2016-04-04

"Tech startups love millennials. Tasty, tasty millennials who get underpaid, overworked, churned up and turned into nourishment for venture capitalists. Millennials are the Soylent Green of the tech world." Excellent New York Post piece on startups and work culture.

"Tech startups love millennials. Tasty, tasty millennials who get underpaid, overworked, churned up and turned into nourishment for venture capitalists. Millennials are the Soylent Green of the tech world." Excellent New York Post piece on startups and work culture.

A great read if you have 5 mins.

http://nypost.com/2016/04/03/millennials-are-being-dot-conned-by-cult-like-tech-companies/

How your newspaper and TV news are brigading and downvoting Trump, and what you can do about it. (Hint: it’s the crooked polls.)

How your newspaper and TV news are brigading and downvoting Trump, and what you can do about it. (Hint: it’s the crooked polls.)

We’ve all seen the polls showing Donald Trump’s large Wisconsin lead melt away in recent weeks, only to rebound at the last minute - but what we’ve been looking at is a house of cards. A house of cards so liable to fall apart, you’ll be amazed that it hasn’t already. By the time you’ve read this, I hope you’ll understand it a little more - and that you’ll be as angry about it as I am. (I hope you won’t be angry about any typos, though: I’m typing fast, I have a plane to catch in a while, and I want to get this post out onto /sub/the_donald in time for the Wisconsin Primary today. I’m typing real fast.)

How do I know that polls are routinely faked? Well, there’s the fact that the latest polls showing Trump ‘doing badly’ in Wisconsin's Republican primary, like the March 24-28 Marquette poll, polled more more than twice as many Democrats and Independents, as Republicans! (If that doesn’t intrigue you, nothing will. And there’s a lot, lot more going on with the polls, besides.)

I’m going to put my cards on the table: I work in the media. I always have. Understanding the public, and persuading the public, is the only thing I do. Theres a good chance you have things in your bathroom cabinet, or your kitchen, that I persuaded you to buy. (Well, if you saw the ads, or heard the commercials.) And over my career as a public persuader, I’ve worked with many public polling companies. I’ve been the ‘public explainer’ of major surveys, writing about the results for national audiences, governments, scientists, and businesses. I don’t work on polls any more, just commercials and ads - and the truth is, that’s because I can sell you soap or pants without lying to you. The average TV commercial or print ad is honest to a fault, in comparison to the ‘impartial’ and ‘scientific’ polls you trust in your newspaper. I started out with good intentions - trying to run good surveys, and explain them honestly. But I spent years in the same room with the politicians, government employees, lobbyists, pollsters and press as the surveys were ordered, faked, and published to unsuspecting readers. I got tired of pushing for honesty, and tired of being ignored.

After a few years, you see that most polls, including political polls, that wind up on your TV news and in your newspaper… well, they’re bullsh-t. Sure, some polling companies are honest, and some private pollsters are honest - but most of the surveys that you see in the mainstream media are baked to order, to pull the wool over your eyes. And, just like the pizza delivery guy, the polling company only gets paid for delivering the correct order. Polls, surveys, statistics… they’re faked in order to justify decisions that have already been taken, and in order to fool the public. (You didn’t pay for that poll. It wasn’t run for your benefit.)

The fake data tells you that, gee, you’re in the minority when you disagree with the government’s latest, greatest idea - so STFU, citizen!

The fake data tells you that unemployment is 5%, when it’s really 25% - because unemployed people don’t riot at 5% unemployment, they just blame themselves.

.

Want more examples of fake data? Let’s get to politics. Let’s get to Trump

Trump’s support is higher than the media shows in polls. One of the reasons it looks lower in the polls is that certain media organisations collaborate with certain polling firms, to release skewed polls immediately before a state primary - maybe a week or two before. Take Trump’s months-long lead in Wisconsin: suddenly it’s dropped! He’s losing support to that nice Mr. Cruz! His TV gaffes have hurt him! …Only, that’s bullsh-t. It’s a media psy-op.

The same ‘familiar suspect’ newspapers and TV channels tend to run the skewed polls. (I won’t mention the newspapers or TV channels by name because I don’t want to be sued - and anyway, the same 6 companies own 90% of the US media, so there are few saints.) These skewed polls dramatically pull down the moving average of polls, making a winner suddenly look like a loser. The picture usually goes like this: Trump was ahead by a lot, now the polls show that he’s behind - but really, he isn’t. Take out the fake poll, and the average is still good. (BTW, not every poll from each of these newspapers is a crummy poll - but the ‘right’ ones are. The ones that need to be, to affect the primary race.)

In the case of Wisconsin, Trump was barnstorming the state, miles ahead of everyone, from September 2015 up to and including the end of February 2016. Cruz was 10 points behind, 15 points behind, and so on. (http://www.realclearpolitics.com/epolls/2016/president/wi/wisconsin_republican_presidential_prima...). So, what happens? The polling organisations release a massive burst of polls in a 2-week window from March 20th, suddenly showing Cruz surging to victory out of nowhere. Just nowhere. And all of that, in the middle of his affairs scandal. Right.

.

It’s more than an establishment newspaper ‘downvoting’ Trump by ‘brigading’. It’s poisonous in lots of ways:

• Partly, the idea is to make you think Trump is a loser. Want the next best thing to stop Hillary? You ‘have’ to go with Cruz.

• Partly, the idea is to make you think attacks on Cruz aren’t working. That no-one cares about his ‘little cruz missile’. (It’s like the fairy tale of the King’s New Clothes: the first person to yell “he’s butt-naked!” pops the fantasy bubble, but the polls are trying to persuade you that Cruz isn’t butt naked… Yeah, sorry for that mental image.)

• Partly, the faked polls help to 'sell' the inevitable vote rigging. If you buy the faked poll data, you’ll buy the rigged vote. They’re training you for disappointment: “Dang it, the polls showed Trump was sliding, and the votes confirmed it! F-ck you Wisconsin!”

.

The Big Steal. The Big Dip. The Unexpected Rise.

This same pattern of faked polling data has been seen time and time again. It works best when no-one’s been conducting polls in a state for a while, so there’s no baseline data, no trend line, that would reveal a suspicious and sudden poll collapse.

The fake polling strategy doesn’t work so well when lots of independent and honest organizations run polls: which is why one or two weeks of faked polls against Trump can suddenly be thrown out by an honest, scientific poll. (In the case of Wisconsin, the latest poll from ARG.) That’s why Trump’s numbers have sometimes seemed strangely low in the week or two before a primary, then an honest company (usually an 'outsider') comes out with a decent-quality poll the day before the vote, and shows that Trump’s figures never fell at all. The honest polling company’s data gets called inaccurate, in our controlled media. The accurate polling data is explained away by presstitutes as ‘an outlier’. Or, they say it shows ‘volatility’ or some other pseudo-scientific phrase, to make it seem like the voters’ support oscillates like a pendulum over a matter of hours. (It does not.)

Sometimes, the strategy would fall apart completely, if the media widely reported all of the polls by all of the pollsters. People would notice that one bunch of pollsters were accurate a lot, and another bunch were inaccurate a lot.... So, the media is sometimes very selective in the polls it chooses to show you, and the polls it chooses to include in moving poll averages. (Too many honest pollsters reporting on their findings this week? Just ignore them completely. The public will never know.)

.

Which polls in that Real Clear Politics table of 20+ polls do I trust, as someone who has run polls? (http://www.realclearpolitics.com/epolls/2016/president/wi/wisconsin_republican_presidential_primary-3763.html)

The Optimus poll on 22-24 March, for one. It surveyed more than six thousand people, and it got a representative sample of the Wisconsin population. Unsurprisingly, the Optimus poll put Trump ahead by 4 points - low, but not bad, considering the Fields-Lewandowski ‘scandal’ had been boiling for a week, flinging sh-t Trump’s way all that time. With the Michelle Fields BS dying down as people realized she’s a lying cook, Trump’s numbers should have climbed back up towards the earlier high (especially with Cruz being outed for his affairs) but the subsequent polls by other pollsters show an astonishing surge in Cruz’s support, that isn't reflected by attendance at Cruz rallies.

The 24 March - 3 April ARG poll is also good, showing Trump leading Cruz by 10%. Not bad, at the tail end of the Lewandowski-Fields BS. ARG has a habit of being accurate, for one thing. ARG surveyed likely Republican voters, and likely Dem/Indy voters in the Repub Primary, in representative age ranges and demographic splits including male-female. Unlike other polling firms, ARG even calls people’s cell phones, not just landlines. So, even with a smallish sample, an ARG survey is usually an accurate survey. (http://americanresearchgroup.com/pres2016/primary/rep/wirep.html)

Which of the Real Clear Politics polls would I avoid? Almost all of the other polls from March to the present. Take the Marquette University Law School poll of March 24-28 (https://law.marquette.edu/poll/wp-content/uploads/2016/03/MLSP34Toplines.pdf). It's not the worst of the worst by any means - I'm selecting it for illustrative purposes. The Marquette poll tries to tell us that Trump will lose the Wisconsin Republican primary, by selecting a survey group that is 68.6% Democrat and Independents (question F1, Page 24, Marquette surveyed people who described themselves as: Republican (27.9%), Democrat (32.1%), Independent (36.5%). Marquette’s data is pure sh-t, if you want to use it to predict the Republican primary winner in Wisconsin, and Marquette's pollsters damn well ought to know this. If they don't, they're in the wrong business IMHO. Draw your own conclusions.

.

How do they fake the polls? There are lots of ways polls are faked, in addition to the brief explanations I’ve given for Wisconsin in particular. The following are less than 10% of the options. (I won’t even go into the many ways you can use statistical sampling, sorting and graphing techniques to mislead people. Math is a deceitful mistress, America.)

• Sometimes, they just make it up. Nice and simple.

• Sometimes, they use the ‘margin of error’ to take votes from one guy, and give them to another. On a poll with a 5% margin of error, you can take 5% from Trump and give it to Kasich or ‘Undecided’ - AND you can give 5% to Cruz, on top. So, if the real poll data was Trump 35%, Cruz 30%, Kasich 20%, Undecided 15%, you tell folks that the poll was Trump 30%, Crux 35%, Kasich 25%, Undecided 10%.

• Sometimes, they run multiple polls with small samples. Smaller samples give more random results. (E.g. to illustrate: instead of running a poll of 1,000 people, run ten polls of 100 people each. Even better, select 100 people from one location, race, sex, or religion. The results of any 100-person group will be dramatically more random than the results for the more representative 1,000-people group. Discard any of the 100-person group results that you don’t like, and keep the one result that shows what you want. It’s like flipping a coin until you finally get heads. With 10 coin flips, you probably will get heads at some point. Yay, you’re a winner.)

• Other times, they ask leading questions. Pollsters are experts at phrasing questions to make you give the answer they want. (Just like lawyers.) “Would you vote for Ted Cruz, who has been proven innocent of adultery - or would you vote for proven adulterer Donald Trump?”

• Did a respondent give you an answer that you dislike? Lose the survey form or dataset for that respondent. Whoops, office error. Survey someone else to make up for the lost data. Repeat as often as you need to, until you get the ‘right’ answers.

• And then we come to another popular technique, as used in Wisconsin recently, by multiple pollsters: ask people if they support Trump, but make sure two thirds of the people you ask are registered Democrats. And if youre feeling really lulzy, of those two thirds that are Democrats, try to make 70% of them female Democrats under 30. (The people least likely to vote for Trump, of the people least likely to vote for Trump.) It’s like asking Islamic Radicals what they think of bacon: you know their answer before you even ask them the question.

• Not to mention one of my favorite techniques, in the sense that Bubonic Plague is one of my favorite illnesses: combined confirmation bias/appeal to authority, by sampling ‘experts’. It’s not a poll, and it’s not illegal or fraudulent - but it does mislead people all the same. A certain major political website, right now (OK, it’s Politico) asked ten (yes, TEN!) ‘experts’ who they would pick as the Wisconsin Republican primary winner: 9/10 said Cruz. (According to Politico, and only Politico.) Now, that sample is so small - a population of just 10 - that it doesn’t mean sh-t. You have no reason to think their views are representative, or even sane. You don’t know who the f-ck they are, or if they’re all Cruz’s poker buddies. (And you’d be right to be doubtful, because Politico doesn’t even tell you which ten ‘experts’ it spoke to.) But - if you don’t know statistics, you’d think that ‘nine out of ten experts support Cruz’. (http://www.politico.com/story/2016/04/wisconsin-ted-cruz-bernie-sanders-insiders-221437)

• Then there’s the Nate Silver technique (named after the famous baker, so the political pundit Nate Silver can’t sue me, different guy): instead of reporting accurate polls of voter intentions, make up your own (not fully explained) rules about how you ‘estimate’, ‘guess’, or ‘calculate’ the winner. It can be as batsh-t as you like - and the more batsh-t, the less it reflects reality, and downplays Trump’s support. For example, you might say that Trump won’t win Wisconsin because men with names begging with ‘D’ or ’T’ didn’t win the Wisconsin Primary for the Republicans before, when a Democrat was in the White House, and more than 10 people started out in the race for the Republican nomination. Dress it up enough in fancy language, and readers will buy it. (Or just don’t explain your scoring system at all. Readers think math is hard, and journalists think math is hard cubed.)

(Pro tip: even better, use multiple scoring techniques, and combine the scores using a formula of your choice. Change the weighting each technique has, and which techniques you use over time, to make it even more like Calvinball. You get to call it 'refining your techniques', which sounds extra swanky.)

• And finally, the newest kid on the block: political betting markets. You see ‘political betting markets’ on the front of major news organisations’ websites, prominently displayed, often way above the actual polling data. What is a ‘political betting market’? People betting on who’s going to win a political contest. All kinds of people. Hillary supporters. Evangelical Cruz supporters. BLM supporters. Bernie supporters. Campaign volunteers, interns, and even ¡Yeb!, now that he has all that free time. Anyone who has an IP, and wants to make bets. (Not necessarily for money, sometimes the prediction market is just for 'points'.) It’s like fantasy football leagues in a way - 95% of the people playing have no f-cking idea what they’re doing. Which is the whole point, when it comes to politics: these idiots crowd-source predictions, using the information they get from the mainstream media, Hillary campaign emails, /sub/s4p, and that dream they had where Cruz visited them and declared he was the Anointed One. Bottom line: they don’t reflect reality, they distort it.

EDIT: WHAT YOU CAN DO ABOUT THIS SHENANIGANS!

Your emails, your tweets, your facebook posts, your posts here, your comments on news media websites - everything is automatically keyword read, and datamined, by the government; and by private analytical companies too; and by media outlets. By which I mean, they use the web as a giant 24/7 second-by-second way to survey all of us. So, give them your feedback. The more people expose their latest scams, the quicker they drop them. (Remember "Kony 2012"? Twitter went crazy calling it a Pentagon psyop designed to get AFRICOM into the mineral-rich nation of DRC, not least because warlord Joseph Kony had been dead for years before 2012. Days after the twitter storm, Kony 2012's chief organiser was found naked and acting deranged in public, was committed to a hospital, and the "Kony 2012" psyop was quietly shut down by the Pentagon.)

• Post comments on news websites, saying that you know the polls are rigged.

• Email the newspapers, and tv stations, and independent news sites, and say you know the polls are rigged. Cancel your newspaper subscription, or tell your cancel news provider you won't subscribe to lies.

• Email the Gop to complain. If you're volunteering in a primary, or at Cleveland, and are worried about the GOPe booting you out - set up a 'burner' email account (you don't need to give your cell phone number with many free email services) and do it anonymously.

• Tweet about the rigged polls. (I need to think of a catchy hashtag, but any tweet is good).

• Tell your friends and family.

• Always vote. Always. Vote rigging isn't always a clean and easy matter, and exit polls make fraud harder: so, the more votes Trump has, the more effort the GOPe has to make, trying to rig the vote in a way that just wins it for the establishment candidate, without looking improbable. Rigging a small percentage of the vote, to tip an election, is much easier to get away with than having to overcome an 80:20 Trump:Cruz split.

(BTW, if you don't use a VPN and practice online anonymity, your political views are already an open book. So, you might as well not be scared of speaking up for Trump in a legal and civil manner. If you do use a VPN and practice online anonymity, feel free to totally sperg out and make dick jokes about Cruz.)

Try one of these subthreads