As Neuropost points out non specific LBP remains full of the unknown when it comes to etiology and pathophysiology. We are probably dealing with a very heterogeneous problem (ie looks superficiallylike one problem but is actually multiple problems with different causes), core stability training does have a good evidence base but it is unlikely to work all the time. Other approaches like McKenzie and alexander also have an evidence base, and no doubt have their successes and failures. As clinicians we need to get much better at correct diagnosis. Until our diagnostic accuracy improves we will continue to offer the wrong technique for the wrong patient.
Personally I think we now have enough treatment options and instead of concentrating on developing better diagnosis we keep dreaming up "new" treatment approaches when in fact what we may end up doing is rehashing old treatment approaches that went out of fashion and became forgotten. I think we need to become more discerning about when to apply or not apply an approach and not get so wrapped up in advocating for one treatment approach over another.