And while I'm on the topic of decreases again, here's a quick little nugget that I think is really clever, and thought I'd throw out:
I have a commercially-knit cardigan that has a section of 2x2 rib that merges into a 5x2 based pattern, so that they needed to get rid of one of the stitches in a "k2, p2, k2" sequence. Now, one of the design issues with decreases is that they do slant visibly -- you can minimize the impact by not lining them up, but you can't really get rid of it. However, what the designers of this cardigan did was to use "k1, SSK, YO, k2tog, k1" across these 6 stitches, and by using two mirrored decreases on either side of a neutral increase, they produced a net decrease of 1 stitch with no resulting slant. Of course, this does produce a hole, which is rather more visible than an increase, but is also rather more easily incorporated as a design element, and indeed that's what was done here, as there's a lace pattern in the upper part of the cardigan. One could use an M1 to avoid the hole; there's a slight slant due to the twist in this increase, but it's less prominent than the slant in a decrease, and the result would still be more symmetrical than a single decrease. I find this to be really interesting as a lesson not to think of my increases and decreases as single units, but in terms of the net increase or decrease to the row overall, and I'm busily thinking up intriguing ways to incorporate it into some of my designs.