I would definitely prefer to have a system in which any leakage of private uses could be controled under a well-known policy requiring a reservation in a publicly accessible registry, like domain names.
Well, you can't. Private Use is Private Use. You cannot restrict it. You cannot control it. You can, as a private person, guide it, as John and I do for some scripts in the CSUR. But that isn't standard, and it isn't going to be. Ever. Guaranteed.
If one designs an open reservation system in a global registry
Like the CSUR? No, that's closed, because John and I decide what we let in and what we don't.
(possibly with small annual fees to maintain this registration in a global registry),
Which no one would ever pay.....
the reservation could be made much more safe.
Not at all.
In addition, this would not prohibit rapid innovation or usage of new characters, and further standardization if needed in the Unicode/ISO10646 space, where these characters, now of public interest, could be assigned more permanently and without renewal= fees, provided that their usage is clearly documented by its author and interested groups of users...
Even *I* (who encourage all of you to contribute generously to the Script Encoding Initiative, which actually *does* manage to get characters encoded) would not dream of trying to charge money in such a loony scheme.
The main reason why those semi-private characters could be standardized later is for conservation of documents which could then be transcoded to the now safe Unicode/ISO10646 space.
There is no such thing as a semi-private character. There are standardized characters (which have particular meanings), and there are private use characters (which are guaranteed to have no meanings at all).
--
Michael Everson * * Everson Typography * * http://www.evertype.com