|
Read the rules in the OP before posting, please.In order to ensure that this thread continues to meet TL standards and follows the proper guidelines, we will be enforcing the rules in the OP more strictly. Be sure to give them a re-read to refresh your memory! The vast majority of you are contributing in a healthy way, keep it up! NOTE: When providing a source, explain why you feel it is relevant and what purpose it adds to the discussion if it's not obvious. Also take note that unsubstantiated tweets/posts meant only to rekindle old arguments can result in a mod action. |
On October 29 2016 02:42 KwarK wrote:
Isn't the point of America that it is a commitment to American values, as defined in the constitution and upheld by American history, that defines what it is to be an American? Not the nation of your birth, the colour of your skin, the religion you practice or anything else.
I thought the point of citizenship was that is that it provides a very clear American vs !American.
Talking about what it means to be American means you've already accepted some cultural pissing match that is going to to be used to screw over some large groups of people.
|
The dream is coming true boys, I'm sure whatever Comey found was pretty fucking huge to reopen a case 11 days before the election. Holy shit. What if she wins and then Comey recommends charges
|
Cayman Islands24199 Posts
On October 29 2016 02:45 LegalLord wrote:Show nested quote +On October 29 2016 00:58 oneofthem wrote:On October 29 2016 00:51 LegalLord wrote:On October 29 2016 00:48 oneofthem wrote:On October 29 2016 00:31 Biff The Understudy wrote:On October 29 2016 00:25 LegalLord wrote:On October 29 2016 00:13 Biff The Understudy wrote:On October 29 2016 00:09 LegalLord wrote:On October 29 2016 00:05 Biff The Understudy wrote:On October 29 2016 00:00 IgnE wrote:
can we agree the election is over and talking about polls is boring? because both are true. Nate Silver at least doesn't agree on the first point. The race is not over. 18% chances is not great, but it's far from 0%. Although in this context, given that it's a one-time event, 18% just means a degree of belief (based on the 538 model) rather than a chance in the more commonly thought-of definition of probability (e.g. that Trump would win 18 out of 100 times). A technical difference, but an important one. I think 18% by 538 is the estimate of their 10 000 simulation of the race, based on the statistical degree of uncertainty of the polls. I don't think it's a degree of belief, but then again, I'm not a statistician. Maybe someone knows better how to interpret the whole thing? I can do that, but I'll be rather simplistic and skim over a lot of the details that I don't think too many people care about. Sorry to anyone who thinks I'm oversimplifying. So there are basically two major "schools" of thought in statistics/probability: the frequentists and the Bayesians. The frequentist interpretation of "probabilities" is pretty much what you intuitively think of as probabilities: the odds of getting the result when you repeat the experiment a lot. This is the classic, and more deeply rooted, interpretation of probability, but it has quite a lot of limits on what you can actually do with it. The Bayesian interpretation(s) are interpretations that try to very quickly make predictions based on a small amount of prior data, and often to predict the chances of events that can happen only once. This is the more "modern" branch of statistics and it's gained a lot of steam because it does work, even if it does have the problem of giving terrible results if your prior information isn't very good. But the interpretation of what probabilities themselves are is very different, and in this case the most appropriate definition is a Bayesian one: a state of knowledge, or more appropriately in this case, a degree of belief. And while those simulations are helpful, they are just that: simulations. The actual event only happens once. Nate Silver is very Bayesian in his analysis. No problem with that but the technical difference is important to note because it just tells us that it's not random chance whether or not Trump wins - it's just how confident, based on the 538 assumptions, the model is that Trump will win the election. That's quite fascinating, thanks. I had no idea  well he is also wrong if he wants to say that a frequentist model (purely poll based) would have trump closer. the bayesian priors in 538's structural model give trump a fighting chance, but it's basically just a guess informed by history and some political science. The point is that you can't really even be frequentist effectively since this is a one-time event, and that saying that "Trump has an 18% chance" isn't true in the sense that people think of probabilities - the interpretation of "what a probability is" is more Bayesian in nature, and of course according to whatever model 538 uses. In the real world, if the election were to be re-run 100 times (e.g. let's say you got 100 ballots at the polling station) then chances are the result would be the same each time. we don't know the true distribution so in that sense it's impossible to be frequentist, but the poll only model that the PEC uses is basically looking at historical poll behavior and using that to construct their model. it's not a structural model in the sense of modeling an idea of how you think the electorate responds to various external conditions. in looking at historical poll behavior, they get around the 'one time event' thing. the claim though is that this election is a pretty unique one, and i tend to buy that argument and look beyond the polls. Man, that PEC methodology seems extremely sketch after reading through it. It's hard to say whether they are just being opaque or overly simplistic about how they do their analysis. I would go with the latter. Besides being a lot more comprehensive, I like Nate Silver's underlying assumptions about which data is more valid a lot more than PEC's here. Though that is certainly a different topic than Bayesian vs frequentist approaches to the interpretation of what a probability actually is.
for data selection PEC only uses state polls. the other major feature of theirs is the use of poll based prior vs economic fundamental priors by other models. they are basically looking at how historical polls have converged to election day and adding an uncertainty to that, then doing some simulations.
http://election.princeton.edu/2016/08/03/why-is-the-pec-polls-only-forecast-so-stable/
|
United States44377 Posts
On October 29 2016 02:54 biology]major wrote:
The dream is coming true boys, I'm sure whatever Comey found was pretty fucking huge to reopen a case 11 days before the election. Holy shit. What if she wins and then coney recommends changes
Nothing is going to happen in the next 11 days. Hell, there are only 6 work days before the election, and voting has already begun in many areas.
|
On October 29 2016 02:54 biology]major wrote:
The dream is coming true boys, I'm sure whatever Comey found was pretty fucking huge to reopen a case 11 days before the election. Holy shit. What if she wins and then Comey recommends charges
The point of reopening investigations is that they haven't even looked yet, so there is nothing "found". Hence the word "investigation".
|
On October 29 2016 02:58 WolfintheSheep wrote:Show nested quote +On October 29 2016 02:54 biology]major wrote:
The dream is coming true boys, I'm sure whatever Comey found was pretty fucking huge to reopen a case 11 days before the election. Holy shit. What if she wins and then Comey recommends charges The point of reopening investigations is that they haven't even looked yet, so there is nothing "found". Hence the word "investigation".
Nah, there is no way he had the balls to open a case 11 days before a fucking election unless he had something.
|
On October 29 2016 02:55 oneofthem wrote:Show nested quote +On October 29 2016 02:45 LegalLord wrote:On October 29 2016 00:58 oneofthem wrote:On October 29 2016 00:51 LegalLord wrote:On October 29 2016 00:48 oneofthem wrote:On October 29 2016 00:31 Biff The Understudy wrote:On October 29 2016 00:25 LegalLord wrote:On October 29 2016 00:13 Biff The Understudy wrote:On October 29 2016 00:09 LegalLord wrote:On October 29 2016 00:05 Biff The Understudy wrote:
[quote]
Nate Silver at least doesn't agree on the first point. The race is not over. 18% chances is not great, but it's far from 0%. Although in this context, given that it's a one-time event, 18% just means a degree of belief (based on the 538 model) rather than a chance in the more commonly thought-of definition of probability (e.g. that Trump would win 18 out of 100 times). A technical difference, but an important one. I think 18% by 538 is the estimate of their 10 000 simulation of the race, based on the statistical degree of uncertainty of the polls. I don't think it's a degree of belief, but then again, I'm not a statistician. Maybe someone knows better how to interpret the whole thing? I can do that, but I'll be rather simplistic and skim over a lot of the details that I don't think too many people care about. Sorry to anyone who thinks I'm oversimplifying. So there are basically two major "schools" of thought in statistics/probability: the frequentists and the Bayesians. The frequentist interpretation of "probabilities" is pretty much what you intuitively think of as probabilities: the odds of getting the result when you repeat the experiment a lot. This is the classic, and more deeply rooted, interpretation of probability, but it has quite a lot of limits on what you can actually do with it. The Bayesian interpretation(s) are interpretations that try to very quickly make predictions based on a small amount of prior data, and often to predict the chances of events that can happen only once. This is the more "modern" branch of statistics and it's gained a lot of steam because it does work, even if it does have the problem of giving terrible results if your prior information isn't very good. But the interpretation of what probabilities themselves are is very different, and in this case the most appropriate definition is a Bayesian one: a state of knowledge, or more appropriately in this case, a degree of belief. And while those simulations are helpful, they are just that: simulations. The actual event only happens once. Nate Silver is very Bayesian in his analysis. No problem with that but the technical difference is important to note because it just tells us that it's not random chance whether or not Trump wins - it's just how confident, based on the 538 assumptions, the model is that Trump will win the election. That's quite fascinating, thanks. I had no idea well he is also wrong if he wants to say that a frequentist model (purely poll based) would have trump closer. the bayesian priors in 538's structural model give trump a fighting chance, but it's basically just a guess informed by history and some political science. The point is that you can't really even be frequentist effectively since this is a one-time event, and that saying that "Trump has an 18% chance" isn't true in the sense that people think of probabilities - the interpretation of "what a probability is" is more Bayesian in nature, and of course according to whatever model 538 uses. In the real world, if the election were to be re-run 100 times (e.g. let's say you got 100 ballots at the polling station) then chances are the result would be the same each time. we don't know the true distribution so in that sense it's impossible to be frequentist, but the poll only model that the PEC uses is basically looking at historical poll behavior and using that to construct their model. it's not a structural model in the sense of modeling an idea of how you think the electorate responds to various external conditions. in looking at historical poll behavior, they get around the 'one time event' thing. the claim though is that this election is a pretty unique one, and i tend to buy that argument and look beyond the polls. Man, that PEC methodology seems extremely sketch after reading through it. It's hard to say whether they are just being opaque or overly simplistic about how they do their analysis. I would go with the latter. Besides being a lot more comprehensive, I like Nate Silver's underlying assumptions about which data is more valid a lot more than PEC's here. Though that is certainly a different topic than Bayesian vs frequentist approaches to the interpretation of what a probability actually is. for data selection PEC only uses state polls. the other major feature of theirs is the use of poll based prior vs economic fundamental priors by other models. they are basically looking at how historical polls have converged to election day and adding an uncertainty to that, then doing some simulations. http://election.princeton.edu/2016/08/03/why-is-the-pec-polls-only-forecast-so-stable/
Their entire shtick is their focus on meta-analysis (lol) and specifically, their idea of the meta margin. I really don't like how they just hide behind "we have these statistical methods to deal with problems" and really just abstract away the issues related to accuracy in polling in a way that doesn't really make sense. They put much less thought into that issue than 538 does, and that is strongly to their detriment.
They talk a lot about how stable their model is, but honestly this election is pretty far from stable. There were times when it genuinely seemed that Trump was likely to win.
|
Sovereign immunity would apply once in office, (pending impeachment of course); not sure how that works with a president-elect though.
I think comey simply found new evidence worth looking at, so he is.
|
if she wins but then gets arrested, looks like we'll be swearing in tim kaine, who is still an infinitely more acceptable president than trump. way more centrist than i'm comfortable with but whatever
|
Kaine is like an even less charismatic clone of Hillary. I am pretty much indifferent between which of the two is president.
|
Cayman Islands24199 Posts
On October 29 2016 03:01 LegalLord wrote:Show nested quote +On October 29 2016 02:55 oneofthem wrote:On October 29 2016 02:45 LegalLord wrote:On October 29 2016 00:58 oneofthem wrote:On October 29 2016 00:51 LegalLord wrote:On October 29 2016 00:48 oneofthem wrote:On October 29 2016 00:31 Biff The Understudy wrote:On October 29 2016 00:25 LegalLord wrote:On October 29 2016 00:13 Biff The Understudy wrote:On October 29 2016 00:09 LegalLord wrote:
[quote]
Although in this context, given that it's a one-time event, 18% just means a degree of belief (based on the 538 model) rather than a chance in the more commonly thought-of definition of probability (e.g. that Trump would win 18 out of 100 times). A technical difference, but an important one. I think 18% by 538 is the estimate of their 10 000 simulation of the race, based on the statistical degree of uncertainty of the polls. I don't think it's a degree of belief, but then again, I'm not a statistician. Maybe someone knows better how to interpret the whole thing? I can do that, but I'll be rather simplistic and skim over a lot of the details that I don't think too many people care about. Sorry to anyone who thinks I'm oversimplifying. So there are basically two major "schools" of thought in statistics/probability: the frequentists and the Bayesians. The frequentist interpretation of "probabilities" is pretty much what you intuitively think of as probabilities: the odds of getting the result when you repeat the experiment a lot. This is the classic, and more deeply rooted, interpretation of probability, but it has quite a lot of limits on what you can actually do with it. The Bayesian interpretation(s) are interpretations that try to very quickly make predictions based on a small amount of prior data, and often to predict the chances of events that can happen only once. This is the more "modern" branch of statistics and it's gained a lot of steam because it does work, even if it does have the problem of giving terrible results if your prior information isn't very good. But the interpretation of what probabilities themselves are is very different, and in this case the most appropriate definition is a Bayesian one: a state of knowledge, or more appropriately in this case, a degree of belief. And while those simulations are helpful, they are just that: simulations. The actual event only happens once. Nate Silver is very Bayesian in his analysis. No problem with that but the technical difference is important to note because it just tells us that it's not random chance whether or not Trump wins - it's just how confident, based on the 538 assumptions, the model is that Trump will win the election. That's quite fascinating, thanks. I had no idea well he is also wrong if he wants to say that a frequentist model (purely poll based) would have trump closer. the bayesian priors in 538's structural model give trump a fighting chance, but it's basically just a guess informed by history and some political science. The point is that you can't really even be frequentist effectively since this is a one-time event, and that saying that "Trump has an 18% chance" isn't true in the sense that people think of probabilities - the interpretation of "what a probability is" is more Bayesian in nature, and of course according to whatever model 538 uses. In the real world, if the election were to be re-run 100 times (e.g. let's say you got 100 ballots at the polling station) then chances are the result would be the same each time. we don't know the true distribution so in that sense it's impossible to be frequentist, but the poll only model that the PEC uses is basically looking at historical poll behavior and using that to construct their model. it's not a structural model in the sense of modeling an idea of how you think the electorate responds to various external conditions. in looking at historical poll behavior, they get around the 'one time event' thing. the claim though is that this election is a pretty unique one, and i tend to buy that argument and look beyond the polls. Man, that PEC methodology seems extremely sketch after reading through it. It's hard to say whether they are just being opaque or overly simplistic about how they do their analysis. I would go with the latter. Besides being a lot more comprehensive, I like Nate Silver's underlying assumptions about which data is more valid a lot more than PEC's here. Though that is certainly a different topic than Bayesian vs frequentist approaches to the interpretation of what a probability actually is. for data selection PEC only uses state polls. the other major feature of theirs is the use of poll based prior vs economic fundamental priors by other models. they are basically looking at how historical polls have converged to election day and adding an uncertainty to that, then doing some simulations. http://election.princeton.edu/2016/08/03/why-is-the-pec-polls-only-forecast-so-stable/ Their entire shtick is their focus on meta-analysis (lol) and specifically, their idea of the meta margin. I really don't like how they just hide behind "we have these statistical methods to deal with problems" and really just abstract away the issues related to accuracy in polling in a way that doesn't really make sense. They put much less thought into that issue than 538 does, and that is strongly to their detriment. They talk a lot about how stable their model is, but honestly this election is pretty far from stable. There were times when it genuinely seemed that Trump was likely to win.
well they have this arbitrary uncertainty parameter that they adjusted to very high for this year. it's just a reflection of the current polling margin that trump is a very bad underdog. it could be possible that the polls are very wrong, or that the race is very volatile (like i see it), and their approach gets upturned.
if trump does do significantly better than what the poll aggregates show, it would not be because 538's model is smart. they have a bunch of economic indicators in there that don't capture the source of the current year volatility.
|
Cayman Islands24199 Posts
On October 29 2016 02:59 biology]major wrote:Show nested quote +On October 29 2016 02:58 WolfintheSheep wrote:On October 29 2016 02:54 biology]major wrote:
The dream is coming true boys, I'm sure whatever Comey found was pretty fucking huge to reopen a case 11 days before the election. Holy shit. What if she wins and then Comey recommends charges The point of reopening investigations is that they haven't even looked yet, so there is nothing "found". Hence the word "investigation". Nah, there is no way he had the balls to open a case 11 days before a fucking election unless he had something.
it's just comey doing his job. they haven't even looked at the emails yet.
|
On October 29 2016 02:59 biology]major wrote:Show nested quote +On October 29 2016 02:58 WolfintheSheep wrote:On October 29 2016 02:54 biology]major wrote:
The dream is coming true boys, I'm sure whatever Comey found was pretty fucking huge to reopen a case 11 days before the election. Holy shit. What if she wins and then Comey recommends charges The point of reopening investigations is that they haven't even looked yet, so there is nothing "found". Hence the word "investigation". Nah, there is no way he had the balls to open a case 11 days before a fucking election unless he had something.
If he didn't have the "balls" to follow thru on his responsibilities then he shouldn't be head of the FBI.
There is ZERO reason not to take him at his word which is that he saw something in an unrelated case that meant they had to reopen it for an unknown length of time. She sent like 100k emails so that seems like a thing that can happen.
|
On October 29 2016 02:59 biology]major wrote:Show nested quote +On October 29 2016 02:58 WolfintheSheep wrote:On October 29 2016 02:54 biology]major wrote:
The dream is coming true boys, I'm sure whatever Comey found was pretty fucking huge to reopen a case 11 days before the election. Holy shit. What if she wins and then Comey recommends charges The point of reopening investigations is that they haven't even looked yet, so there is nothing "found". Hence the word "investigation". Nah, there is no way he had the balls to open a case 11 days before a fucking election unless he had something.
There's no "balls" involved here and there never were any. He just does his job. He is following protocol and doing his job. This could hurt Clinton if something bad is proven, but this could also be a big bump for her if Comey declares her innocence. Comey declaring her innocence would also disincentive republicans from pursuing this as a criminal investigation.
|
On October 29 2016 03:10 Mohdoo wrote:Show nested quote +On October 29 2016 02:59 biology]major wrote:On October 29 2016 02:58 WolfintheSheep wrote:On October 29 2016 02:54 biology]major wrote:
The dream is coming true boys, I'm sure whatever Comey found was pretty fucking huge to reopen a case 11 days before the election. Holy shit. What if she wins and then Comey recommends charges The point of reopening investigations is that they haven't even looked yet, so there is nothing "found". Hence the word "investigation". Nah, there is no way he had the balls to open a case 11 days before a fucking election unless he had something. There's no "balls" involved here and there never were any. He just does his job. He is following protocol and doing his job. This could hurt Clinton if something bad is proven, but this could also be a big bump for her if Comey declares her innocence. Comey declaring her innocence would also disincentive republicans from pursuing this as a criminal investigation.
That is naive, it's not him just doing his job, it's him throwing a nuke into the political process, and I don't think he would do such a thing unless they really had something.
Whatever though, keep believing!
Edit: this case isn't going to be resolved in 11 days. So this will only have a political effect until election.
|
United States44377 Posts
Throwing a nuke into the political process after all the debates?
|
headlines: THE FBI HAS FOUND NEW EMAILS AND MUST REVIEW THEM AS PART OF ITS INVESTIGATION
what probably happened: some analyst had a stack of emails to review printed out and left them in the print room. some guy was cleaning out the print room and noticed that they were left there, so they gotta read them now.
|
On October 29 2016 03:14 ticklishmusic wrote:
headlines: THE FBI HAS FOUND NEW EMAILS AND MUST REVIEW THEM AS PART OF ITS INVESTIGATION
what probably happened: some analyst had a stack of emails to review printed out and left them in the print room. some guy was cleaning out the print room and noticed that they were left there, so they gotta read them now.
Holy shit if this is really what you think, the level of delusion is out of this world.
|
On October 29 2016 03:16 biology]major wrote:Show nested quote +On October 29 2016 03:14 ticklishmusic wrote:
headlines: THE FBI HAS FOUND NEW EMAILS AND MUST REVIEW THEM AS PART OF ITS INVESTIGATION
what probably happened: some analyst had a stack of emails to review printed out and left them in the print room. some guy was cleaning out the print room and noticed that they were left there, so they gotta read them now. Holy shit if this is really what you think, the level of delusion is out of this world.
i'm not going to get into an argument about if my reality or your reality is actual reality because it'd be pretty futile and at worst end in an existential crisis for one of us.
|
United States44377 Posts
Bio, out of curiousity what, for you, is the last date at which someone could throw a nuke into the political process? Day of the election? Day before?
|
|
|
|
|
|
EPT
