I agree that the null should be handled by the data producer. My issue is that I update my data from a join to an oracle database on a regular basis. Once you have rows in your feature class, you CAN NOT add a non nullable field to the feature class (so, why does it even give you the option? just so it can give you an error I guess). So, every time I update it, it sticks all the nulls in there.
Also, any time you add a feature, it automatically assigns a null to every field you don't populate (since I can't set to not accept nulls)
I had my script, after the updates, go in and converted all the nulls, and modified the fields to not allow them. However, it seems that each service pack (except for 5) breaks something in the arcpy package so I have to go re-do all my scripts again. Finally just decided to let ESRI tell me what I "want" and live with it. So, now I have both blanks and nulls.
personally, I don't think the nulls should even exist. Well, for that matter, they don't :). In other words selct for item <> "non null value" does not select nulls. Seems like if it isn't equal to "non null value" it should select it, but appearantly, it is not not equal to it either. So far I've never came across a case where I would prefer a null over a blank.
Maybe now that version 10.1 is out and I have sp5 on my 10.0 I can re-visit this without it breaking (doubt there will be a service pack 6).
R_