home studios equipment staff/friends booking/rates for sale forum contact

Dude: Nate Silver

Vote and debate.

Moderators: kerble, Electrical-Staff

Dude: Nate Silver

Person who renders all pundits, talking heads, and self-proclaimed experts useless
32
71%
NEEEEEEEEEEERDD!!!!
13
29%
 
Total votes : 45

Re: Dude: Nate Silver

Postby jimmy spako on Wed Nov 09, 2016 9:50 am

Can we please be done with fantasy football politics forever now?

Done with the retreat into the last bastions of snarky humor and armchair wonky überhyperanalysis of the horse race? It's pathetic.
Isabelle Gall wrote:'Do not go smoothly into that dark night'
User avatar
jimmy spako
Man with Encyclopedic Knowledge
Man with Encyclopedic Knowledge
 
Posts: 5376
Joined: Sun Dec 02, 2007 10:49 am
Location: schnitzel circuit

Re: Dude: Nate Silver

Postby cerebralheadtrip on Wed Nov 09, 2016 11:54 am

J Temperance wrote:It's hilarious how badly Silver misjudged the Trump phenomenon. Literally every fucking step of the way, Nate's been wrong. By a lot.


in fairness, pretty much everyone misjudged Trump. probably his own campaign included. one of the big takeaways from today is that despite all the advances in technology, we are still completely incapable of predicting, or even accurately observing, voter sentiment. the media needs to take a long hard cold look in the mirror because they are routinely missing the biggest elephants in the room. and that means the portrait of reality we as citizens are being presented is fundamentally flawed.
cerebralheadtrip
forced to change shirt
forced to change shirt
 
Posts: 1913
Joined: Fri Jan 22, 2010 3:10 pm

Re: Dude: Nate Silver

Postby Gramsci on Wed Nov 09, 2016 2:24 pm

Nassim Taleb, the author of Black Swan has been having an ongoing low level twitter war with Silver for years.

Needless to say, Taleb was right.
clocker bob may 30, 2006 wrote:I think the possibility of interbreeding between an earthly species and an extraterrestrial species is as believable as any other explanation for the existence of George W. Bush.
User avatar
Gramsci
World's Greatest Writer
World's Greatest Writer
 
Posts: 10705
Joined: Thu Aug 28, 2003 8:43 am
Location: The People's Republic

Re: Dude: Nate Silver

Postby Wood Goblin on Wed Nov 09, 2016 2:35 pm

Taleb said that a Trump victory was "very possible."

Which is, um, the same thing Silver said.
User avatar
Wood Goblin
Heaven-Sent Hero
Heaven-Sent Hero
 
Posts: 6081
Joined: Tue Jun 06, 2006 12:43 pm
Location: South Loop, Chicago

Re: Dude: Nate Silver

Postby Anthony Flack on Wed Nov 09, 2016 3:14 pm

Yeah I have been looking at those numbers all week thinking "30% = extremely fucking possible". Russian roulette with two bullets.

A bit like the predicted possibility that we suffer the worst, most catastrophic projected outcomes for climate change.
Anthony Flack
Present-day God
Present-day God
 
Posts: 9596
Joined: Fri Aug 03, 2012 12:27 am
Location: New Zealand

Re: Dude: Nate Silver

Postby Anthony Flack on Wed Nov 09, 2016 4:23 pm

prowler wrote:yeah the only good part i can see right now is watching how Nate&co will explain themselves out of this one.


The only good part I can see is that at least the Clintons can fuck off now.
Anthony Flack
Present-day God
Present-day God
 
Posts: 9596
Joined: Fri Aug 03, 2012 12:27 am
Location: New Zealand

Re: Dude: Nate Silver

Postby Angus Jung on Wed Nov 09, 2016 5:07 pm

Anthony Flack wrote:[The only good part I can see is that at least the Clintons can fuck off now.

Not just the Clintons - the whole neoliberal order of the U.S. just got a giant middle finger.
The entire Democratic Party Establishment is finished.
Good article.
Angus Jung
Present-day God
Present-day God
 
Posts: 9454
Joined: Mon Aug 04, 2003 7:04 pm
Location: High Desert

Re: Dude: Nate Silver

Postby jbrobertson on Wed Nov 09, 2016 7:11 pm

Maybe the real problem is not so much the forecasts themselves -- compare the surprise freak storm in the UK some years back -- as that people trust in the polls too much. That is to say: Did too many Democrats see the huge lead Clinton had in the polls and decide they didn't need to vote? Would more people have voted if the polls didn't exist? From what I have read, it seems Clinton lost not because of some kind of demographic shift or Republican surge but rather because fewer Democratic voters showed up overall. Romney actually got more votes than Trump did, and yet Romney lost.

In any case, I wouldn't be quite so angry if the majority of Americans had voted, instead of only about half.
jbrobertson
envoy
envoy
 
Posts: 38
Joined: Thu Jul 28, 2016 6:53 pm

Re: Dude: Nate Silver

Postby prowler on Thu Nov 10, 2016 6:45 am

jimmy spako wrote:Can we please be done with fantasy football politics forever now?

Done with the retreat into the last bastions of snarky humor and armchair wonky überhyperanalysis of the horse race? It's pathetic.

this.

obviously most pundits and aggregators are worse than Silver, but just for a moment let's not grade this on a scale. the simple fact is that there's no point at which 538 "got" Trump.* the numbers and the modelling helped, but they weren't enough.

*judging by the first reaction podcast, they still don't. i should probably unsubscribe...
User avatar
prowler
meatball enthusiast
meatball enthusiast
 
Posts: 1259
Joined: Wed Jun 27, 2012 4:54 am
Location: Bucharest

Re: Dude: Nate Silver

Postby matthew on Fri Nov 11, 2016 6:27 am

prowler wrote:...the simple fact is that there's no point at which 538 "got" Trump.


I think Mr. Silver totally "got" Trump. That's why he campaigned against him using probability and mathematics as his guise so as to sound "detached" and "objective". LOL.

In the end, he's just another goofball who is a discredit to both the study of probability and the field of mathematics. Shame on him.

He might want to move to Ashland, WI or Iron City, MI or East Grand Forks, MN and take bets from the locals on what the Powerball drawing is gonna be.
...diesque nostros in tua pace disponas...
User avatar
matthew
Master of Lit., Arts, & Architecture
Master of Lit., Arts, & Architecture
 
Posts: 3503
Joined: Sat Jan 29, 2005 3:38 pm
Location: Gens non sancta.

Re: Dude: Nate Silver

Postby jbrobertson on Mon Nov 14, 2016 8:27 pm

There's a good postmortem of the election, including a talk with Nate Silver, on the latest On the Media podcast.
jbrobertson
envoy
envoy
 
Posts: 38
Joined: Thu Jul 28, 2016 6:53 pm

Re: Dude: Nate Silver

Postby Gramsci on Wed Nov 16, 2016 3:07 am

Wood Goblin wrote:Taleb said that a Trump victory was "very possible."

Which is, um, the same thing Silver said.


No, Silver said is it was highly unlikely but not impossible.

To be fair on Silver he did put his thumbs on the scale of his model a little to make it more conservative (small c). He said so in his podcast. I listened to every one in the lead up to the election and yes, Silver and is colleague said "it's not impossible" but they definitely did not say it was "highly likely"

As for Taleb, his schtick is "never discount anything" based on a very long term view of history and he is deeply distrustful of any "predictive science", especially Economics. I've read all his books and listened to a lot of interviews with him. He is a partial elitist dipshit and oddly enough self describes as a conservative (small c) leftist.

The disagreement Taleb has is with the idea of predictive science in general, not individual results.
clocker bob may 30, 2006 wrote:I think the possibility of interbreeding between an earthly species and an extraterrestrial species is as believable as any other explanation for the existence of George W. Bush.
User avatar
Gramsci
World's Greatest Writer
World's Greatest Writer
 
Posts: 10705
Joined: Thu Aug 28, 2003 8:43 am
Location: The People's Republic

Re: Dude: Nate Silver

Postby Anthony Flack on Wed Nov 16, 2016 4:13 am

Maybe it really was highly unlikely.
Anthony Flack
Present-day God
Present-day God
 
Posts: 9596
Joined: Fri Aug 03, 2012 12:27 am
Location: New Zealand

Re: Dude: Nate Silver

Postby matthew on Wed Nov 16, 2016 5:21 am

Gramsci wrote:
As for Taleb, his schtick is "never discount anything" based on a very long term view of history and he is deeply distrustful of any "predictive science", especially Economics. I've read all his books and listened to a lot of interviews with him. He is a partial elitist dipshit and oddly enough self describes as a conservative (small c) leftist.

The disagreement Taleb has is with the idea of predictive science in general, not individual results.


I've read some of Taleb and I largely agree with his anti-economist economics, though his philosophical foundation leaves room for a lot of questions in that they are ultimately Kantian (but that's neither here nor there. I'm a mere realist who excised myself from Kant.).

Anyway...he has a point about predictability. It's been an ongoing debate within the philosophy of science for decades. I side with him in that debate in that I think that predictability is not a prerequisite for all sciences. That said, I question the legitimacy of probability studies as a science in the first place.
...diesque nostros in tua pace disponas...
User avatar
matthew
Master of Lit., Arts, & Architecture
Master of Lit., Arts, & Architecture
 
Posts: 3503
Joined: Sat Jan 29, 2005 3:38 pm
Location: Gens non sancta.

Re: Dude: Nate Silver

Postby Gramsci on Wed Nov 16, 2016 2:10 pm

Anthony Flack wrote:Maybe it really was highly unlikely.


The point we see from polls is that they are not nearly as predictive as pollsters would have you think.
clocker bob may 30, 2006 wrote:I think the possibility of interbreeding between an earthly species and an extraterrestrial species is as believable as any other explanation for the existence of George W. Bush.
User avatar
Gramsci
World's Greatest Writer
World's Greatest Writer
 
Posts: 10705
Joined: Thu Aug 28, 2003 8:43 am
Location: The People's Republic

Re: Dude: Nate Silver

Postby joelb on Wed Nov 16, 2016 2:45 pm

Gramsci wrote:
Anthony Flack wrote:Maybe it really was highly unlikely.


The point we see from polls is that they are not nearly as predictive as pollsters would have you think.


I still don't get this discussion. Silver gave Trump 28% the night before the election. He was talking regularly up to election day about the lack of recent polling in MI and WI and trying to make sense in a vacuum. He was giving Trump nearly even chances in FL and NC. He predicted Clinton winning the popular which she did. PA seems the most egregious example of being off but the rest not so much.

Saying "The odds are that Clinton will win" is a helluva lot different than saying "Clinton will replicate recent polling data." Silver said the former not the latter. I don't always think people want to hear this.

Bottom line 28 percent is a lot. A lot of things had to break Trump's way and they did.
DrAwkward wrote:If SKID ROW likes them enough to take them on tour, they must have something going on, right?
joelb
tired of birds
tired of birds
 
Posts: 1119
Joined: Mon Dec 19, 2005 12:16 pm
Location: City by the sea

Re: Dude: Nate Silver

Postby Anthony Flack on Wed Nov 16, 2016 2:53 pm

I think a lot of people see something like "20% chance to win" and think that means "no chance to win". The reality was it was extremely close and Trump only just squeaked in.

I think Clinton is right, in a way, to say the Comey letters cost her the election at the last minute. Which is outrageous, but the polls showed a huge swing from that. Exit polls said most people claimed to have already made up their mind before the Comey letters, but if you can't trust a poll to tell you what people think right now, I would put even less faith in a poll that tells you what people retrospectively claim they were thinking weeks ago which is different to what they said at the time.

I think it's equally true to say Clinton lost the election herself before it even began, by taking all that speech money. And several other factors, ALL of which combined into a perfect storm that narrowly pushed Trump over the top.
Anthony Flack
Present-day God
Present-day God
 
Posts: 9596
Joined: Fri Aug 03, 2012 12:27 am
Location: New Zealand

Re: Dude: Nate Silver

Postby jbrobertson on Wed Nov 16, 2016 3:26 pm

Anthony Flack wrote:I think a lot of people see something like "20% chance to win" and think that means "no chance to win". The reality was it was extremely close and Trump only just squeaked in.

I think Clinton is right, in a way, to say the Comey letters cost her the election at the last minute. Which is outrageous, but the polls showed a huge swing from that. Exit polls said most people claimed to have already made up their mind before the Comey letters, but if you can't trust a poll to tell you what people think right now, I would put even less faith in a poll that tells you what people retrospectively claim they were thinking weeks ago which is different to what they said at the time.

I think it's equally true to say Clinton lost the election herself before it even began, by taking all that speech money. And several other factors, ALL of which combined into a perfect storm that narrowly pushed Trump over the top.


Exactly. Quite a few people, including many journalists who should know better, seem to misunderstand how statistical forecasting works. Forecasters look at hundreds or thousands of possible futures and make their predictions based on that. When Nate Silver says that Trump has a 28% chance to win, he means that in 1,000 possible futures, Trump wins in 280 of them. Seems pretty high to me.

Who hasn't gotten wet on a day the forecast gave a 30% chance of rain?

The freak storm that hit the UK a while back did not cause any UK meteorologists to change their forecasting methods. Their methods were sound and based on the available data, they made the right call. As you say, people make the mistake of thinking "unlikely" means "impossible." Also, "accurate" does not mean "perfect" or "infallible." Silver makes mathematical, not psychic, predictions. Further, as he says in the On the Media interview, the polling data he had to work with was not as good this year.

What we saw may indeed have been the Perfect Shitstorm.

If Silver can be blamed for anything, it may be that he should have done a better job educating the public about the true nature of statistical prediction and making clear that the polling data was not as good this year.

Finally, as I wrote above, it seems likely that the poll numbers and forecasts could have convinced at least some people that they didn't need to vote, which could have flipped the election by itself. Hard to account for these kinds of attitudes in a forecast.
jbrobertson
envoy
envoy
 
Posts: 38
Joined: Thu Jul 28, 2016 6:53 pm

Re: Dude: Nate Silver

Postby Gramsci on Thu Nov 17, 2016 3:07 am

joelb wrote:
Gramsci wrote:
Anthony Flack wrote:Maybe it really was highly unlikely.


The point we see from polls is that they are not nearly as predictive as pollsters would have you think.


I still don't get this discussion. Silver gave Trump 28% the night before the election. He was talking regularly up to election day about the lack of recent polling in MI and WI and trying to make sense in a vacuum. He was giving Trump nearly even chances in FL and NC. He predicted Clinton winning the popular which she did. PA seems the most egregious example of being off but the rest not so much.

Saying "The odds are that Clinton will win" is a helluva lot different than saying "Clinton will replicate recent polling data." Silver said the former not the latter. I don't always think people want to hear this.

Bottom line 28 percent is a lot. A lot of things had to break Trump's way and they did.


Here's a discussion about Taleb vs Silver:

https://www.quora.com/Whats-Nassim-Tale ... at-he-does

Which gets rather granular here:

https://mishtalk.com/2016/08/07/nassim- ... of-tweets/

At the root Taleb's problem is some making money as a Probability Sage. I listened to all of the 538 podcasts during the cycle. Silver was consistently way off during the Republican primary regarding Trump's chances, as we almost everyone. The actual race between HRC and Trump was almost entirely focused on how much she would win by, even when it was getting 50-50%.

That said Silver and team covered their asses with the occasion mention that Trump could win, but it was highly unlikely. In the end his model was way off compared to his near perfect predictions on 2008. Silver has rarely had the same level of success predicting since that election making it seem that was more a fluke than anything else. Remember that Silver was once lauded as a predictor, not wobbly probably wonk. To his credit since his accuracy decline since 2008 he has rebranded 538 as more about probability than prediction. Maybe it's just the media that haven't caught up.

What good is predictive analytics if they don't actually predict anything?
clocker bob may 30, 2006 wrote:I think the possibility of interbreeding between an earthly species and an extraterrestrial species is as believable as any other explanation for the existence of George W. Bush.
User avatar
Gramsci
World's Greatest Writer
World's Greatest Writer
 
Posts: 10705
Joined: Thu Aug 28, 2003 8:43 am
Location: The People's Republic

Re: Dude: Nate Silver

Postby Parbuckle on Thu Nov 17, 2016 2:45 pm

Anthony Flack wrote:I have been looking at those numbers all week thinking "30% = extremely fucking possible". Russian roulette with two bullets.

Yep.

joelb wrote:
Gramsci wrote:
Anthony Flack wrote:Maybe it really was highly unlikely.


The point we see from polls is that they are not nearly as predictive as pollsters would have you think.


I still don't get this discussion. Silver gave Trump 28% the night before the election...Bottom line 28 percent is a lot.

Yep.

Anthony Flack wrote:I think a lot of people see something like "20% chance to win" and think that means "no chance to win".

Yep.

jbrobertson wrote:Quite a few people, including many journalists who should know better, seem to misunderstand how statistical forecasting works. Forecasters look at hundreds or thousands of possible futures and make their predictions based on that. When Nate Silver says that Trump has a 28% chance to win, he means that in 1,000 possible futures, Trump wins in 280 of them. Seems pretty high to me.

Who hasn't gotten wet on a day the forecast gave a 30% chance of rain?

The freak storm that hit the UK a while back did not cause any UK meteorologists to change their forecasting methods. Their methods were sound and based on the available data, they made the right call. As you say, people make the mistake of thinking "unlikely" means "impossible." Also, "accurate" does not mean "perfect" or "infallible." Silver makes mathematical, not psychic, predictions. Further, as he says in the On the Media interview, the polling data he had to work with was not as good this year.

What we saw may indeed have been the Perfect Shitstorm.

If Silver can be blamed for anything, it may be that he should have done a better job educating the public about the true nature of statistical prediction and making clear that the polling data was not as good this year.

Finally, as I wrote above, it seems likely that the poll numbers and forecasts could have convinced at least some people that they didn't need to vote, which could have flipped the election by itself. Hard to account for these kinds of attitudes in a forecast.

Exactly.

Gramsci wrote:To [Silver's] credit since his accuracy decline since 2008 he has rebranded 538 as more about probability than prediction.

There's really no difference between "probability" and "prediction". I suppose you could say that "probability" refers to the numbers themselves (usually expressed as percentages) and "prediction" puts the same information into everyday language. For example:

Probability: Clinton 80% Trump 20%

Prediction: Clinton is very likely to win, but a Trump win is not impossible.

The moment you make a statement of what you believe the probabilities are, you are in effect making a prediction.

Whilst I can claim no great insight into the specifics of the recent election (being English/British/European I found the whole thing depressing enough at a distance), I'm inclined to think the "failure" of the pollsters can be attributed at least as much to the public's misunderstanding of them as to the pollsters themselves (as more than one FM that I've quoted above has effectively said).
Paragraph Eleven, Confirm
Parbuckle
chimp
chimp
 
Posts: 707
Joined: Wed Apr 06, 2016 12:28 pm
Location: Turn left at Europe

PreviousNext

Return to Crap / Not Crap

Who is online

Users browsing this forum: DrAwkward and 16 guests