As the networks cover returns from California, you can be sure they'll offer projections, even after only a small percentage of precincts have reported.
Imagine, hypothetically, that California had 130 precincts. Imagine if, after returns were in from 10 of these, 100 percent of those returns showed voters in favor of the recall, with 0 percent opposed.
There's little doubt, in such a scenario, that the news networks would (appropriately) inform us that: "With 7.6 percent of precincts reporting, 100 percent of voters support the recall." The networks would reasonably project from these early returns and make a safe prediction about the eventual outcome.
For example, there are approximately 130 known Iraqi Ammunition Storage Points (ASPs), many of which exceed 50 square miles in size and hold an estimated 600,000 tons of artillery shells, rockets, aviation bombs and other ordinance. Of these 130 ASPs, approximately 120 still remain unexamined.
Kay is 0-for-10. He wants us to believe that this makes it likely he will end up 120-for-130.
That's not how these projections are usually done.