Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Lord Shonus]

That has more to do with the fact that the Republicans had a ton of weak candidates and Trump (who dominated the field as much because he was so solidly there) while the Democrats had two fairly strong candidates. If the Republicans had only had one or two "not-Trump" candidates, Trump might still have won the primary, but the margin wouldn't have been anywhere near as solid.

[origamiscienceguy]

I don't think bernie sanders could have won the general election against trump. He would pull in lots of millenials (which were voting democrat anyways) But push away the general working class (which won trump the election.)

I think we know by now that polling when trump is involved is always wrong...

[hector13]

Polls have been wrong for a lot of things recently.

Not sure Bernie would've lost the working class vote, considering he was championing workers' rights for his entire career.

His message was also a bit more positive as opposed to Trumps "fucking immigrants" stance.

Edit: then again, I thought Republicans at large would reject the hatred of Trump's message despite his vague promises to bring jobs back and grow the economy to a ridiculous extent, so what do a I know.

[Reelya]

I also think a lot of libertarian/independent types might have jumped for Sanders rather than Trump or Clinton. There are a lot of people who just want a massive shake up of the establishment, probably on the feeling that a fresh slate can't be any worse than the humongous mess of bureaucracy and special interests that exists.

[WealthyRadish]

I don't think bernie sanders could have won the general election against trump. He would pull in lots of millenials (which were voting democrat anyways) But push away the general working class (which won trump the election.)

I think we know by now that polling when trump is involved is always wrong...

All three of these statements are wrong.

1) Clinton didn't get the millennial vote
2) This "silent working class" vote element is greatly overstated
3) The polls predicting a high likelihood of someone winning and then having that not happen does not always mean they were wrong. What happened was very unlikely, and it happened by an extremely thin margin, and the fact that it did happen does not change how unlikely it was.

[wierd]

Why trump is popular, and why his major demographic is "working adults without a degree"

1) This demographic gets paid the least, because the jobs are shit, or deal with shit. (literal shit. The kind you flush.) They come home exhausted, and the last thing they really want to do is go "self-educate" about some pontificating politician who claims to have their back, when really, nothing really changes for them, because shit jobs are essential to the functioning of society. SOMEBODY has to make sure grandma's ass is clean at the nursing home. SOMEBODY has to assure that the trash is taken from the dumpster every week. Etc. For some perverse reason, these people, who's jobs are the most essential to the functioning of society, are always the least paid, assuring that people in this demographic are not able to rise above it without gargantuan efforts. No matter how much Clinton, or Sanders, or any other champion of the left claims to have their backs, these people cannot stop doing their jobs, because the well educated will never take those jobs. Never. One day they might be replaced with robots, but on that day, these people will just be out of work permanently, and society will ignore them.

2) Since these people are drained by their jobs every day, they lack the energy needed to self-educate about the shyster falsehoods of Trump and Co. Keeping up on the endless lies, the lies about lies, the almost truths, and the past histories and voting records of just more people who are so distant from the kind of life that these people lead that they might as well be living in ivory towers by comparison, is just too much energy for very little gain.

3) Trump and Co offers a simple, easy to accept, superficial message that is appealing, given the above preconditions.


Nevermind that Sander's Euro-socialist platform would have provided far more security for this demographic than any of the other candidates this cycle, (given that he would have tried to get lasting social support networks to transition such working classes out of those essential but highly intensive and undesirable vocations as automation eventually replaces them, rather than throwing them to the wolves), or that of the candidates, he was among the least disingenuous with his statements. His policies were complicated (because policy and politics are complicated), so his message was complicated.  That made him undesirable.

When people have just enough mental or physical energy to barely pop the top of a can of beer, and pass out in the recliner, maybe change the channel a few times before hitting a state of unconciousness-- they dont have the energy to be properly political.

If you dont like the consequences of this kind of society, then those that have the time and energy to properly think about it, should take the time to properly think about and consider why people are this way, and endeavor to make it so such people can actually have the needed resources to also be properly political.

The problem, is that these essential services would become vastly more expensive if that happened, because if you reduce those people's hours, you have to increase their pay, because they are already the bare bottom earners (and barely survive financially), AND are overworked. If you cut hours, you have to hire more of them to cover the workload-- Together, they handily increase the base operating costs of supplying those services.  That is EXACTLY THE OPPOSITE of what the industry wants to do, and what society wants from those services. (Industry does not want to pay ANYONE, ANYTHING-- and society wants the services for as close to free as is possible.)  That is why automation is so desirable in these industries. Currently, it takes veritable armies of these people to keep society running. People that society wants to pretend do not exist. People that the top tiers treat as disposable and non-people. Those people have a good chunk of the vote.

A dangerous populist got their attention.

Will he keep it? Probably not. He's exactly the opposite of what they need. However, he regurgitates rhetoric that is simplistic enough for their overtaxed abilities to comfortably ingest and react to.

Overall, these people have a very negative view of the world (wouldnt you, if your day was spent cleaning up other people's filth and being paid slave wages-- day in, and day out?) and so, negative rhetoric resonates with them more than progressive promises, which they view as just more pablum and lies, again, because society cannot actually afford to change their plight.

It has very little to do with "hate", or "misogyny" or any of those topics that are convenient for the middle to attribute it to, and for the top tiers to distract with.  It has everything to do with the fact that these people are ignorant, lack the resources to stop being ignorant, they are angry, they are tired, and they have had it up to "here" with promises that society has no interest in upholding. More than anything, they voted for trump because 1) it was convenient, 2) it as something other than the status quo we have had for the past 20 years that has seen this demographic go from near-middle-class to severe poverty by failure to regulate industry, and being whores to international commerce, and 3) they are at the breaking point, and are desperate. Even a shyster with a fake spray on tan, and a horrible hairpiece, that is an obvious shyster, seems better to them than very capable, intelligent, and conniving foxes that have been running the henhouse.

If society wants to replace them with robots, fucking do it already. Stop stringing them along. -- That's about the mindset they have now.


You can add to that, the presence of aloof millennials who felt comfortable, and thus did not leave home to vote, and the slobbering morons that will vote for anything that supports more guns, and a rose-colored version of the 50s (complete with racism), and then you get things like trump as president.

[Reelya]

Yeah, that's the thing about polls. If polls say "candidate X has a 70% chance of winning", and that's about as accurate as polls ever get, then the polls are going to be right 70% of the time. 30% of the time the other guy will win. That's already accounted for. It in no way indicates the polling was "wrong". The limits of the polling method were already accounted for in the very phrase "70% chance".

It's the same as weather prediction. If each day in a 10-day time period you say "10% chance of rain today", and it doesn't rain for 9 days, but then on the 10th day it rains, people are going to lose their shit and accuse weather predictors of "getting it wrong". People are wrong when it comes to understanding basic probability statements, not the people working out the odds. People convert slight variations in odds to personal yes/no prediction of whether something will happen, then go with that belief. But then they get angry when their own prediction is wrong a predictable amount of the time.

[wierd]

I dunno. That assumes the model is accurate.

The weatherman has pretty reliable models to work with, which are based on impartial real-world physics. The uncertainty just comes from the complexity of the system.

With politics, it is not so.  We dont even have a valid model of human behavior, how the hell can there be reliable predictions about high level outcomes like this?

Instead, the polls hedge this with weasel words like "Based on exit poll reports", and the like.  Never mind the notorious inaccuracy of self-reporting.

[Reelya]

But politics has repeated trials that you can form models about. There are town, council, state, federal elections, and multiple types of bodies at these levels, at regular intervals, and you can data mine elections going back over a century and create machine learning models that can predict particular years by training them on data from different years. It's has the required Big Data component that makes fairly accurate models possible that don't rely on domain-specific reasoning which could be faulty.

To do data analysis you don't need "valid models of human behavior" at all. Because big data machine leaning is an aggregate situation where you're looking at covariance in variables. Parameters that don't have a high correlation with what you're trying to predict get eliminated from the data set automatically, so you can just pile as much data in as you can get, and the black box works out which ones are good predictors or not. And you always leave random years out of the training data, so you can use them as a validation set: does this model predict years it's never seen before from the data for other years.

[wierd]

Big data is not magical. It relies on inferences, and not proper causalities.

"Group X tends to do Y", but does not consider any causative factors of why this is so.  If that causative factor changes, the model that makes use of that outcome will fail.

That is exactly what seems to have happened here. "These states tend to vote Democrat consistently", without contemplating the whys and what fors.  When those whys and whatfors did not get presented, the expected result did not happen.

[Reelya]

But the thing is, big data doesn't really rely on inferences. Any input data which has a low covariance with the variable you're trying to predict gets automatically eliminated. So you just throw everything in there and see what's left after the irrelevant crud is rejected. That's the beauty of the method. A person doesn't need to decide "this is relevant" at all, just "oh we got some more data, maybe this stuff will be useful, throw it in". It's the data mining engine that tells you "sorry but that data is junk, it doesn't improve the prediction".

Causality only matters if your goal is explicitly to prove a causal link between one thing and another. If the goal is purely to predict a single output variable, correlation is all you care about. Sure, correlations can be coincidence, but if two variables have been highly correlated for the last 1000 days, then it's a good bet they'll be highly correlated tomorrow, meaning it's useful to measure one to get a better estimate of the other. Perhaps both variables (the one you're predicting and your data point) are actually correlated with a third element. Even better, just collect that data and throw that into the model, too. Having two correlates for the desired prediction is only going to be more helpful than having one (since your inference engine can automatically calculate the cross-correlation between them, making them more accurate than they would be alone).

As long as the thing you're trying to predict is something objectively measurable (such as the percentage of votes one candidate will get), then having more data of any type is going to give a more reasonable estimate. It's certainly is a lot like weather prediction actually, since somewhere some data would be a better predictor, but under-sampling is the main problem. If you literally asked 100% of people "who are you going to vote for" the day before the election, I bet you that data (along with other data you know about the people) would give you a very strong prediction. So it's not that different. The problem with both polling and weather prediction is under-sampling. i.e. how to extrapolate from partial data, which is where modeling comes in. It's not a case of "humans are unknowably complex" here. The vast majority of people have probably made their minds up well in advance of the election date whether their going to go for the binary Option A or Option B. It's just not feasible to ask everyone.

Personally, I think that data mining social media usage patterns might end up being as useful as polling. The issue would be that there are very few years of sampling data that you could collect.

[Gentlefish]

This debate from 2012 is scarily relevant.

[origamiscienceguy]

um, I think she did get the millenial vote:

Spoiler (click to show/hide)

[Reelya]

um, I think she did get the millenial vote:

Percentage of millennial vote is a separate issue from voter turnout for that age-group:

Spoiler (click to show/hide)

As you can see, while she "won" the total millennial vote, youth votes declined for Democrats but not Republicans.

[origamiscienceguy]

Trump is mad at chicagos mayor