I don't feel strongly about the actual numbers; I think if development continues and features are added, the coin will have value in the long term regardless. Arguments for one number or another ultimately depend on speculative assumptions about DNET's future value anyway. But whatever numbers are chosen, I'd prefer the reduction follow a predictable pattern. For example, in the proposed system, the drop from 25 to 5 (1/5) is out of keeping with that between the other phases (1/2). Perhaps strict linearity (50-25-12.5-6.25-3.125...) is no good because it entails awkward fractions and a slower path to 2.5, but even something like 50-25-10-5-2.5 (so, one extra phase but linear-ish) might soften the blow. At the other extreme, something like 50-15-5-2.5 (so, nearly exponential decay) could also work. I just find the 1/2 - 1/5 - 1/2 pattern seems arbitrary, and I think it would create unnecessary drama around the 5x reduction.