kd,
Actually, the more I read my post, the worse it appears to convey my thoughts.
As I am sure Teddy and Q would agree, getting to an end point is as much "gut feel" as anything, based on multiple factors. It then becomes difficult to justify it easily in any written form. It's a bit frustrating really, which is why I rarely post on the subject.
As for USCIS processing speed, I would disagree. In February, they have managed to approve far more EB2-I cases than any QSP could provide. Even being conservative, EB2-I alone received 4.5k approvals - then add EB2-C on to that. That level of approvals is not sustainable. In fact, it is not inconceivable DOS/USCIS may have blown the February allotment and used some of the March numbers. The law only talks about quarterly numbers, so that is not an issue. It may be that the numbers are "ahead of the curve" and some throttling back is required.
The dates must retrogress at some point to a date that can handle the number of visas available for FY2012. I agree that when that becomes necessary is a point for debate and differing opinions.
I don't think USCIS have the capability to throttle approvals with the level of finesse required. If the dates don't retrogress, then USCIS might just pick the low hanging fruit from ever later cases to meet the numbers.
In that sense, retrogression is a good thing, because it will force USCIS to adjudicate cases they feel are complicated, or don't generate a quick adjudication towards the ISOs targets. The sooner the better, once it is deemed necessary, so that the maximum time is available to adjudicate those cases. It will act as a focus for USCIS. If it doesn't happen, then people with earlier PDs are going to be left waiting, while later PDs are approved.
Even if the dates do retrogress, all approvals are still being made under the spillover laws, so I didn't quite understand your comment. It merely limits the timespan of PDs that USCIS has to choose from.