Congratulations to all concerned with the publication of An Asynchronous Hybrid Genetic-Simplex Search for Modeling the Milky Way Galaxy using Volunteer Computing.

One remark in particular struck me on reading the paper: “[…] BOINC is much more accessible as it is dedicated to the project at hand, while use of the BlueGene is shared among many researchers.” From a point of view shared by many—probably most—participants, this is actually far from the case: we run a variety of other BOINC-based projects as well. For example MilkyWay@home gets between one-third and about one-sixth of my attached hosts’ CPU time, and its contribution to my overall RAC is about one-tenth.

Yet evidently BOINC as a system succeeds in creating the illusion of dedication from the project’s point of view, despite the partial availability of the participating hosts and the issues of volatility & heterogeneity also mentioned in the paper. IMO this constitutes a remarkable achievement.

---