I would think technical users would prefer some help to make automation of tasks easier, since however good the UI is manually adding data to all the fields MusicBrainz supports is always going to be hard work. At least for new releases we have a way to seed a release, the trouble is that the database doesn’t treat the release and the tracks on the release as an atomic unit and so if the database is overloaded it is very easy to end up adding a release but without any tracks (MBS-7296), and it doesn’t seem to take much at all for adding new releases to get overloaded, I wonder if the code can actually only add one release at a time ?
Another site in a completely different domain but with similar issues is http://wikitree.org - they want more people to add more data but they are wary of users adding bad data so they have a similar fear of automation. They have recently introduced a new gedcom importer this allow a user to export his ancestry tree and import it into Wikitree without typing, BUT for each person they have to check against possible matches to the same person already in wikitree and only then can they add in new people, family by family.
The advantage of this process is if the original data is good then little manual editing is not required , but all the data has to be checked before it can be imported, and the system can cope with one user adding alot of new people. I think the technological ability of wikitree (developers and users) is much simpler than those of MusicBrainz, but they have managed to create a system where multiple people can be added fairly painlessly, it would be great if MusicBrainz could also achieve this for releases.