The dirty secret of most news Web sites is that their biggest viewership comes during the day, when people are at work and should be working. On Election Day in the U.S., you can't blame people for spending their work hours hunting around online for exit poll numbers, the supposedly closely guarded numbers that came this year from the National Election Pool, a group effort created by ABC, the Associated Press, CBS, CNN, Fox and NBC.
So a certain friend of mine spent a good part of his day Tuesday simply refreshing Slate's special page with exit poll numbers, keeping up on how well Kerry was doing throughout the day in various battleground states. Lucky for him, his boss was even more of a political junkie and was probably doing the same thing.
The problem? The numbers were far from prescient, and Kerry was really headed for defeat on Tuesday. Though Slate and Wonkette and Daily Kos (but not Drudge Report) all ran some sort of disclaimer about the exit poll numbers, saying to take them with a huge grain of salt, most readers who passed the numbers around excitedly were in for a real shocker that night when they turned out to be dead wrong.
While the mainstream media might be seen as a bit conservative in their efforts to hold back on calling elections and giving out early exit poll data, they were the ones who ended up looking good, while Slate was caught with a mid-day headline on exit poll numbers over a huge happy photo of Kerry. (Hey, why not title it "Dewey to Defeat Truman"?)
To add insult to injury, many of these sites crashed under the weight of all the curious-voter traffic, costing them readership and stability at perhaps their most crucial moment on Election Day. While it was the 2000 elections that brought down many of the big news sites and made them reconsider infrastructure, 2004 brought traffic concerns to independent media and political bloggers.
As for rationales, Slate media critic Jack Shafer, who ran the exit poll page, told me that the reasoning for publicizing the numbers was all about demystifying a process controlled by the media elite.
"Think of the exit poll as a secret tracking poll conducted for the elite," Shafer said. "All Slate is doing is giving civilians a look at the process that they've been locked out of previously. The exit poll numbers are being swapped from NEP to its clients to politicians and journalists to boardroom big shots today like crazy, so why shouldn't civilians have access to the information? I trust readers and voters to see the exit polls for what they are."
But if Slate truly saw them for what they were -- "a snapshot of an extremely fast-moving object," as Shafer says -- why trumpet them so prominently on their front page with a photo of Kerry looking triumphant? And even run a headline about how Bush might still pull it out (as if he were losing)?
According to Nielsen/NetRatings, Slate had the fourth biggest jump in traffic of all sites on Election Day, growing 169 percent to 412,000 unique visitors, faster growth than even on MSNBC.com that day. The Drudge Report reportedly had 36 million visits on Election Day, its best showing ever.
Blogger and political consultant Markos Moulitsas, who runs the top left-wing blog Daily Kos, was quick to post the exit poll numbers he found at colleague Jerome Armstrong's MyDD blog and from his own unnamed sources. Moulitsas says that, unlike the mainstream media, he isn't hemmed in by corporate rules or journalistic ethics.
"If someone says I'm being irresponsible, I can just say 'screw you,'" he told me. "I don't have a responsibility to my stockholders. I said these were for entertainment purposes. I know people got a lot of false hopes, but oh well. Most of the people who were really paying attention this time -- it was their first time tracking an election as closely as they did. ... All the caveats in the world didn't seem to accomplish much."
Holding the line -- and not
Not all bloggers were passing around exit poll numbers, however. Democratic pollster Mark Blumenthal, who only just recently started his Mystery Pollster blog, declined to run the numbers and threatened to take down any comments from people if they exposed them. Blumenthal did post probably the most in-depth explanation of how exit polls work and why they might be problematic.
"I just want you to know that those leaked exit polls really don't tell us much more about the outcome of the race than the telephone polls we were obsessing over just a few hours ago," Blumenthal wrote on the morning of Election Day. "Even if we wanted to call a race on unweighted, unfinished, mid-day exit polls alone (something the networks will not do), we would need to see differences of 10-15 points separating the candidates to be 95% certain of a winner."
Glenn Reynolds, law professor at the University of Tennessee and preeminent poli-blogger at Instapundit, also refused to run any of the exit poll numbers or link to them.
"I don't think [exit poll numbers are] all that useful unfiltered, and we'll know the real numbers (I hope) soon enough anyway," Reynolds told me earlier on Election Day. "Besides, that's why we have Drudge! On exit polls, there's a certain baffle-the-experts pleasure in seeing them turn out wrong, but on the other hand they are -- or should be -- a useful check on fraud."
Perhaps, this time it was the election results that were a useful check on the exit polls.
As the exit polls made the rounds online, stockbrokers caught wind of a possible Kerry win and pushed the broader U.S. markets downward. That gave financial wires like Bloomberg and CBS MarketWatch the opportunity to run stories that mentioned the exit polls in a second-hand way. The Australian was less circumspect about the polls in a story titled, "Blogs send jitters through Wall Street," noting that Drudge had posted numbers showing Kerry leading in Florida, Ohio, and Pennsylvania.
Drudge, for his part, ran no caveat about the numbers as he splashed them on his front page. However, perhaps after getting depressed about them, the conservative gadfly took them down in the early evening and ran the banner headline, "Enough of the media exits; let's count the people's votes."
The pain of popularity
Meanwhile, the prominent sites that did run exit poll numbers got more traffic than they bargained for. Visitors to MyDD.com had trouble getting access, and Daily Kos had to take measures to streamline the site, as well as expanding its bandwidth. Moulitsas told me he now spends $3,500 to $4,000 for hosting and bandwidth per month and has added three temporary servers to the five he already had to handle the increased traffic this week.
Moulitsas stripped some graphics, turned off traffic meters and even stripped out ads to make sure the site didn't go down. Daily Kos held up pretty well under the strain, though Moulitsas said there were some early outages when the exit poll numbers were first posted.
"When the first exit polls started coming out, all hell broke loose," he said. "I turned off my traffic meters because they were a drain on my resources. I was averaging about 800,000 visits per day before that. I would guess I was getting three to five times the normal traffic that I get."
Reynolds told me that Instapundit likely had its second most trafficked day on Nov. 2 -- after its high point during the RatherGate days -- even though he wasn't running exit poll numbers. He said he also had to take action to prepare for the heavy traffic in advance.
"I lightened the graphic load a bit, and my hosting company did some prep by increasing the number of connections servers are allowed to have open," Reynolds said. "It's been mixed -- access to the site was intermittent, though some of that problem is heavy traffic elsewhere on the Internet."
Throughout Tuesday, I had trouble getting to Wonkette, Instapundit and Daily Kos, and Josh Marshall's Talkingpointsmemo.com blog was down for a number of hours in the afternoon. The deluge of traffic for sites with and without exit polls shows just how much political blogs are becoming part of media consumption for a big event like an election. However, it's easier for a site like Slate, backed by Microsoft, to boost bandwidth than a small one-person blog operation.
Still, Slate editor Jacob Weisberg told me his site's traffic "swelled to a multiple of what we anticipated" due to people looking for exit polls and estimates the traffic was in the millions on Nov. 2. He admits that the site's performance was slower than usual, but after rejiggering some servers, they were able to handle the inundation.
The future of exit polls
No one knows exactly what will happen with exit polls in the next big election and whether the mainstream media will crack and start running them in newspapers and on TV in a reaction to the popularity of the Net.
Doug Feaver, executive editor at Washingtonpost.com, says that exit polls are useful for news organizations in finding out who the electorate was and why they voted the way they did. However, those numbers aren't released until after all the polls have closed, and Feaver doesn't expect that this longtime policy will change in the future.
"There are at least two excellent reasons not to release horse-race exit poll data," Feaver said via e-mail. "There is a distinct possibility it will be wrong, which doesn't help the credibility of the news business. Additionally, early release of the information might unfairly influence an election in a state where polls are still open. Just because you can do something doesn't mean you should."
So did the release of exit poll numbers online do more than just move the stock markets? Did they somehow affect voter turnout or hurt the credibility of bloggers or online media? Bill Grueskin, managing editor of Wall Street Journal Online, says that it's probably impossible to know. The Journal didn't run exit poll numbers, but did run an in-depth story on how they spread online.
"As a practical matter, it's impossible to say whether, or how, these numbers affected voting," Grueskin told me via e-mail. "Did the positive Kerry numbers encourage his supporters to vote, or did it make them feel they didn't need to? Did the numbers spur Bush supporters to get more people to the polls, or did it demoralize some of them and cause them to stay at home? I don't know the answer to those questions, but I suppose someone could find anecdotal evidence to back up responses to any of them."
Slate's Weisberg doesn't buy the theory that the poll numbers affected voter turnout and says he has no regrets about running them. "There is much dismay in the old media about how the Web and bloggers have broken their cartel, preventing them from withholding this kind of information on a dubious theory that it suppresses voter turnout," he said via e-mail. "But I don't think that publishing leaked exit poll data hurt our credibility with the public or our readers at all. To the contrary, readers come to Slate in part because of our transparency and directness on issues like this."
Reynolds, for one, thinks that the mainstream media will "probably" change their policies in the future for running exit polls. "And once they do, the numbers will lose their allure as forbidden fruit," he said, "and just become additional data points."
Slate's Shafer won't predict what the media companies will do down the line, but he thinks the NEP consortium might crack down on all the leaks from excited journalists e-mailing exit polls to friends.
"What more likely after this election cycle, I think, will be heightened security among NEP's owners and subscribers," Shafer said. "They'll restrict the number of people who can view the data and they'll poke out the eyes of anybody unauthorized to view it. They might even come hunting for me with a fiery red poker!"
Not bloody likely. Trying to staunch the flow of previously taboo information on the Net has proven difficult if not impossible on everything from videos of beheadings to names of sexual assault plaintiffs. More likely, this will become the accepted nature of exit polls: leaked to friends and bloggers, run up the pole and then likely ridiculed just like every other poll that ends up being wrong.