Being the political junkie with endless demands for information that I am, I find it emotionally difficult to accept his case. Intellectually, however, I must admit that it is a good one:
Remember 2000? Early exit poll numbers released midday on Election Day showed George W. Bush tied with Al Gore in Mississippi, Colorado, Arizona and Alaska, and had Bush losing Florida. It was virtually impossible for Bush to prevail without Florida, yet those early poll results encouraged commentators to put the state into Gore’s column as early as 7:48 p.m., while voting continued in the Florida panhandle.
This affected turnout. The 24 states where polls closed after 8:30 p.m. Eastern time saw turnout rise by 2.3 percent over the 1996 election. But in the 26 states and the District of Columbia where polls closed before 8:30, turnout increased 2.9 points over 1996. This meant that more than 400,000 voters stayed home in the central and western United States, most of them likely Bush voters. This potentially affected New Mexico (which Gore won by 366 votes) and Oregon (where he won by 6,765).
[. . .]
On election nights, networks feel pressure to display early exit poll numbers in snazzy graphics to explain what groups are breaking what way. But these numbers almost always differ from the final version of the exit polls, after gurus crunch the data during the evening.
Rove states that if we cannot break our addiction to exit polls, we ought to at least wait to see them until actual polls close. Sounds reasonable to me; though I suppose that this may heighten the temptation on the part of some to leak the polls.