As a theorist, I do not plan to use the BOINC platform The reason is that me, our any collaboration of theorists, don't need as much computing power as what our experimental colleagues do.

In general, we can achieve doing anything we want with small, very affordable, clusters. If you take the simulation of a single LHC collision by the big experiments, it amounts in minutes of CPU time.... For one single collision. And we need billions of these simulations. As you can see, here there is a need :)