@Greg in this particular case the issue isn't really about a manufacturing error rate. From personal experience I think Ruger's QC, at least in their semi-auto operatiosn, is far too poor for an MIA product. We're talking quality being compared (and not particularly favorably, sadly) with Taurus. Even so, you have valid points about us not having good numbers for a "big picture" view...but that applies only to initial quality.
This isn't unique to Ruger, though from my POV it appears that Ruger is leading the trend to manufacturing mediocrity for the American firearms industry. The bottom line, unfortunately, seems to be that American manufacturing pretty much across the board is not what it should be and could be. Probably plenty of blame to go around for that, from labor to management to, yes, we consumers and our demand for inexpensive products.
But, this thread is about a gun that went back for service for a specific problem, was supposedly repaired, but came back to the customer with exactly the same problem. That makes it a whole 'nuther horse. And it's not even an esoteric problem that might be difficult to duplicate. To be fair, this too is not a problem unique to Ruger. I've heard of one case where an early S&W .40 Shield went back for repair and was returned as repaired, only to fail again on the first trip to the range, whereupon the owner noticed a hairline crack in the poly frame. Under recoil the frame was flexing and binding the slide.
But, when a product comes in with a complaint, and one that is pretty easy to duplicate, it is inexcusable for that product to go back to the customer without being repaired and then tested extensively. We don't know what happened at the plant in this case...whether it was a case of the paperwork for a different gun being mixed up with this one so the gun went back to the owner untouched...or it was a case of a sloppy tech (or a sloppy process) that meant that a standard set of fixes got applied to the gun but then it wasn't tested beyond firing a couple of test shots...etc. The bottom line, though, is that the process failed and that seems to be happening enough that the process needs to be addressed.