See also my earlier pieces: "The floresiensis mess" and "Homo floresiensis update".
Yet more about whether the Flores specimens discovered in 2004 constitute a new species, this time on the positive side. In an upcoming article in the Proceedings of the National Academy of Sciences, paleoanthropologists Adam Gordon, Lisa Nevell, and Bernard Wood of George Washington University argue a statistical analysis of one of the skulls (LB1) reveals the Flores specimens cannot be shrunken Homo sapiens. Intriguingly, they say the skull most closely resembles Homo habilis, a very primitive hominid indeed. Says Gordon: "This is particularly exciting because ... it suggests that we really do have a hominin lineage that split off from our own as much as 1.7 million years ago, yet persisted up until the time when modern humans started peopling the Americas. That's pretty cool."
When the study goes online, I'll update this entry with a link to it.
Update: the article can be found here.