For both fun and work I am very interested in getting a community around map/reduce technologies (hadoop/mahout/mongodb) and building out n sized clusters using Chef. There are already two of us here trying to work on this goal with modest/lower power systems (pi, i.MX5, atom) - but this is way more shiny/powerful/dense . I think that the Parallella presents us with an opportunity to extend the affordable hardware efforts represented by this little board toward massive computing and it's potential for disruptive/democratizing impact when that potential for data analysis is made accessible to all.
clark
too bad I wasn't able to back it up on Kickstarter on time.