The article is by Jay Caspian King, who writes:
[Silver] said the site’s big launch was also accompanied by “a lot of bragging that was kind of stupid relative to the quality of the product.” But, like any honest gambler who can see the swings in his career with some clarity, he attributed many of FiveThirtyEight’s later troubles to the inevitable regression back to the mean. “You’re just on this crazy winning streak,” he said, referring to his proficiency in picking the 2008 and 2012 elections. “It’s like making the World Series of Poker final table in back to back years or something—where you have two cycles where you’re outperforming your expectations. Inevitably, you’re going to get shat on.”...
“It’s very weird to become very well known for the wrong reasons,” Silver said. “People say, ‘Oh you have numbers and therefore a lot of certainty’ and they can’t quite process the fact that you can use numbers to quantify uncertainty as well.”
Silver, for his part, doesn’t quite seem to know what to do next...
22 comments:
The 2008 and 2012 elections were probably the most predictable ones of the last 50 years. I suspect that other polling firms lagged behind the social media and smartphone revolution and may have stuck with older data collection methods too long.
Survey research is just a combination of data collection and mathematical models. It's never as precise as "rocket science," per vague human opinions as shown by the published margin of error (e.g., 2% to 4%). But, Nate Silver told the urban left what they wanted to hear about Obama and thereby became a living saint.
I will never forget election night 2016. I could've cut diamonds with my schadenboner if I'd wanted to. For a long time my disqus profile avatar was a picture of a haggard and crestfallen Silver looking like he'd seen the dark side of 7 or 8 bad vodka-tonics with stale limes.
...Then there were the crying feminists at the Hillary balloon drop that didn't happen, le piece de resistance. I haven't felt that good since my threesome with 2 Singaporean supermodels at the Swiss-Hotel Merchant Court. Come on 2024...I need to feel that feeling again!
I will never forget election night 2016. I could've cut diamonds with my schadenboner if I'd wanted to. For a long time my disqus profile avatar was a picture of a haggard and crestfallen Silver looking like he'd seen the dark side of 7 or 8 bad vodka-tonics with stale limes.
...Then there were the crying feminists at the Hillary balloon drop that didn't happen, le piece de resistance. I haven't felt that good since my threesome with 2 Singaporean supermodels at the Swissotel Merchant Court. Come on 2024...I need to feel that feeling again!
diversity (i.e. color judgment, class-based bigotry)
Silver was a useful tool once. He is no longer, so he gets dumped.
Silver's star started to dim when he gave Trump a better chance of winning than Democrats wanted to hear in 2016, and even then Silver only gave Trump a 30% chance of winning. His fans wanted to hear 0%, and I think that they blamed him in some way for Shelob's loss.
Enigma is correct there was no doubt about the outcome of the 2008 election, and predicting Obama's reelection wasn't much a reach either. I did both predictions without collecting any data whatsoever, and in presidential election years, all close Senate races generally follow the which Presidential candidate wins the state. That isn't polling either- it is just electoral historical analysis.
Silver, for his part, doesn’t quite seem to know what to do next...
Silver's only reason for existing was telling people Hillary Clinton was going to beat Donald Trump.
That's why he can't find a job any more.
Once Silver's numbers stopped telling leftists what they wanted to hear, he was no longer useful to them.
If people understood the difference between statistics and reality, there would be no climate change (global warming) cult.
"... Hey, we're going to take these nerdy white guys..."
Great. Now I can't get Weird Al's 'White & Nerdy' out of my head.
The great ones don’t revert to the mean.
Flash in the pan.
Somehow, among the pundit class, Nate Silver got a reputation as a stats genius for applying the Law of Large Numbers to the work of actual pollsters. Congrats to him for being the guy who picked up that $100 bill that had been lying on the sidewalk for so long.
As caller id became ubiquitous and unbiased sampling became much more difficult, the sampling error for Nate's samples became much larger than he initially realized.
Wonder what the relationship was between "a lot of bragging that was kind of stupid relative to the quality of the product" and becoming "very well known for the wrong reasons". Seems like he rode those wrong reasons to the bank.
100% agree with Not Sure. Business boomed when adding up all the poll results and dividing by the number of polls was new and effective. When it became old and ineffective, as polling tanked in accuracy and others copied the simple stats, Nate became a pundit just like any other.
Read "Positively Fifth Street" by James McManus to get a feel for what successful poker players do. They play the odds, but they don't win every time. They absolutely know the odds on every hand.
Nate's issues are the issues that anyone interpreting public opinion surveys has today. It is hard to get a good sample.
Like the rest of us mortals, Nate can just roll the dice again and hope for the best.
Got to agree about the easy calls on the Obama elections--nearly as predictable as the . . . sunrise!
"Silver's only reason for existing was telling people Hillary Clinton was going to beat Donald Trump.
That's why he can't find a job any more."
boom.
Perhaps the surprise in 2008 and 2012 was that Obama was able to get such high turnout of blacks--in the full knowledge that if they voted, they would vote D--and was able to do so well among whites. The latter means that the U.S. was much less racist than it used to be, it is not true, as Obama has lied recently, that it is still a Jim Crow society, etc.
Why do blacks keep voting D when the Dems keep screwing them? The sprinkles of cash, which Republicans seem to object to? Culturally the Dems have a way about them, going to the funerals and saying the right things, Republicans would rather stay away?
Obama got at least some votes from white folks who thought that voting for him would serve as evidence that they weren't racist.
Thomas Sowell and others have a theory that traces the dysfunction of urban B/black culture to their appropriation--with modifications--of the old South's shiftless po'-white redneck peckerwood ways of being.
In politics, the modern D party's B/black leaders revive their Southern white predecessors' racial divisiveness, ignorance, greed, ruthlessness, and fanatical urge to power.
Perhaps the surprise in 2008 and 2012 was that Obama was able to get such high turnout of blacks
Today, the People of Color are deported from his community in Martha's Vineyard. Hope and change. Yes, he can... throw another baby... fetal-baby on the barbie, it's over.
"Obama got at least some votes from white folks who thought that voting for him would serve as evidence that they weren't racist."
Not racist? Race is just about all Democrats obsess over.
@Daniel12
Nope, Silver never did anything that simple, that was RCP I think that just aggregated poll numbers. Silver always claimed he had a 'special sauce' of weighting polls by 'reliability' which really meant, as he was making his bones in 2008 and 2012, they were closest to agreeing with the internal data from the Obama campaign that he was being feed.
Christopher, yes it was marginally more complicated (mostly by starting with a fundamentals driven model and the weighting polling more and more, the closer to the election).
Re Obama, I guess you're saying his campaign had pretty accurate polling.
Post a Comment