The problem with looking at the discovery curve is that you have to believe the reserve estimates for these discoveries.  Those numbers are very dicey - especially for fields that are not far along their depletion curve.  The discovery curve only counts recoverable reserves for that portion of the in-place volume that is considered recoverable under the production technique being used.  How do these estimates account for reserve changes later in life due to application of secondary and tertiary recovery methods?  I doubt that they account for changes in understanding of the geology and the amount of in-place volume also.
Bubba, there is great uncertainty as you say. There is some ultimate URR from these new discoveries but early in the lifetime of the field in question--as in any Hubbert Linearization-- this is not known until things "settle down". This takes years to happen.

What would you say about the accuracy of the initial "reserves" in new discoveries are? What are people talking about concerning new discoveries when they reference "reserves"? P5? P50? Whatever they are, are they good within some reasonable margin of error? Are they essentially worthless? In between? Let's take Kashagan (Kazakhstan) and Azedegan (Iran), the last two very large fields (>= 5 Gb) that have been discovered in my current research, as examples in this case.

I'm trying to work up a post on new discoveries and any insight you might give me here would be appreciated.
Of course this all depends on what you mean by reserves.  As you may recall, I posted about this earlier about reserves.  Where do the reserve estimates in these tabulations come from?  Are they from governments like Saudi Arabia or Nigeria that have reasons to inflate them?  Are they SEC published proved reserves that are likely underestimate of ultimate recovery?

A good rule of thumb for a field that is large and complex without significant production history is that P90/P10 ratios of EUR (estimated ultimate recovery) would range from 1.5 on the low end to as much as 5 on the high end.

Prudhoe Bay is a good example.  Early estimates of UR were on the order of 10 Gbo.  I am pretty sure the field has already exceeded 14 Gbo.

Estimating UR from fields is a tricky business because involves the extrapolation of sparse and imperfect data along with predicting the future of technological enhancements to recovery techniques within a world of economic uncertainty.