I wouldn't necessarily say risk difference has a superior meaning, but it certainly has a different meaning.  Here's an example I use in a lecture.

Two scenarios that yield the same RR:

• RR1 = (2 per million / 1 per million) = 2.0
• RR2 = (200 per 1,000 / 100 per 1,000) = 2.0

RRs are the same, but the attributable risks are very different in these two cases:

• AR1 = 1 per million = 0.000001
• AR2 = 100 per 1,000 = 0.1

Note that attributable risk (AR) is another name for risk difference. 

HTH.

Thanks for your helpful answer