Well, the most likely cases is:
1; someone grabbed a full 14-day-cache while the system was up, and has of course not managed crunching all yet.
2; someone is only crunching a couple of hours/day, and since the "active-frac" on install is always 1 he's got assigned too many wu to manage crunching these within the deadline.
3; someone for various reasons just dumped their cache.
4; due to server-problems verifying results is either disabled or goes currently very slowly...
The important thing to remember is that if you yourself crunches correctly you should get credit after some days, but due to the anti-cheating-measures & more fair crediting added to BOINC it can take a couple weeks to get this credit.
Then all the initial server-problems & database-optimizations is fixed, expect a more constant credit-flow, but you'll still lag some days. Except in team-races this really isn't a problem.
edit... "if it ain't broke, don't fix it..."
Well, the seti@home-classic-crediting is really broke, remember the doublet-cheaters & the 99%-"give-to-friends"-cheaters. Also, even with switching from v3.03 to v3.08 it's already a 10% difference, not to mention v3.00. Even the normal time-difference due to AngleRange is fairly huge, add on top some "turbowu" taking under a minute to crunch. Remember, before v3.03 was released Anandtech was being beaten by MURC and just look there they're today...
So anything that removes the cheaters & gives more differential crediting than "1 wu = credit" really is an improvement.
Of course the benchmarking can be better, but even with 10% variance here it's better than the old seti@home. AFAIK atleast on the same machine the benchmark-variance is under 1%, so the only huge difference is between machines.
Oh, lastly, with only delivering the same wu to 3 crunchers BOINC is over 100% more effective than "classic"...
(edit... fixing tyipgn errro)