Peppy mentioned today in the #osu chat that a Hidden SS score is better than a 96% Hidden+DoubleTime score in terms of Performance Points by about 50%. I'd just like to point out that if this is truly the case and it remains, this ranking system will be inaccurate for a number of reasons.
The main reasons are that the actual score and individual map ranking offered by Hidden+DoubleTime scores are quite a bit higher and it is typically much more difficult to achieve a 96% Hidden+DoubleTime score than a 100% score with Hidden in most beatmaps (e.g. Cookiezi's score on
http://osu.ppy.sh/b/95360 as opposed to ragelewa's). Because of these two reasons, skilled players will typically strive to set low-accuracy Hidden+DoubleTime scores as opposed to high-accuracy Hidden ones, even if a number outside of the actual score tells them that they are doing worse.
Another flaw in this system of ranking is that because Hidden SS scores are worth less points than a Hidden+DoubleTime score with low accuracy, it is impossible to "fix" Performance Points by replacing a former Hidden+DoubleTime score with a normal Hidden score. This means that even if a player wants to increase his or her Performance Points after setting a higher score, he or she will be unable to. This will make moderately skilled players who can hit accurately lower ranked in terms of Performance Points than high-level players like Cookiezi and White Wolf who set ridiculous 95% Hidden+DoubleTime scores, at least for some individual maps.
I won't be able to take this ranking system seriously if accuracy is so important to it.