As I mentioned to Dinesh, the IP address is not a good solution, it's just better than no solution. But you are right that a hash of the cookie may be a better solution yet.

As to having to upload the file in order to import it, is this just a temporary implementation detail, or a fundamental design decision?

I don't think it's the way most users would expect it to work, and it certainly isn't the easiest way. Importing a SCORM file should be a simple matter of sending it to the seerver for import in a single operation triggered by a single form. Otherwise, the process involves too many steps:

1) upload the file
2) import the uploaded file
3) delete the uploaded file

This is too complicated, uncalled for, and brings with it the problem that most users will likely forget to perform step 3, which will leave us with a server full of large SCORM objects we cannot delete.

I guess I should open another ticket for this one?

Would the wikipedia way of doing things (attributing the change to an IP) be too much trouble? I realize we don' t want to have a Plone user for each IP out there... I'm thinking along the lines of creating a single user called "anonymous", and to log the IP address together with the change.

CAPTCHAs for anonymous are a Good Idea, too.

On anonymous uploading: maybe a bad idea. If the file manager is related to a user anyway, it doesn't even make much sense to allow anonymous to upload stuff to an either non-existing or simply humongous pile of stuff. I guess anonymous SCORM import should be allowed.

The system currently forbids anonymous users from editing LOs, but it shouldn't.