I have a couple of questions about the LR4 softproofing compared to that of CS5. My last version of LR was 2.7, at which point I found little real advantage for me of using LR, since I do so much work in CS5 plus 3rd party filters. I softproof in CS5 and use adjustment layers to make the softproofed image look as close as I can get it to the original image, with the two of them side-by-side. I use Qimage for printing, which gives a fairly accurate softproofed view of the image, and I compare that with the printed image which I view in a dimmable lightbox (Just Normlicht.) The reason I use Qimage is that it seems that I can easily override the size limitations of my Epson 4900 printer, and there are some other aspects of Qimage that I find more helpful than printing from CS5.
I still have no overriding reason to use LR, but I have tried the LR4 beta. As far as I can tell, the only advantage it has over the softproofing in CS5 has to do with the convenience of setting up the side-by-side pair of images, but I would be more likely to get LR4 if:
1) The accuracy of the softproofing is higher than that of CS5 (some new paradigm in simulating the softproof from the printer profile), and,
2) Would there be a way of exporting the image with the softproof adjustments to Qimage? Or would I still have to open the original image and the softproofed virtual copy in CS5 and somehow make the adjusted softproofed virtual copy into a layer or layer group on top of my original image, which would add steps to the workflow?