If you think of the payroll survey as measuring the employment level rather than the change in the employment level, the revisions are actually tiny.
33