As any fan of the TV show "The West Wing" knows, for decades there has been a debate over whether it is appropriate to use sampling in the taking of the decennial US Census. Those in favor believe that sampling—taking a cross-section of the population and extrapolating data out for the whole country—is more efficient and, in fact, more accurate. Those against proffer the slippery-slope argument: if we use sampling for the US Census, why not conduct, say, elections that way?
Why not, indeed. While the idea of using what are essentially polls to determine the winners of general elections—when there is true power at stake—is certainly inappropriate, it might be a different story for a type of election that is already considered a bit of a farce in America today: the primary election. Although it could never be taken seriously by either major party, what if presidential nominees were chosen not by Iowa, New Hampshire, and the like, but by a poll instead?
The idea is not totally without merit. The current process is quite obviously flawed, giving undue weight to the two rural, non-diverse states of Iowa and New Hampshire. How would a poll be any worse? (The sample size might be smaller, yes, but it would be far more representative.) And there could be no intellectually honest arguments about how a primary-by-poll would subvert democracy; presidential nominating contests were never meant to be democratic. Until the late 20th century, parties' nominees for president were chosen in the famous smoke-filled rooms by party elders. Any nominee-selecting system that considers the input of the general public would be better than the system we thrived on for 200 years. And any other nominee-selecting system that considers the input of the general public would probably be better than the one we have, which still usually picks a nominee before a majority of the country has had a chance to cast their ballots one way or another.
Here's how it would work. A pollster, or series of pollsters, is contracted by each party. (Remember, the parties are private entities; they're free to choose whichever partisan, inaccurate, or book-cooking pollster(s) they wish. Again, I contend it's no more jerry-rigged than the current process.) The pollster prepares to conduct a massive, country-wide poll—of, say, 2,000 registered voters (normal polls have sample sizes of 500-1,000). These 2,000 voters are chosen by lottery to make the sample size as representative as possible of the American public—by race, sex, religion, educational background, family income, what have you. In this way, the nominee of the party is guaranteed to be chosen by a true cross-section of the nation—a nominee for all.
The parties then have two options: either go ahead and conduct the poll or announce the "winners" of the lottery. The latter has the potential to be brilliant and revolutionary or foolish and corrupting. By announcing the lottery winners—and therefore the 2,000 people who will decide their next presidential nominee—the party creates a Swing Vote–type scenario where each candidate is required to solicit the lottery winners' votes individually. (After all, if the random sampling was done properly, they are scattered all over the country and thus can't be reached en masse.) This removes the big-money element of campaigns, such as political advertisements, and (in an ideal world) puts the focus on each candidate's demeanor and positions on the issues. Namely, it facilitates the kind of retail politics that Iowa and New Hampshire are praised for—voters confront candidates face to face, while candidates are constantly reminded of the roots of their support and the concerns of everyday Americans.
True, it is somewhat likely that announcing the lottery winners would lead the campaign to devolve into a series of personal promises on the 2,000 voters' pet issues—not so different from big donors getting access to politicians that isn't available to poor commoners. And it's possible that the retail politicking could get so retail that it is conducted outside the view of the general public—resulting in a nominee who isn't as well known. So perhaps the chosen pollster should just conduct the poll immediately after all, and the winner of a plurality of votes would simply and easily be declared the nominee. (With such a large sample size, it's almost certain that the poll would be a close approximation of the results of a nationwide vote.)
Make no doubt about it, the primary-by-poll reform idea would radically transform the way presidential campaigns are run—but that's part of the point. Sure, it's a fanciful proposal. Yet inside-the-box thinking doesn't appear to be getting us any closer to a perfect presidential selection process. I'm happy to add one more kooky idea to the pile, just in the name of intellectual exploration. Maybe, just maybe, someday we'll come up with something so crazy, it just might work.