Preface: A second used to be defined by 1/86,400th of a solar day. This wobbles a bit, so a more stable definition was given as 9,192,631,770 periods of the radiation corresponding to the transition between the two hyperfine levels of the ground state of the cesium 133 atom. Fine and dandy. This is presumably because, unlike the earth's rotation, which can wobble and is slowly slowing, cesium always behaves the same, no matter what... it is what it is and it doesn't change.
So, my question is mostly a philosophical one. I understand the supporting evidence for the math showing that these atoms slowed down, but why is it preferable to believe that these atoms are incapable of being inaccurate and that time is malleable? Obviously, because the definition of a second is based on the clock, we can logically say that a "second" has slowed down... but with the old definition of a second, if the earth's revolution were to slow, we wouldn't assume time to be slowing, but the earth's rotation (as evidence by the fact that we realized that the earth's rotation altered rather than assuming that time was slowly slowing down).
Realistically, it makes little difference, the math stays the same even if you say "time stays the same, but everything (including atomic actions) slow down by a certain factor." ... but I wonder why the definition einstein chose is so preferred.