C 4 04 referee report
This paper is suitable for the Journal Concurrency:Practice and Experience.
However some changes are needed before publication. Firstly the authors need to discuss the
grain size of the tasks that they are producing the programming model for. For instance systems
like AVS are designed to address very large grain size (like full programs). Other approaches
like openMP are designed for quite small grain sizes such as loops inside a program. Do you
address everything or one of these limits.  The examples suggest a large grain size.
Further some key references are missing.  I suggest adding a reference to Opus, openMP and also
to literature on low balancing such as the work at San Diego from Fran Berman. Of course you
should describe how your work improves on the previous literature.