The way account requests work on ProleWiki is that only a few select people are able to see the requests itself. Mainly because it contains sensitive details about the machine requesting an account which we prefer to preserve to protect the privacy of incoming requests. However, those sensitive details can be useful to verify sockpuppet accounts, prevent accounts with illegitimate or harmful intent, etc.
Recently, we received a request which some of us could tell it was written by a particular user which has attempted to infiltrate ProleWiki for months. We could also see the user was trying its best to conceal their identity. But we proceeded with the request as usual, voting on it and collectively discussing requests. The administrators pointed out the similarities with Wisconcom. At first we decided we were going to allow them in, patrol their edits, before they attempted to wreck the wiki. But some editors took issue with their edits, and their behavior further increased our suspicions it was Wisconcom, and thus a new voting session was held and there and then the majority of editors chose to ban Commissar/Wisconcom again.
I was against this decision, because I'm more pragmatic when it comes to improving and developing the encyclopedia. If someone is interested to work, why not? Provided they do good edits, there's no problem in letting them in. In the case of Wisconcom, some editors took issue with his edits. In my opinion, they weren't too bad, most of them were irrelevant, some were fine, and some were mildly bad. Still, it was only a short experiment, and I think no true lesson can be learned from this.
The major problem I see is the matter of expectation. How can we expect good faith edits from an user which has actively found ways to infiltrate, sabotage and disrupt our work? It's okay to make mistakes, everyone does, but how can we tell it's a mistake, and not an act of disruption? When you tarnished your image with your actions, it's hard to tell.