I have a new column up arguing that it's much more likely that the BLS undercounted job growth at new small firms over the summer than that they overcounted September employment. Basically if you extend the "rebenchmarking" analysis of job growth through March forward, then it all adds up.
I might add that this has become an excessively partisan question. The basically shape of the river is that 2008 was horrible, 2009 was even worse, and since mid-2010 we've had an about-average pace of job growth rather than a rapid catch-up pace of job growth. There's no read of the September data that changes that analysis. The whole question of whether things have been "good" or "bad" under Obama comes down to whether you blame him for 2009 not what exactly happened in the summer/fall of 2012.