FiveThirtyEight.com and why Election Night will never be the same

Take a look at this map. This is the map generated on the eve of Election Day 2008 for the website FiveThirtyEight.com, the brainchild of baseball statistics guru Nate Silver, who decided to apply the same complicated regression analysis and whatnot he developed for baseball to the higher-stakes game of presidential polling.

I should have also downloaded the map for the day before, because then you'd see how incredibly accurate Silver's models were. Basically, they were finding that in 48 states, voter sentiments had hardened to the point that the computer could safely predict (based on 10,000 daily election simulations) whether that state's electoral votes would go to John McCain or Barack Obama. And in two states the race was too close to call, even after 10,000 tries:Missouri, seen here without a color, and North Carolina, which is barely going blue in this graphic but was as pale as the Show-Me State in the previous night's map.

And here's how it turned out:

Silver's models came up with a final projection of 348.6 electoral votes for Obama. He currently has 349 electoral votes won. Forty-eight states divvied up their ballots almost exactly as Silver's models predicted -- and in the coup de grace, the computer even nailed the two states that are still, as I type this, too friggin' close to call.

Silver was all over TV in the days leading up to the election. Everyone knew his models were working, and wasn't this the same pasty-faced kid who said the Tampa Bay Rays would not only have its first winning season but win 90 games this year?

OK, the Rays won 97, but here's my point. In four years, I think the tables will be turned. People won't be turning to TV to see a dot-commer give them the election scoop. People will be going online and watching live video at pure-play dot-com sites. TV will not rule the roost. If the YouTube phenomenon can completely revolutionize viewing of streaming recorded video in just one election cycle, what idiot would deny that its live -- even live high-definition -- component won't be on its way before President Obama runs for a second term?

Previously on TV Barn: And that's why I argued earlier this week that it makes no sense for the networks to hold onto their quaint practice of not calling states until the polls there have closed.

Comments

[Joe Hass]

Indeed, what we saw last night was a massive overcorrection from 2000 and 2004, where networks seemed fearful of calling anything. Anyone with a calculator knew that at 10:00 pm ET, when Obama broke 197, the ballgame was over. That hour of "coverage" was wasted breath, networks killing time while ignoring the elephant in the corner of the room.

[jeano]

Next big election, I'm turning off the TV and going online. Absolutely awesome!

[DonBoy]

I loved the site all year, but here's a slightly contrarian take: how much of it comes down to "the late opinion polls were right after all"?

[Peyton]

"In four years, I think the tables will be turned. People won't be turning to TV to see a dot-commer give them the election scoop. People will be going online and watching live video at pure-play dot-com sites."

And that's exactly what I and my friends did Tuesday night; one window open to fivethirtyeight.com to read the comments, one window open to MSNBC.com, one window open to Daily Show coverage and one window open to surf.

I'll never go back to televised coverage again.