Seth Rich story resurfacing

[fskimospy]

lol 538 gave Trump losing pop vote by 2% but still winning probably only about a 1% chance.

This is false, see above.

 

[Maxima1]

Well it sounds like you need to talk to Mr. Wang about this because HE stated that it was either a 99% or 95% probability many times. You appear to be misreading the article you are citing as well as he explicitly states that taking an average of recent elections is a bad idea:



He states right after that part that he was NOT using a 1.5% error term in his model:



You are claiming things about his model that he explicitly states are not the case. I would suggest you go back and read your own link more carefully.

Well duh. He was being silly. Using 0.5-0.8% obviously wasn't supported. His judgment was clouded by trying to get people to focus on the other races instead of the presidency.

 

[Maxima1]

This is false, see above.

That's not false. I remember Nate Silver mentioning how the EC could produce a winner with a 2% pop vote deficit, but it was unlikely. Sorry, but I don't know why people give his model so much praise. This is why he was hedging because he knows not slanting too far either way will get people to say his model is correct no matter who wins. He was already humiliated completely by the Cubs and Trump prior to this and was on the ropes.

 

[Jhhnn]

You seem to be unaware of all the details and now you are making stupid assumptions. Assange doesn't need to tell us Seth Rich is the source. He has given us facts that lead to that conclusion. And there is absolutely no way for the details Assange has given to apply to "Russian hackers".\

Assange is giving the chumps enough conspiracy theory rope to hang themselves. Even Conservapedia allows that possibility-

Certain statements by Assange open the possibility that Seth Rich was a source (or a believed source) for Wikileaks; however, the possibility is not discounted that Assange was giving disinformation:

http://www.conservapedia.com/Murder_of_Seth_Rich

 

[fskimospy]

Well duh. He was being silly. Using 0.5-0.8% obviously wasn't supported. His judgment was clouded by trying to get people to focus on the other races instead of the presidency.

Then how isn't his model wrong if one of its primary parameters was way off in your estimation?

 

[Maxima1]

Then how isn't his model wrong if one of its primary parameters was way off in your estimation?

He shouldn't have picked that margin. He even says:

"As you can see, a less aggressive approach to estimating the home-stretch error would have given a Clinton win probability of 91-93%. That is about as low as the PEC approach could ever plausibly get."

His model is much better. It doesn't just turn on the dime, and Hillary still won the pop vote by 2%, while Nate's model assigns significant probabilities to Trump pop vote wins.

 

[fskimospy]

That's not false. I remember Nate Silver mentioning how the EC could produce a winner with a 2% pop vote deficit, but it was unlikely. Sorry, but I don't know why people give his model so much praise.

People give his model praise because it has been the most correct since he created it in 2008. Again, you can see from the link that I gave you that he considered a Trump PV/EV split to be about a 10% possibility.

This is why he was hedging because he knows not slanting too far either way will get people to say his model is correct no matter who wins. He was already humiliated completely by the Cubs and Trump prior to this and was on the ropes.

His forecast has always been a probabilistic one, which means that inherently you can't say from one result whether or not it's wrong (unless you are off as badly as Wang's prediction, I guess).
It's odd to say that the person with the best performing model was somehow putting his thumb on the scales. That makes no sense, or if he was it was a smart move that made his model more accurate.

If you wanted to look for evidence that his model is wrong I would look for his predictions of close races to be wrong more often than other models and/or wrong with a frequency in excess of what should be expected given the odds he gives.

 

[Maxima1]

People give his model praise because it has been the most correct since he created it in 2008. Again, you can see from the link that I gave you that he considered a Trump PV/EV split to be about a 10% possibility.

Almost all of that with very narrow margins. He assigned only about 1% to a 2% pop vote deficit. So how can you praise it so much? The 1/100 chance outcome is what happened.

 

[zinfamous]

You are an idiot. 538 predicted a Hillary win. They were wrong. If Hillary won, you would be saying 538 made a correct prediction and nobody would be arguing about it. They made an incorrect prediction, end of story.

you're still lying about this. why are you so fucking stupid?

 

[Victorian Gray]

you're still lying about this. why are you so fucking stupid?

Genetics?

 

[fskimospy]

He shouldn't have picked that margin. He even says:

"As you can see, a less aggressive approach to estimating the home-stretch error would have given a Clinton win probability of 91-93%. That is about as low as the PEC approach could ever plausibly get."

Yes, he said if he relaxed his assumptions further her odds would go down. He also spent several paragraphs stating why he thought that was wrong to do. Again, the margins he picks are integral parts of his model. You can't say his model was the best while saying he made decisions that made his model bad.

His model is much better. It doesn't just turn on the dime, and Hillary still won the pop vote by 2%, while Nate's model assigns significant probabilities to Trump pop vote wins.

No, his model was poor. He tuned it to give a >99% chance of something happening that did not happen. That's about as clear a mark of failure as you're ever going to get from a model for a single event. I think 538's model is too aggressive in incorporating new information, at least in the early to mid parts of the campaign season. I wouldn't be surprised if they modify that some for next time. Still, 538's vastly outperformed PEC and there's no getting around it.

 

[fskimospy]

Almost all of that with very narrow margins. He assigned only about 1% to a 2% pop vote deficit. So how can you praise it so much? The 1/100 chance outcome is what happened.

What output are you referring to, specifically?

 

[Maxima1]

Yes, he said if he relaxed his assumptions further her odds would go down. He also spent several paragraphs stating why he thought that was wrong to do. Again, the margins he picks are integral parts of his model. You can't say his model was the best while saying he made decisions that made his model bad.



No, his model was poor. He tuned it to give a >99% chance of something happening that did not happen. That's about as clear a mark of failure as you're ever going to get from a model for a single event. I think 538's model is too aggressive in incorporating new information, at least in the early to mid parts of the campaign season. I wouldn't be surprised if they modify that some for next time. Still, 538's vastly outperformed PEC and there's no getting around it.

He clearly made a bad judgment due to him wanting people to focus on other things.

As I said at the top, my motivation in doing these calculations is to help readers allocate their activism properly. Whether the Presidential win probability is 91% or 99%, it is basically settled. Therefore it is a more worthwhile proposition to work in Senate or House campaigns. Get on over to IN/MO/NC/NH/WI, or find a good House district using the District Finder tool in the left sidebar.

What output are you referring to, specifically?

I'll have to look. I remember reading an article from him talking about the biggest deficit plausible.

 

[fskimospy]

He clearly made a bad judgment due to him wanting people to focus on other things.

As I said at the top, my motivation in doing these calculations is to help readers allocate their activism properly. Whether the Presidential win probability is 91% or 99%, it is basically settled. Therefore it is a more worthwhile proposition to work in Senate or House campaigns. Get on over to IN/MO/NC/NH/WI, or find a good House district using the District Finder tool in the left sidebar.

I'll have to look. I remember reading an article from him talking about the biggest deficit plausible.

So we are now in agreement that his model was bad?

 

[Maxima1]

So we are now in agreement that his model was bad?

How does that suggest his model is bad? He picked the wrong margin. Good grief. He also updated it to 95%, so the >99% thing isn't even true. Looking at what I just bolded, he also didn't have a strong dismissal against a number suggesting a 10% chance of Donnie winning.

 

[Doc Savage Fan]

Remember when 538 had "serious, if not fatal flaws"? lol

http://www.huffingtonpost.com/entry/whats-wrong-with-538_us_581ffe18e4b0334571e09e74

https://projects.fivethirtyeight.com/2016-election-forecast/?ex_cid=rrpromo

 

[brycejones]

Genetics?

Paint chips were a food group for him while growing up.

 

[fskimospy]

How does that suggest his model is bad? He picked the wrong margin. Good grief. He also updated it to 95%, so the >99% thing isn't even true. Looking at what I just bolded, he also didn't have a strong dismissal against a number suggesting a 10% chance of Donnie winning.

It suggests his model was bad because the parameters he used in it were wrong. I mean that's how models work. That's like saying 'the engine was only bad because this one part was designed incompetently'. If that part is necessary for the engine to function, that's a bad engine. Maybe he can fix up his model to be better in the future, but as it stood for the last election it was very, very bad.

I am baffled as to why you are defending such an obvious failure.

 

[fskimospy]

Remember when 538 had "serious, if not fatal flaws"? lol

http://www.huffingtonpost.com/entry/whats-wrong-with-538_us_581ffe18e4b0334571e09e74

https://projects.fivethirtyeight.com/2016-election-forecast/?ex_cid=rrpromo

Why are you linking to an editorial by some random finance guy as if it has any meaning? I am consistently baffled that you think editorial pieces are meaningful contributions.

 

[Jhhnn]

I'm trying to figure out how election predictions have a damned thing to do with Seth Rich, other than to serve as diversion.

Meanwhile-

http://hotair.com/archives/2017/05/...connecting-hillary-clinton-seth-richs-murder/

Seth Rich's involvement in the release of DNC & Podesta emails is pure conspiracy theory.

Yeh, sure, it might be true... and Bigfoot might come out of the woods to give a press conference, too...

 

[Maxima1]

It suggests his model was bad because the parameters he used in it were wrong. I mean that's how models work. That's like saying 'the engine was only bad because this one part was designed incompetently'. If that part is necessary for the engine to function, that's a bad engine. Maybe he can fix up his model to be better in the future, but as it stood for the last election it was very, very bad.

I am baffled as to why you are defending such an obvious failure.

Quit being obtuse. The 0.5-0.8% margin wasn't supported. It shouldn't have been in there. Get it?

What's surprising is why you praise Nate's model so much. His model changed on the dime when it's obvious public opinion doesn't work like that, and assigning probabilities toward the middle hardly shows squat once there is a winner. He was also completely wrong on the Cubs and Trump prior, so quit acting like he's so great at predicting.

 

[fskimospy]

Quit being obtuse. The 0.5-0.8% margin wasn't supported. It shouldn't have been in there. Get it?

THEN HIS MODEL WAS BAD. WHY CAN'T YOU ADMIT THIS.

What's surprising is why you praise Nate's model so much. His model changed on the dime when it's obvious public opinion doesn't work like that, and assigning probabilities toward the middle hardly shows squat once there is a winner.

His model isn't attempting to model public opinion, it's attempting to say what the odds are of someone winning given the polling information available. The polls showed a huge amount of uncertainty, which is why his model showed a large amount of uncertainty. I agree that he should tamp down the aggressiveness in the middle of the campaign season, by the way.

Showing probabilities towards the middle isn't inherently right or wrong, with high uncertainty it's the right call. If you think that models are only right or valuable when they show high percentages then you're missing the point of forecasting models.

He was also completely wrong on the Cubs and Trump prior, so quit acting like he's so great at predicting.

The Cubs have zero to do with his election forecasting model and his primary prediction on Trump was in OPPOSITION to his model, not in concert with it. If anything, Trump's primary win was yet another victory for his model as it said Trump had a very real chance of winning long before conventional wisdom did.

Again, I don't care about 538's model except that it is very clearly the best one based on the results. If some other, better model comes along in the future I'll be fine with that one too. There's no point in clinging to failed models like Sam Wang's unless he is able to show convincing results in the future to make up for past failures.

 

[Maxima1]

THEN HIS MODEL WAS BAD. WHY CAN'T YOU ADMIT THIS.



His model isn't attempting to model public opinion, it's attempting to say what the odds are of someone winning given the polling information available. The polls showed a huge amount of uncertainty, which is why his model showed a large amount of uncertainty. I agree that he should tamp down the aggressiveness in the middle of the campaign season, by the way.

Showing probabilities towards the middle isn't inherently right or wrong, with high uncertainty it's the right call. If you think that models are only right or valuable when they show high percentages then you're missing the point of forecasting models.



The Cubs have zero to do with his election forecasting model and his primary prediction on Trump was in OPPOSITION to his model, not in concert with it. If anything, Trump's primary win was yet another victory for his model as it said Trump had a very real chance of winning long before conventional wisdom did.

Again, I don't care about 538's model except that it is very clearly the best one based on the results. If some other, better model comes along in the future I'll be fine with that one too. There's no point in clinging to failed models like Sam Wang's unless he is able to show convincing results in the future to make up for past failures.

Holy **** What's your reasoning to say Wang's model is bad even at 5%? How do we know the 5% didn't occur? After all, Trump only won by about 81k votes. Had the Comey letter not come, it's obvious that Wang's model would have been very high for Hillary win, but Nate's model would have still assigned asinine probabilities to Trump winning.

Silver assigned a very low probability to a EC win with a 2% deficit. His model is ****! His model changed drastically over trivial ***. His model assigned laughable probabilities to large Trump pop vote wins. Wang's model didn't do that with the orange buffoon which was supported by the outcome as he still lost by 2%

 

[fskimospy]

Holy **** What's your reasoning to say Wang's model is bad even at 5%? How do we know the 5% didn't occur? After all, Trump only won by about 81k votes. Had the Comey letter not come, it's obvious that Wang's model would have been very high for Hillary win, but Nate's model would have still assigned asinine probabilities to Trump winning.

That's just the thing though, had Clinton gotten 82,000 more votes Wang's model would have been 'right' in that it called the election correctly, but a result that close for something you gave a 95 or 99% probability to would strongly indicate that your model was bad. Results that close for something you are so certain about are usually signs you shouldn't be so certain.

You also seem to be awfully selective in the probabilities you consider small or large. Silver gave Trump a 10% chance of winning the EC while losing the PV, something you dismissed as crap. Now when Wang's model gives a 5% chance of something happening (at best, and a binary outcome no less) it's suddenly "well how do we know it DIDN'T?!". That's motivated reasoning.

Silver assigned a very low probability to a EC win with a 2% deficit. His model is ****! His model changed drastically over trivial ***. His model assigned laughable probabilities to large Trump pop vote wins. Wang's model didn't do that with the orange buffoon which was supported by the outcome as he still lost by 2%

Still waiting on a source for that, and his model changed based on polling inputs. I don't know what else to say other than you seem to be getting angry about this and digging in. Wang's model was bad and Silver's was considerably better. This much can easily be seen simply by comparing their predictions to the eventual results.

 

[Maxima1]

That's just the thing though, had Clinton gotten 82,000 more votes Wang's model would have been 'right' in that it called the election correctly, but a result that close for something you gave a 95 or 99% probability to would strongly indicate that your model was bad. Results that close for something you are so certain about are usually signs you shouldn't be so certain.

You also seem to be awfully selective in the probabilities you consider small or large. Silver gave Trump a 10% chance of winning the EC while losing the PV, something you dismissed as crap. Now when Wang's model gives a 5% chance of something happening (at best, and a binary outcome no less) it's suddenly "well how do we know it DIDN'T?!". That's motivated reasoning.

No, I'm partly saying this because that's what you've been doing. Pot meet kettle! Though, considering the unique outcome, it isn't that unreasonable to think his model isn't as bad as you think. I realize there were faults with Sam's, too, but he did in fact suggest an estimate upwards of 10% (on last day, he gave 7%, btw). Donnie was an unconventional candidate, so the assumptions with the pollsters were likely off with the uniqueness of this election. Because of that, a more conservative error margin would need to be chosen, since the pollsters are more likely to heavily favor one candidate.

Still waiting on a source for that, and his model changed based on polling inputs. I don't know what else to say other than you seem to be getting angry about this and digging in. Wang's model was bad and Silver's was considerably better. This much can easily be seen simply by comparing their predictions to the eventual results.

i can't find the specific article I was thinking of, but this should do.

https://fivethirtyeight.com/feature...ection-can-end-and-3-involve-clinton-winning/