Following is a quote from a portion of the Class IX podcast. I ran the whole section past @dirvine before doing it because I wanted to ensure I had the whole thing straight before I did and he gave the whole bit the thumbs up.
“Upon implementation, all safecoin’s will exist in the network as specific type_tag’d network addresses. Initially, they are unowned. They contain fields for current owner and previous owner. (If there’s any other data in them, it’s not relevant to this example).”
Perhaps he skimmed and missed this point, but it made sense to me that it would be this way because of the fact that elsewhere it says a random safecoin address is queried to see if it is owned. This would mean that the address had to be at least accounted for, but if not already there, a mechanism for creating it at that moment would have to also be in place. That’s much more complex than just creating them all at the same time, so that they are then placeheld, etc.
Of course, I definitely could have this wrong. It’s also possible that this point hasn’t really been decided in exact detail.
EDIT: I also preambled the whole lifecycle of a safecoin bit with the following disclaimer:
“The following description is not accurate in exact technical details, but is effectively how it will work.”
It doesn’t really make a difference from a higher level of safecoin function, but when it gets down to discussing exact implementation, it does. So I’d have been better off not making the point here, I guess.