Nate Silver rages at 'fucking idiotic' Huffington Post editor
48 replies, posted
[QUOTE=Smoot;51317567]lol you guys are way too into these statistics and polls. Honestly, just go vote on Tuesday and whoever wins, wins. Obviously alot of you are solid on who you want to vote for, so reading all this crap is just going to stress you the fuck out.[/QUOTE]
what's wrong with predicting? It can be fun and promote discussion.
Huffington Post is full of idiots anyway so idk why he's so pissed at them. They're free comedy.
[QUOTE=JaegerMonster;51316880]Is it just me or are people fundamentally ignorant of chance?
The average person seems to, if not think, certainly treat something that has a low % of happening as something that [B]won't[/B] happen, and vice versa, something that has a high % as something that will definitely happen.
This seems to be the source of a lot of either the schadenfreude or the misplaced anger at Silver's models.[/QUOTE]
This crops up all over the place. People have the most tenuous grasp of how probability works. The classic example is a deck of cards. Shuffle it, then lay them out in their shuffled order. The odds of that specific arrangement occurring are 1 in 52!, which is a fucking enormous number. 80,658,175,170,943,878,571,660,636,856,403,766,975,289,505,440,883,277,824,000,000,000,000, to be precise.
The "impossible" happens constantly, people just aren't aware of it.
polls are fucking dumb
[QUOTE=General J;51319793]polls are fucking dumb[/QUOTE]
Why?
[QUOTE=General J;51319793]polls are fucking dumb[/QUOTE]
Trump what are you doing here get back on the campaign trail
Trump is as likely to win Pennsylvania as Clinton is Alaska (ie. extremely improbable), and it seems to be often agreed that Trump needs both Florida and Pennsylvania to realistically win. From this position, I don't see why Trump is getting such a high percentage in their model. Still better than the ridiculous HuffPo model.
[QUOTE=FlashMarsh;51320030]Trump is as likely to win Pennsylvania as Clinton is Alaska (ie. extremely improbable), and it seems to be often agreed that Trump needs both Florida and Pennsylvania to realistically win. From this position, I don't see why Trump is getting such a high percentage in their model. Still better than the ridiculous HuffPo model.[/QUOTE]
Probably because there isn't an 100% chance Clinton will win PA and that factors into their non-100% chance Clinton national win.
Their percentage on a Trump win is 35%, but their percentage on Florida is 50%, and 25% for Pennsylvania, of which he needs to win both in almost all situations. How does this translate? I don't get it. And I think 25% is generous for Trump in Pennsylvania.
[QUOTE=FlashMarsh;51320030]Trump is as likely to win Pennsylvania as Clinton is Alaska (ie. extremely improbable), and it seems to be often agreed that Trump needs both Florida and Pennsylvania to realistically win. From this position, I don't see why Trump is getting such a high percentage in their model. Still better than the ridiculous HuffPo model.[/QUOTE]
afaik at this point his chances mostly just represent the possibility of the polls being wrong. the margin is slim enough that he could feasibly be ahead by one or two points nationally without us knowing, but if the polls are remotely accurate or they're off in a way that favors Clinton then he hasn't got a snowball's chance in hell.
[QUOTE=FlashMarsh;51320108]Their percentage on a Trump win is 35%, but their percentage on Florida is 50%, and 25% for Pennsylvania, of which he needs to win both in almost all situations. How does this translate? I don't get it. And I think 25% is generous for Trump in Pennsylvania.[/QUOTE]
Literally it means that in the 10,000 simulations they ran, Trump won the election 3,500 times
[editline]7th November 2016[/editline]
Florida is a knockout blow, but NBC have [url=http://www.nbcnews.com/storyline/2016-election-day/electoral-college-map-potential-paths-victory-clinton-trump-n678266]proposed[/url] this 'creative' path for Trump to get around a Pennsylvania loss
[t]http://i.imgur.com/jmVfVAO.png[/t]
[QUOTE=smurfy;51320249]Literally it means that in the 10,000 simulations they ran, Trump won the election 3,500 times
[editline]7th November 2016[/editline]
Florida is a knockout blow, but NBC have [URL="http://www.nbcnews.com/storyline/2016-election-day/electoral-college-map-potential-paths-victory-clinton-trump-n678266"]proposed[/URL] this 'creative' path for Trump to get around a Pennsylvania loss
[t]http://i.imgur.com/jmVfVAO.png[/t][/QUOTE]
Huh, they must think WI and MI are way closer than 538 because over there the path of least resistance is to not lose ME 2 and NC (54.7% and 51.2% Trump) and flip NH (61% Clinton).
The saving grace for Clinton on 538 is that her riskiest projected win is 61% (and she'd win if she got the easier ones as well) but a bunch of Trump's projected wins are almost 50/50.
[QUOTE=JaegerMonster;51316880]Is it just me or are people fundamentally ignorant of chance?
The average person seems to, if not think, certainly treat something that has a low % of happening as something that [B]won't[/B] happen, and vice versa, something that has a high % as something that will definitely happen.
This seems to be the source of a lot of either the schadenfreude or the misplaced anger at Silver's models.[/QUOTE]
Do people really believe this? Low chances are still chances unless it's extremely low. Trump's chance of winning is greater than me rolling a die and getting a 2.
[QUOTE=FlashMarsh;51320108]Their percentage on a Trump win is 35%, but their percentage on Florida is 50%, and 25% for Pennsylvania, of which he needs to win both in almost all situations. How does this translate? I don't get it. And I think 25% is generous for Trump in Pennsylvania.[/QUOTE]
I guess there (first of all) situations where Trump wins Florida, loses Pennsylvania, but still manages to win.
States are also interconnected in 538's model - Trump winning some other state in some simulations may impact his chance of winning Pennsylvania. As in, Trump may have a 25% a priori risk of winning Pennsylvania, but if he wins some other state that says something about the polling in Pennsylvania, and how off it might be - that's new information that is put into the model. I obviously haven't got a clue, but it's definitely more complicated than simply the risk of Trump winning a state he "needs" to win.
[QUOTE=Daniel Smith;51320744]Do people really believe this? Low chances are still chances unless it's extremely low. Trump's chance of winning is greater than me rolling a die and getting a 2.[/QUOTE]
Well, not necessarily. You need something to happen to build probabilities and what Nate has done is tried to emulate the election, and simulate it thousands of times to come up with the probability. So the dice comparison doesn't really work because we actually know the chances of that based on reality
It's an extreme example but it's like when creationists say "THINK OF THE ODDS OF LIFE HAPPENING." All that is is speculation.
[QUOTE=smurfy;51320249]Literally it means that in the 10,000 simulations they ran, Trump won the election 3,500 times
[editline]7th November 2016[/editline]
Florida is a knockout blow, but NBC have [url=http://www.nbcnews.com/storyline/2016-election-day/electoral-college-map-potential-paths-victory-clinton-trump-n678266]proposed[/url] this 'creative' path for Trump to get around a Pennsylvania loss
[t]http://i.imgur.com/jmVfVAO.png[/t][/QUOTE]
He doesn't need PA in the first place if he takes NH and ME 2nd District
[QUOTE=Stick it in her pooper;51321678]He doesn't need PA in the first place if he takes NH and ME 2nd District[/QUOTE]
And not lose NC, as I said.
[QUOTE=FlashMarsh;51320108]Their percentage on a Trump win is 35%, but their percentage on Florida is 50%, and 25% for Pennsylvania, of which he needs to win both in almost all situations. How does this translate? I don't get it. And I think 25% is generous for Trump in Pennsylvania.[/QUOTE]
Outside of Pittsburgh and Philly, PA is a fairly conservative state. The central part of the state always votes red.
[QUOTE=Emperor Scorpious II;51323114]Outside of Pittsburgh and Philly, PA is a fairly conservative state. The central part of the state always votes red.[/QUOTE]
Same way with New York usually.
Sorry, you need to Log In to post a reply to this thread.