I may be misconstruing what you're saying (please correct me if I am), but I don't see how MIT's and JSTOR's dismissing of the case makes them any less "primary witness material", should the authorities decide to continue prosecuting.
None of those facts require MIT or JSTOR to testify. A charge related to a denial-of-service, would require the cooperation of a victim...I assume...because if the "victim" isn't going to say that their network was terribly inconvenienced, then it's a hard case to prosecute.
The feds may be pursuing this hard because even if JSTOR doesn't care, there may be many people (i.e. the authors of the journal articles) who feel ripped off.