What Netflix’s ‘Great Hack’ Gets Wrong About Cambridge Analytica

This is frankly par for the course in data science. I work as a data scientist/statistician for a relatively newly established (~5 years) analytics team at a large company, and the first order of business was weaning our teams off of the snake oil that grifter consulting companies were making millions selling them. The things I've seen referred to as "machine learning" and "AI" are absolutely bonkers. The reality is that most of the analytics services being sold B2B is total garbage, and yet it thrives because most customers aren't aware of why its bad.

To be fair to these consulting companies, my impression is at least some of the time they aren't knowingly selling BS, and that often times they just have very poor analysts/data scientists who have a very surface level knowledge of what they are doing, but not enough depth to realize why they can't do/aren't doing what they think they are. And the actual sales people know literally nothing but buzzwords.

These days we've been able to largely get the company away from vendors where possible and to rely on internal analytics teams, but it is still an issue when our team will come back to business partners with an answer someone doesn't want to hear, and they have external vendors essentially promising they can solve literally any problem with AI, and magically always produce the safe answer the customer wants to hear (but they never are willing to share code, the low level details of their process, or really anything that can make their analyses reproducible).

It's one thing to claim two teams _don't_ work together regularly; it's another to say there's a moat between them, indicating it was difficult or impossible for the teams to interact due to an external factor designed into the organization's structure. What does this moat mean practically, and why is this claimed division pertinent to the points the documentary makes?

Mind blown. The lengths campaigns will go to for the win. Can you share any more stories like that?

I tried searching for this but wasn't able to come up with anything. Does anybody know what this is referring to?

Edit: Found this in another comment https://news.stanford.edu/2017/11/28/neighborhoods-cars-indi...

You want a bad guy, with no regard for privacy, making crazy inferences (and getting a lot wrong)? That’s your guy.

Over the weekend I watched American Factory[3]. It may be my new favorite. I don't know if I will watch it as much as Smartest Guys, just because of the storytelling in that.

I like these because they clearly represent different viewpoints. Most documentaries feel like propaganda, and these don't feel that way to me.

0: https://en.wikipedia.org/wiki/Enron:_The_Smartest_Guys_in_th...

1: https://en.wikipedia.org/wiki/Dirty_Money_(2018_TV_series)

2: https://en.wikipedia.org/wiki/Alex_Gibney

3: https://en.wikipedia.org/wiki/American_Factory

Non-documentaries can do it to, like books or magazine articles, but nowhere near as strongly. Non-documentary video content can do it too, but usually without such a strong veneer of factuality. (Counterexample: An evening news program, which are probably even worse than documentaries, but are at least more limited in scope.)

I might make an exception in the future for something with no effective political or scientific content, but that's about it. Most documentaries are just more dangerous than can be justified.

"But who is this man, and what is this much-vaunted 'magic kingdom' he peddles to children?"

Now sometimes this is mostly benign or even "good." Such as detailing an event in a in-depth matter-of-fact style. Works such as the early Planet Earth series are simply trying to capture and show the wonder of life around the planet! And maybe hopefully encourage you to try and protect them as a secondary upshot. As time has gone on, they've produced works are that much more blatant with their "save the environment" messaging.

However, most of the most popular ones have a significant editorial bias. Heck, one of the first landmark pieces, Nanook of the North, was almost completely fabricated! Often quite easy to spot as well. Some are very up front about their bias, such as Michael Moore's work. Some less so, such as the editorial narrative in features like Forks over Knives or Blackfish. And some it's quite subtle. The recent Making of a Murderer comes to mind where certain facts weren't presented and it's framed in a very sympathetic light towards the defendants. Sure they have a lot of true facts within them. And often you learn quite a bit! But they are often presented in a particular fashion or certain points are talked up while others are ignored or not presented.

It's an unfortunate thing, but even the most innocuous of documentaries have some small bias. And it's up to the user to try and understand how much is narrative editorializing and how much is "truth." It's all part of making the documentary feel more true and tap into our need for a strong narrative arc via romanticizing the story we see unfold.

It would seem then that documentaries might be an effective form of propaganda for those with the resources to produce them and a public perception to change.

Markets in general are also chock full of unproductive Red Queens Races; you don't have to constrain yourself to politics to see how organizations can spend billions of dollars only to leave themselves in exactly the place they started. Look for instance at public outreach marketing expenses, writ large. Lots of these things turn out to be self-justifying; people spend money on them because money is generally spent on them.

You can also just read the meta-analysis study this piece cites, "Minimal Persuasive Effect Of Campaign Contact In General Elections".

The smaller the number, the less clearly one can point at a single cause though, right? I mean, to an extreme, say it had been a 1 vote difference. There's millions of factors that can make a vote happen or not (that one voter could've slept in because their alarm broke, that one voter could've gotten sick that day, etc). One vote is an extreme case, but for 80k I can think of a lot of factors that would normally be "small" but suddenly play a sizable amount of importance.

- That's 80k Democrats who didn't go to vote. - That's well within the territory of gerrymandering. - That's a tiny fraction of the African American population, where Hillary was notoriously weak. - That's well within the viewership of many Fox News shows and segments. - As the article stated, you could fit these people in a mid-sized football stadium. That's within the territory of having a weak last month of campaign touring, or aggressively getting more rallies done.

Just to mention a few. And yes, also means it COULD have been CA— perhaps they weren't as influential as they claimed but 80k is not a large number, they perhaps could've influenced that many people.

I'm just wondering if there's a lot of attention being put on this factor, vs other factors that at this scale could've feasibly swayed in the election.

To many, the “real issue” is losing the election. This is what the piece focuses on, and why it draws the conclusion that focusing on blaming CA for losing the election is not helpful for the purpose of trying to win future elections.

Meanwhile, in my experience, statistical understanding and the general skills required to understand why an analysis would or would not be valid is virtually non-existent for most people. Also, the majority of parties involved have a vested interest in the story of the effectiveness of marketing being perpetuated. Data vendors and advertisers, including giants like Google and Facebook, no longer have a product if it isn't. Consulting companies and internal analytics teams do ton of work solely focused on marketing. That all goes away if people realize its a useless exercise. The companies buying the advertising, who one would think would absolutely be interested in knowing if their dollars are going to waste, are filled with people whose jobs are contingent on the idea that marketing is driving business.

I'm not convinced marketing does nothing. But I think the effect is greatly overstated. And I think its becoming less effective over time. I can't remember the last time I clicked on an advertised result on Google, or a banner ad. Perhaps they were useful at one point years ago, but I think people are increasingly learning to tune them out over time, the same way one would acclimate to a noisy environment. Coincidentally, I remember seeing a link here on HN relatively recently about how the New York Times completely dropped their targeted marketing efforts, and found that their subscription growth remained unchanged. I think that were more companies to follow suit, many would see similar results.

So, it's not just about flipping voters.

It's also about encouraging people who agree your politics to head to the polls & discouraging the dissenters.

I didn't find their use of the data for marketing particularly novel but using that to create a campaign to discourage black youth in Trinidad from voting seems extremely unethical to me.

You gather the data required to make a good probability prediction for voter preference ((soft) labels for this easier to find than swing voter labels). Then when the model is uncertain, those are your swing voters / on the fence voters.

> Postcode? Age? Race? Gender? Income?

When it is found to be cost-effective: All and everything that is allowed by law and then some. In its pitch deck, Facebook boasted about its advertisers being able to target and identify: university, degree, concentration, course history, class year, housing/dormitory, age, gender, sexual orientation, zip (home and university/work), relationship status, dating interests, personal interests, club membership, jobs, political bent, friend graph, site usage/addiction level.

Likes make this very easy (with a little luck, you can deduce all of zip, age, race, gender, income from a list of Likes).

> What is it about CA's methods that were so effective?

Hillary Clinton: “The real question is how did the Russians know how to target their messages so precisely to undecided voters in Wisconsin or Michigan or Pennsylvania – that is really the nub of the question. So if they were getting advice from say Cambridge Analytica, or someone else, about ‘OK here are the 12 voters in this town in Wisconsin – that’s whose Facebook pages you need to be on to send these messages’ that indeed would be very disturbing.”

FBI: Using those techniques in June 2016, “the GRU compromised the computer network of the Illinois State Board of Elections by exploiting a vulnerability in the SBOE's website,” the report said. “The GRU then gained access to a database containing information on millions of registered Illinois voters, and extracted data related to thousands of U.S. voters before the malicious activity was identified. Similarly, in November 2016, the GRU sent spearphishing emails to over 120 email accounts used by Florida county officials responsible for administering the 2016 U.S. election,” the report said. “The spearphishing emails contained an attached Word document coded with malicious software (commonly referred to as a Trojan) that permitted the GRU to access the infected computer.”

> After all someone had to do something similar for Obama.

Obama's digital campaign was very successful, but the above seems to indicate that Kushner's campaign was way more aggressive and less scrupulous (and may have had connections with - or help from foreign adversaries).

It may also be that propaganda and smears works better depending on your political preference and level of education and neurosis: Even if Hillary had spent the same amount of money and energy (some reports indicate that Hillary's digital campaign was a waste of money and displayed poor management), efficiently, it may be easier to sway a voter to vote Republican, if you can target their fears of immigrants, religious beliefs, distrust in gun regulation from the government, and conspiracy theories. Surely, the many wolf cries about fake news, and retweeting of conspiracy theories, has set up the Trump base for easier manipulation (you can simply create a meme to counter a story in a respected journal or keep them guessing on the alternative truth of it).

1% of what population? 1% of Facebook users? 1% of the American population? Both of those are pretty staggeringly large numbers of people...

> It felt like the filmmakers were trying to put her into Snowden or Assange type circumstances

Given they clearly showed the quote by the data analyst that she was not a whistleblower and given they clearly showed her questionable morals, I disagree. She seems to me like the classic spoiled narcissist American woman, but then again Nix is also from an upper class family. She even admitted she was in it for the money, with a sob story about her family (as if that justifies it). No, this woman wasn't portrayed as a heroine. The documentary quite explicitly described her flawed character.

A realization I had is that Trump becoming president is just a 4-year thing. The UK leaving the EU is going to have a stronger effect on the long term.

I found it a good documentary which ultimately describes an example of where it went wrong (in many elections). It also clearly describes the long-term worries.

By the standards of these kinds of articles, it is the opposite of what you said it is.

That doesn't make it right, but does refute the one criticism you managed to marshal.

This sets the precedent of the entire article, and avoids mentioning the circumstances around the Bush Al Gore election of 2000.

And unfortunately, it also ends with such polarizing words:

> The Great Hack wants to make its viewers care about data rights and the dangers of modern misinformation campaigns but unfortunately is itself a slick piece of misinformation that plays artfully on the prejudices and misunderstandings rife in its targeted audience. If the weaponization of misinformation was such a powerful tool for changing hearts and minds, why did the left win recent elections in Mexico and Istanbul? Why did Democrats retake the House in 2018? We aren’t losing the war for the future because of some new media masterminds. If we’re losing, it’s because our message—and messengers—aren’t connecting.

Because 'the left' isn't excluded in social media manipulation. The narrative that 'the left' (whatever that exactly is) is somehow crying wolf is quite clear, and also tells the observer who this article is trying to convince (hint: those who got manipulated)

> The film gives an inordinate amount of attention to Brittany Kaiser, a Democrat who worked for CA’s US branch as its business development director and then—after the revelations of the Facebook breach—apparently had an attack of conscience.

They're omitting the data scientist who started whistleblowing. The very same data scientist who said in a Tweet which is quoted, that Kaiser was no whistleblower.

Does Kaiser get a lot of attention in this documentary? Yes. Is it unwarranted? I don't know. Is it all positive, is she put in a positive light alone? No way. If you watch the documentary carefully, you can find examples where she is being criticized. The description of events of her 180 u-turn is not pleading her free at all.

> According to her, until CA started working for Trump in June 2016, his campaign had no data infrastructure or digital marketing apparatus and no digital strategy. The implication, never blatantly stated but simply conveyed by all the tricks of modern documentary-making—striking digital graphics meant to illustrate how our data leaks into the hands of others, ominous music, and alluring close-ups of Kaiser as she watches the scandal unfold on television—is that Trump won because Cambridge Analytica gave him a secret edge.

> All of this is garbage. Kaiser first worked in politics as an intern on the 2008 Barack Obama campaign, helping its social media team, but The Great Hack implies that she ran his whole Facebook operation.

Classic straw man, keywords being: implication, implies.

The amount of people who had to be manipulated were carefully chosen, and had to be very little amount in US thanks to the way the election system works (with the districts).

Just because something is the status quo doesn't imply that the it is the optimized reality. I work in data science and quite a bit with marketing data, and in my experience, its highly ineffective most of the time. More surprising to though, is the sheer number of people, companies, and dollars, on both sides of the marketing business, with major investments in the idea of marketing being effective.

People believing it works and failing to verify?

Having worked with this technology, I question its effectiveness. Do you have any data to back up your claim that it can sway people's opinion? It seems dubious that it's more effective than traditional brand building through mass media advertising (which itself has never proven to be particularly effective).

In other words, CA lied about the intended use. From a data collection standpoint, it was the same thing Obama's campaign did in 2012 for get out the vote efforts.

"Parscale said he asked the Facebook “embeds” to teach staffers everything the Clinton campaign would be told about Facebook advertising “and then some”. Parscale told CBS he was told the Clinton campaign did not use Facebook employee embeds. “I had heard that they did not accept any of [Facebook’s] offers,” he said."

So really this is all due to a complete incompetence on one campaigns digital strategy while another actually put in work.

The documentary, at the end, with the teacher in and students drawing a conclusion makes the point that this isn't just a one company thing. It makes the point that big data companies are the problem. They even quote in an interview the names Google and Facebook.

However, as someone else already replied to you, this API which was used by CA wasn't meant to be used anymore. That's a big difference.

https://www.cnn.com/2017/09/13/politics/susan-rice-house-inv...