@quid (and all) I'm going to talk to the guy that actually wrote the algorithm before I edit / comment further, I want to be sure that what I think is correct is actually correct before offering more focused advice on a premise that might be erroneous. Probably not going to be able to chat with him until early next week so please, stay tuned. I still assert that the benefits outweigh the headaches and will continue to do so, but I want just a little more confidence in my understanding of how it works before I respond to these (very well thought out) comments.

Another interesting thing, unregistered accounts can recover their cookie automatically by visiting this page if they remember the email address they used - if you ever need to guide someone on how to do that.

This is pretty much it. With only a few exceptions, you can ask or answer a question on our sites without registering at all, this was a corner stone design decision because other Q&A sites at the time not only wanted you to sign up, they wanted you to pay to get the answer you thought you found in your search results. We do encourage folks to register within the UI, but we're not too forceful about it.

Just wanted to let you know that we saw this, there are similar proposals on MSO - each site seems to have different needs and wants when it comes to thresholds, and editing guidelines in general. However, this is something that we really want to be consistent everywhere. I'll follow up here when we've gotten through the discussions.

It was an automatically rejected migration, 4/5 close votes were sending it to Math SE, where there was a problem with tags not existing.

Okay. The "sending" site (in this case, MO) counts actual rejections, while the destination counts all questions closed or deleted - even duplicates, thus it's typically higher on the destination. The metric on the sending site is what's important, which is 11%. more here

I'm looking into the discrepancy, I can't readily explain why it differs between sites.

@JohnBentin They also quickly move a question off the front page (there is some caching involved so it might take a minute), but down votes are definitely the tool for the job.