IOW when I left the experiment, the execution graph was determined statically up-front, that was a liability as pointed out at summit by ... Seibel? and I thought that was solved by Stus later work