0.5.3b for two reasons:
1. Easier to develop axioms for it
2. ++ and --
I fancy ++ and -- too, and I use 0.5.3 too, although I guess not too many people use it because of the 16bit/16bit division bug.
Actually if I remember correctly those bugs were fixed. I don't have an 83+ series calc. at the moment, but when I get my 84+SE(this week hoppefully) I'll be using the newest version.