It doesn't look like betweenness centrality normalization is working.
The result seems the same with and without normalization. I expected the solution to be 0,0,1 (after normalization by (n-1)*(n-2)).
> E(g)
Edge sequence:
[1] 0 -> 2
[2] 2 -> 0
[3] 1 -> 2
[4] 2 -> 1
> E(g)$weight
[1] 0.3 0.2 0.1 0.3
> betweenness(g,v=V(g),directed=TRUE)
0 1 2
0 0 2
> betweenness(g,v=V(g),directed=TRUE,normalized=TRUE)
0 1 2
0 0 2