiggles about .015 laterally (left to right) and about .010 up and down. The front of the barrel will go about .015 -..020 in any direction.
Re-install the spring: the spring force drives the locking block to the rear and wedges the barrel "wings" into the frame grooves. There is a lot of slop in the groove/barrel fit, so this is the first error point. The barrel (chamber end) has a fair range of freedom of placement when the gun cycles. You can see the side to side play in the chamber end of the barrel, as the rearward force of the recoil spring does not restrict movement in those directions.
When in battery, the recoil spring forces the bottom side of the muzzle end of the barrel to rest on the slide. Spring force holds it down. Of course, at firing, the barrel muzzle will rotate upward since the bore axis is above the holding point so there will be a torque. Another source of error, the barrel is moving at discharge.
If this is a "floating barrel" design, I think it's pretty adequate. Reliabilty is high and accuracy is good. It just seems to me that it is (by design) less accurate than a 1911 where the barrel is locked very tightly to the slide both by bushing and barrel lugs (held at both ends).
