How about putting a list of batch size numbers into the documentation? I think this might help people to avoid certain problems.
For instance I use a dual CPU system with 4GB memory and I run into troubles when I try to import library data when I do batch import of more than 20'000 tracks at a time. Then I once did the 'analyze track' function for more than 2'000 tracks in one batch and the program died on me regularly. Analyzing below 2'000 tracks just works like a charm...
That list could help people to spend a lot of time to investigate problems which are simply no problem

just a suggestion...
Martin