RAVE implementation different from paper

[dsmic]

refering to

http://users.soe.ucsc.edu/~dph/mypubs/AMAFpaperWithRef.pdf

If I understand explanation and Fig.1 correctly, the RAVE values are never updated for nodes of the actual playout?! If I understand our implementation correctly we do update all children of the nodes in the actual playout, which includes the node played in the actual playout.

This has an effect on the RAVE formula of cause...