The classic "rule of 6" has a few problems which I have resolved, at least to my satisfaction. As you probably know, there is no real difference between TN = 6 and TN = 7. I started with changing the "rule of 6" to adding only 5 instead of 6 to all re-rolls. That way, if you roll a 6, the second roll will result in a final roll in the range of 6-11 instead of 7-12. This applies to all re-rolls. Another roll of 6 will add 10, so the 3rd roll will be 11-16, etcetera.
This worked for a while, but considering the infrequency of rolling a 6, I further modified it to smooth the curve a little bit. I mean, if you graph it out, the probability curve "stair-steps" something wicked. I fixed this by re-rolling on 5 and 6, and adding 4.
Sure, there may be a little more wrist action, but the odds work much better as far as I'm concerned. The curve is smoother, but in the same general shape. I don't have the exact numbers at the moment, but if I have some free time, I'll crunch 'em and post 'em here for illustration.
Of course, you could just switch to SR4.... when it hits the shelves
Opinions are somewhat mixed as to how good it is, though most people are sticking with the new stuff because obviously no more SR3 books are being published, and the number of good changes in places like Hacking and Rigging somewhat mollify the fact that the total overhaul of the system introduced many inherent problems and limitations. 