The original measured frequency of the cesium transition energy was 9,192,631,830 ± 10 Hz (cps), based upon the duration of the second maintained at Greenwich. When it was decided to use the duration of the second based upon Ephemeris Time, the technicians had to adjust the frequency counter time base (which had been based upon the Greenwich second) to that of the Ephemeris second. The Ephemeris second is slightly shorter than the Greenwich second, thus the frequency counter would count fewer cycles, and this came out to 9,192,631,770 ± 20 Hz. The energy level of the cesium transition did not change, just the way it was counted. See page 6 of the following report:
http://www.cl.cam.ac.uk/~mgk25/time/...leapsecond.pdf
All this means is that physical scientists in different locations can compare frequency values with others, it is a "convenient" reference. The report implies that technology will improve upon the precision and at some time a decision eventually will be made to correct the "error", and the world will then have a new value for the duration of the second, as well as all the other values that are dependent upon that definition.