Thinking ahead, services could suffer Denial of Resources/Services attacks via:
1. Repeated join, post, 'request deletion' from individual or multiple data subjects acting in concert.
2. Individual data subjects may join, post, 'request deletion' to multiple services simultaneously.
Because the 'deletion' data is PII that should make it possible to identify these activities either by single services or services sharing intelligence.
That brings up an interesting issue. In order to operate effective anti-spam measures to deal with this, data controllers/processors would need to retain some PII. That retention would likely come under the heading of 'needed to operate the service'.
Hashing the identity items individually (email address, nick-name, real-name, telephone, etc.) would mean the anti-spam data was anonymous. The only time it could be linked to an identity would be when the data subject tries to join a service or request deletion using the same data.
Which brings up the other difficult issue with casual pseudonymous services: verifying the requester is the data subject.