Why did I get so many duplicate items with data load?

I'm moving an org to a new clean NPSP installation and that involved loading about 22,000 Account records and about that many Contacts using the apex data loader. Before doing that I set up the dupe management system using the standard match rules and a clone of the NPSP match rule (matches fuzzy first name, exact last name and email on Contacts). Note I cloned it because it shows as activating but never actually activates!

When I was through I ran a duplicate Accounts report and it shows more than 47,000 duplicates. That's not possible, first in terms of our data and second in terms of sheer numbers. Even if I allow that each Account were reported twice as a dupe, that might reach as many as 45,000 dupes. But because there is only one matching rule catching the Account dupes (and one Account duplicate rule running) that would only happen if every Account record were a dupe, and I assure you that is not the case.

When I review some of the "dupes" there appears to be nothing in common among most of them. Names are completely different, addresses are mostly or completely different. At a glance I'd say there's no way most of them should have been assessed as dupes.

So what's going on? Why so many dupes? And why so many duplicate record sets with records that are clearly not dupes? More importantly, now what? I'd hoped the system would help me deal with the dupes I know we have, but now that I've loaded them, it seems I'm out of luck. Please tell me there's something else...

Thanks for any input!