Sigma was going backwards for a while with its 70-200 f 2.8 lenses. The ORIGINAL "DG" version (BEFORE they started to add "Macro" "features") was actually better than the "DG Macro" and "DG Macro II" lenses, and apart from the CA numbers, was actually comparable with the Nikon VR lens (no VRII results to compare yet). It equaled the Nikon's resolution (58 lpmm), distortion (0.3%), and was close on transmission (3.3 Tstop vs. 3.4 for the Nikon), while being BETTER on vignetting (-1.3 ev vs. -1.8 ev for the Nikon), falling short ONLY on CA (15um vs. 8 for the Nikon).
The original DG 70-200 BEATS the "OS" version on resolution (58 lpmm vs. 54 for the OS), transmission (3.3 T Stop vs. 3 for the OS), distortion (0.3% vs. 0.4% for the OS), vignetting (-1.3 ev vs. -1.4 for the OS) and falls short ONLY on CA (15 um vs. 8 um for the OS). SO in reality, the OS has "progressed" on ONE performance measure (CA) while REGRESSING on every other, compared with Sigma's original DG version, while simultaneously suffering a massive price hike and a massive downgrade of ergonomics compared with (all of) the older versions. As far as I'm concerned, Sigma would be better off shedding the gimmicks (like "macro" focusing and OS) and concentrating on improvement of the ORIGINAL DG lens design (the best of the lot) to improve on the CA. Then they could sell a lens with comparable performance to Nikon at a lower price than Nikon, which has always been Sigma's strength. As their pricing climbs into the stratosphere with OS gimmicks, they are losing their appeal as a less costly but viable alternative to C & N. At the very least, they should offer non-OS versions of their lenses in addition to the OS versions (with their crappy ergonomics) for those who won't find the (much smaller) price differential a compelling enough reason to choose their lenses over N & C.