This is my reasoning. Unsaturation increases the electron withdrawing properties. Therefore, the pKa decreases from ethane (~50), ethylene (~44), to acetylene (~26). I argue similarly if an unsaturated atom is a neighboring group. The pKas of propylene, acetone, or the enol of acetone are also decreased. The enol of acetone would be more acidic than the ketone as the proton is attached to a more electron withdrawing element (more protons).
I prefer this argument to an argument about the stability of the resultant anion, because I prefer to argue why a proton should be released from a compound over an argument about the stability of a resultant anion. If you apply this reasoning to cyclohexane and benzene, benzene should still be more electron withdrawing.
Where this becomes confusing is how this affects different reactions. If a benzene ring is involved in a reaction in which the electrons of benzene are donated, then substitutents that can contribute to electron donation will increase the reaction rate proportional to their donation ability (and compared to benzene). Therefore, a methyl group (toluene), an amino group (aniline), and an hydroxyl group (phenol), all have higher reaction rates even though a nitrogen and oxygen are more electron withdrawing than carbon (or hydrogen). If these same compounds are deprotonated, then the pKas are in the expected order as these reactions require a different reaction mechanism and different electron demand.