I think we're in danger of running away with this notion that SR puts motion back in when there is none.
Personally I don't think the evidence for it is at all compelling.
There have been a few reports of on-tripod SR causing blur. However, these are anecdotal, and not usually backed up with solid evidence.
I've investigated this fairly carefully, and found no actual evidence of additional blur caused by SR, although I have found a temporary camera malfunction that is reason not to carry out manually-released, tripod-mounted, very long exposures with SR on.
What people sometimes don't realise is that the K10D automatically turns SR off in almost all the cases where you'd be using the camera on a tripod - so all the mirror-up delay modes, and all the remote-release modes have SR off, regardless where the body switch is set.
So, if anyone has done a carefully conducted experiment and found good evidence for SR-induced blur, using the camera on a tripod and releasing the shutter, undelayed, using their finger on the button, I'd be very interested to hear details of it.