Basic analysis of open source software's source - for things like "bus factor". Started as a research project, probably going to die as a research project.
That exists, it is just hidden right now. If my intent was to offer this up and a for-the-public-good I'd turn it back on. But like most of the other comments here, at some point it just doesn't make sense for me to foot the bill to run it. It would be one thing if it was a simple front-end, with a simple back-end. But processing this insight is a little bit resource intensive which creates a cost burden. Academically it wasn't a big deal to run this in a true HPC, the costs have already been covered. Commercially is a different calculus.