I don't understand the re-weighing of the poll numbers. The changing Rep/Dem breakdown is reflective of candidate support and partisan energy, and should not be tampered with. The 2000 turnout model will not apply to the 2004 election, and therefore is irrelevent. Gallup is historically the gold standard in election polling, and they are one of the few organizations that do not fall into the trap of artificially manufacturing partison samples based on the 2000 turnout.
The re-weighting exercise is to show what would happen with the Gallup internals if a Zogby-style (or ARG) variation of the use of the VNS year 2000 exit polls were used in the re-weighting. If you carefully read my post you will see this is for comparison purposes ONLY, not predictive.
dvwjr