Hacker News new | comments | show | ask | jobs | submit login
Cambridge Analytica: The Data That Turned the World Upside Down (vice.com)
160 points by lakeeffect on Feb 1, 2017 | hide | past | web | favorite | 99 comments

I'm loath to uncritically accept the claims of this article. It just tells you what they did and how Cambridge Analytica thought it had an impact. How do they actually know it worked? The claims are pretty strong, yet the article reads a little like marketing copy for the company. After all, a lot of things happened in the election, really a lot. Plus, the whole thing about "shadowy private company relying on subconscious decision-making" does kinda set off my bullshit alarms.

Cory Doctorow agrees: https://boingboing.net/2017/02/01/trumps-big-data-secret-sau...

And a more thorough debunking: http://civichall.org/civicist/will-the-real-psychometric-tar...

With this quote: "I have described them as the Theranos of political data: I think they have a tremendous marketing department, coupled with a team of research scientists who provide on virtually none of those marketing promises." :)

But on a second thought. Planting this story is a masterpiece of promotion - and isn't this kind of similar to promoting a politician?

It looks like it's more that Cory Doctorow agrees with Cathy O'Neil https://www.bloomberg.com/view/articles/2017-02-01/trump-s-s...

Yeah, I was recently thinking of applying to work at Cambridge Analytica, so I had reason to do some digging on them. Like a lot of people applying data science to problems, they're riding off hype, buzzwords and clients not fully understanding what they're doing.

One reason I didn't end up applying is, though I lean to the right politically, I'm not at all a Trump supporter and couldn't bring myself to work on getting him elected. The second reason is that, even if I was working on a non-Trump project, it wasn't clear that their product added that much value. Some media references to them agreed with this, pointing out that, e.g., their model requires campaigns to create multiple adverts targeting different personality profiles, but most don't have the resources.

I think the primary evidence is near the end of the article:

"Kosinski has observed all of this from his office at Stanford. Following the US election, the university is in turmoil. Kosinski is responding to developments with the sharpest weapon available to a researcher: a scientific analysis. Together with his research colleague Sandra Matz, he has conducted a series of tests, which will soon be published. The initial results are alarming: The study shows the effectiveness of personality targeting by showing that marketers can attract up to 63 percent more clicks and up to 1,400 more conversions in real-life advertising campaigns on Facebook when matching products and marketing messages to consumers' personality characteristics."

The subject of big data and the role of data brokers in elections seems to be trendy. I've seen a few pieces around on Cambridge Analytica due to Trump ending up POTU but also on the bigger players: Experian, Equifax, Epsilon, Acxiom.

Those are very real and do play a role in influencing targeted voters in elections, but the question is not about who do what and how it works. It is about how this way of campaigning is changing the way candidates convince voters ( I was gonna say democracy but there are no actual democracy anywhere right now ).


Believing Cambridge Analytica are a bunch of geniuses doesn't only benefit CA, but also to the people that CA defeated. If Trump had a crazy amazing data science team, it means that Clinton and the "Mook Mafia" didn't mess up, they were just outplayed.

I think this whole pro-data story is incorrect. Trump's huge unpopularity right now shows that what happened this election was Clinton messing up her own campaign by religiously following data and not doing basic things like visit Wisconsin. And from the journalism end, the NYT telling readers that Clinton had an 80+% chance of winning, sending a loud and clear message: Stay home if you wish to protest vote.

Even after massive, massive predictive failures, fortune-tellers always manage to give excuses and stick around.

"Trump's huge unpopularity right now shows..." After watching what happened with the election, I don't think it's safe to say Trump is unpopular. Sure the mainstream media and loud social media voices say so, but that looming "silent majority" are still silent. Assessing his popularity among those willing to speak only tells us how popular he is among that group.

He lost the popular vote by 3 million votes.

His disapproval rating has eclipsed 50% in record time for a modern president - 8 days. 42 vs 51, approve/disapprove https://www.rt.com/usa/375715-trump-majority-disapproval-8-d...

President Obama had an approval rating twice as large following his inauguration. 84/14

Over 3 million people marched against him just two weeks ago and continue to mobilize.

I can't find any evidence that there is some silent majority supporting him. I'm more inclined to think it's a vocal minority.

> He lost the popular vote by 3 million votes

Maybe if it mattered he would have campaigned differently. You can't judge on a metric that isn't the target.

> His disapproval rating

opinion polls?

> I can't find any evidence that there is some silent majority supporting him

He won the election...

What, do we use the Electoral Popularity College now, too?

And Trump actually got amazing positive opinion polling 6 months ago. Now that it's not in his favor, though, ignore?

And you can absolutely win an election with both minority vote share and a lack of popularity.

> Now that it's not in his favor, though, ignore?

I've never payed attention to them. They seem to be used more for the political purpose of manipulation, than for producing an innocent, yet unnecessary, prediction of results

> And you can absolutely win an election with both minority vote share and a lack of popularity.

No, you still need a certain amount of voter popularity, with the right distribution; unless you are talking about something else.

If they repeat it long enough, they hope to make it true.

He didn't even get more votes.

Whether or not I agree with the premise of Cambridge Analytics's services directly leading to the Trump campaign emotionally manipulating loads of disparate groups into electing him, it's alarming how easy it is to conceive of someone intentionally doing such a thing at such a scale.

YC made a request for startups targeting news and democracy the other day. I'd like to propose additional underlying unmet societal needs:

Emotional resiliency & nonviolent communication.

We've all known that this has been coming for a long, long time. Education won't get the job done. Strong business ethics won't either.

No, this shit needs to be regulated into the ground. No more of this opt in bullshit, no more selling data that was never yours to sell, we have to cut out immediately. We MUST limit the amount of datapoints that can be used for advertising and we must limit access to that data with much stronger privacy laws by outlining EXACTLY what data is being collected, providing consumers access to that data, and not allowing business to sell data that to other business. Internal use only, that has to be the rule.

Freedom of thought is at stake, we have to act fast or we are totally boned.

Building a system that bypasses the problem entirely by disintermediating media consumption from advertising and tracking would be even better.

any clever ideas?

Universal content syndication, with distributed delivery.

Static pages would help loads.


You lose the "freedom of thought" simply because a computer does exactly the same thing that a good salesman would do, simply at scale?

A hydrogen bomb does the same thing that Greek Fire does, at scale.

The gas turbine does the same thing the ox does, at scale.

Mustard gas does the same thing a bee does, at scale.

Scale fucking matters.

Your examples illustrate my point quite well. It is (and should be) illegal to pour greek fire on even 1 person - small scale doesn't make the action ok.

Does this mean we should pass a law against salesmen, and just general interpersonal persuasion? If not, why not?

I'm going to try to put my finger on the nature of your comments that's troubled me for quite some time.

A key element that's lacking seems to be the principle of charity: seeking a sympathetic understanding of an idea presented, and interpretation of ideas in their most persuasive form.

I don't believe I've ever seen you do that. You might care to give it a thought.


I'll through your questions back at you:

How is Greek Fire unlike a hydrogen bomb? How might considerations of these be different? What else that shares elements of what a hydrogen bomb is, or does, still substantively different in a way that would not require some sort of regulatory treatment?

Why is it we pass laws, generally? What are the hallmarks of a good, or a bad law?

Gah. "Throw", not "through".

If I sell you something I know you have the money for, but can't afford, that's a predatory action.

If I convince you to harm yourself, that's a predatory action.

We have laws of various forms making these actions illegal. They may not be federally regulated, but could probably stand to be, especially when it's done at scale.

Who needs protection of the law in those situations? The emotionally exploitable or the unethical salesman?

no one is talking about passing a law against campaigning or politicians.

What we are talking about is passing a law that says your data can't be used against you.

How does this mesh with the idea that information wants to be free and DRM is evil?

Indeed if the security of control of your government is fragile enough that a fancy telemarketer can acquire significant power perhaps a bit of governmental re-architecture is in order.

Well, what's problematic is democracy. Representative government doesn't protect against this sort of data-driven populism. If people can be manipulated so efficiently, how can democracy be workable?

One answer: restricting the vote to people who have skin in the game. If you have to personally pay for your favored policy you might consider it more carefully.

Yes, people with less to lose are arguably easier to manipulate. But I'd rather see some sort of "I am not a bot." test for independent thinking ;)

Not only are they easier to manipulate - they have a conflict of interest when they (for instance) demand that the government pay them money when they don't pay taxes.

That, I believe, is among the least of our problems.

By shifting societal norms toward learning nonviolent emotional independence?

That seems a good start. But maybe that's game-able too. If you're dealing with an AI (or proto-AI, at least) that understands you better than you do yourself, you're pwned. So then you need your own AI, to filter input and flag exploits. Rather like anti-malware, for your mind.

That's what I'm developing :D

The idea is to create a human programming language to encode activities for learning/practicing principles found in the book Nonviolent Communication and publish that before publishing the language as a way to protect oneself.

Cool :)

Sounds expensive

No, you lose your freedom either way. But a society loses its freedom when software does it at scale.

For those who want to get started on developing their own, here are some resources that have really helped me:

"Nonviolent Communication" - Marshall Rosenberg

"Mindsight" - Daniel Siegel

I've also been working on a new mathematical model (rooted in category theory) for how our brains, bodies, and minds work together with the goal of developing a natural language based way to generate practices for the sake of improving myself in targeted ways. I only just finished reading Mindsight & have been practicing some of the techniques in the book by accident for months prior after coming up with the practices myself through my model. I've found my other attempts at programming myself using the model also develop my ability to focus & direct my attention, as well. As a result, I've had initial successes with learning echolocation and learning a form of synesthesia I haven't heard of (seeing a stick figure that moves with my body), both in under 10 minutes of my first attempts.

I'm not an expert in any of this stuff. I'm an information addict in recovery who saw connections among various recent research findings in different fields and started making connections. If love to collaborate with any Neuroscientists or Category Theorists. My contact info is in my profile.

I think about the future and it's relation to mind. For example, we hear a rustle in the jungle, consciously or not we take action against a future in which we are eaten by a tiger. Siegel points to this regulation over time in his definition, good stuff.

Your self improvement is related to the future too. Do you mind if I ask, what kind of tigers do you run from?

BTW. Thank you for the info, this might be worthy of a detour :)

My tigers are all activities dealing with household chores, organizing just about anything, and producing things people will see. I've been avoiding related tasks since I was a kid because of blame/shame/judgment resulting from my attempts. It sounds silly and spoiled, but more reasonable after considering it started when I began shutting down after multiple sexually traumatizing events and a bully "best friend" next door. I'm only now developing the tools needed to overcome the deeply embedded anxieties. Luckily, they're mostly due to fears of tigers that aren't around anymore, so the main trick is to identify what the beliefs are surrounding the tasks and choose to believe the opposite.

As for the future, I consider this year the start of the neural age because it'll be the first year in which a general programming language for the brain/mind/body will be available.

I am afraid of determinacy, like you, my actions are related to my past. I see it all around me. Science embraces this notion of causality, my friends and everyone else embrace it as well. Though the tendency is to compartmentalize causality, sure people explain why things happen all around them. It goes right up to an inch in front of their nose, but they don't want to take it any farther.

Truth is, I don't know who I am. Where I end and you begin.

Cya around friend.

We royally fucked up with the internet. I literally have to question my own motives for things, and often wonder if I have been manipulated into my beliefs, opinions, and desires. It is kind of scary to think about!

How do we move to a less centralized internet and is it too late?

Fun question: where does the idea "decentralization good" come from?

What big players could have an interest in this and why?

Decentralization is a way of removing dependencies and adding redundancy. It's often less efficient, there's way more duplication of effort, but can be more resilient.

Big players don't necessarily have an interest in this, but some do. It's always a balancing act between too much centralization (single target/point of failure) and too much decentralization (impossible to control).

I'm trying to think of specific actors that have interests in this, but I'm finding it hard. Not the NSA, right?

I don't understand why you think decentralization is a balancing act. What do you mean by "impossible to control", and why would that be bad? What would "full" decentralization even entail? A network of sentient computers doing whatever they want, except using supernodes "too much"?

(My definition of supernodes: nodes that transfer significantly more traffic than others structurally, in practice nodes that accept incoming connections, like servers or BitTorrent supernodes, but also AS's.)

The military and certain branches of the government (e.g. CIA) are very interested in having modular, autonomous structures that can survive in the event of a decapitation attack, or if communications are severely disrupted through other means.

They're also interested in having centralized control so the various components of their organization are not running rampant and in conflict with the others.

It's a delicate balancing act. Look at the US as a whole: If states are too independent they might get into wars with the others, not unlike The Holy Roman Empire. If they're too much under the thumb of the Federal government they'll feel oppressed.

If (my example) instead of people messaging each other on facebook, every single person needed to run a server (such as an IRC server) at home - no centralization at all. (If five people want to talk to each other, there are a total of 10 connections because 4+3+2+1 to fully connect every node with every other node). That would be an (extreme) example of decentralization, and obviously centralizing chat so I can just use a thin facebook/whatever app is hugely convenient. Likewise the blockchain is a joke, a single $200 pc can trivially do the work of the entire bitcoin network, trivially - if it were centralized. This second example also shows what is meant by control. So it's a balancing act.

The internet is both centralized, we need to agree on protocols, coordinate address spaces, issue domain names, and decentralized, where elements of it can survive independent of the others if they're temporarily disconnected.

The Bitcoin blockchain is an example of something that's a complete contradiction: It's a highly distributed highly centralized system. There's only one Bitcoin block chain, and everything must be recorded there for it to be valid. If, for example, Iran is cut off from the greater internet then that means you can't use Bitcoin in Iran.

BitTorrent is a much better example: There's no central authority, it's truly distributed, and highly fault tolerant. If Iran was cut off for some reason, all the Iran-hosted seeds of any content would still work.

I agree with everything you've written.

"Discordians shall not believe anything they read."

That is only half the problem, from the article: "Cambridge Analytica buys personal data from a range of different sources, like land registries, automotive data, shopping data, bonus cards, club memberships, what magazines you read, what churches you attend."

Data collection like this still occurs without the internet. In addition, decentralization of the internet does not necessarily prevent targeted ads. IE, now instead of buying an ad from facebook that targets everyone with interest x in geographic area y, now we buy an ad on a local website for interest x in geographic area y.

Since targeted ads (and sponsored Facebook posts) are the vectors for influencing people, wouldn't it be a great deal easier if people just blocked ads? No matter how much accurate a profile is built off you (through other companies selling your data), you can't be influenced by ads if you don't see them. Especially since we know that ads are not just trying to persuade you to buy things (which is probably still morally acceptable, at least to me), but also to influence your political affiliations without you realising it.

Not sure if being influenced unbeknownst to you is a n entirely new phenomenon though. What was its extent with print or TV media where everything is centrally controlled?

At least one core problem is that no one wants to read primary sources and make decisions themselves. We seem to prefer an illusion of independent thought when we're actually choosing from a set of prepackaged partitions that are neatly presented to us.

I mean Facebook for example, at its core for an end user it really just mirrors what the Internet was built to do. I can easily host my own profile and allow people to log in to it and talk to me. How did it become so dangerously centralized?

The average user is lazy, cheep, and unaware of how valuable their freedom is.

Did you mean "sheep" or "cheap"?

I think we've just witnessed the birth of a new technology portmanteau on HN- "Cheep: (noun) Member of one or more online herds of users who are exploited for personal data". I like it.

(sadly, cheep.com is already taken or I'd be on that like white on rice...)

> How did it become so dangerously centralized?

Network effects and there's nothing out there better than Facebook.

Facebook owns the directory, and the notifications channel.

Freestanding sites lack this, and it turns out that behavioural costs matter.

I think Urbit is trying to tackle this problem of decentralizing the internet. I'm not sure of their current status however.

Decentralizing won't help much.

The who really knows what is fake news? And what about the filter bubbles that keep the full picture hidden from you?

A decentralized search engine with configurable or even just inspectable filter bubble would be a good start?

(Although I think the correct abstraction is not "search engine", but "decentralized content index" + "decentralized content ranking" with tweakable parameters.)

> I literally have to question my own motives for things, and often wonder if I have been manipulated into my beliefs, opinions, and desires.

Its possible to be immune for such manipulaions, I believe. If you subject you thoughts to strict formal rules of rational thinking, than it would be hard to manipulate you.

But the first and most important thing is constant questioning own motives. So you are on the right track. ;)

> If you subject you thoughts to strict formal rules of rational thinking, than it would be hard to manipulate you.

If the brain was fundamentally a linear processor, you might have a chance. But the brain is almost entirely a pattern-machine, and linear, logical thought is mostly an artifact.

Humans can be programmed like any AI. It just requires the correct stimulus.

I'm not saying that one can be completely rational. I'm saying that you can be much more resistant to mind influencing technologies. Moreover you can be resistant for such a level, that you'll become immune for technologies which targets 95% of population.

In addition to what riprowan said, if you are perfectly logical and rational, but I can control your input data, I can control your conclusions.

Really? What you know about my consclusions? How can you check that you control them? By looking at my behaviour? Ha-ha. How you can control them if you have no precise knowledge about them and processes leading to them? Oh... And how can you control my input data, if I'm not reading your ads?

> How do we move to a less centralized internet and is it too late?

Oh great, divide et impera.

Not again this story. This is a very sensationalist take on the impact of micro-targeting. It probably has some impact, but every political campaign has been doing this for years and there isn't something inherently special about Cambridge Analytica's approach (other then being funded by the Mercers and hence Trumps go-to firm).


Thanks for sharing. Cathy O'Neil is one of the most credible voices in the room when it comes to understanding the unintended consequences of applied big data algorithms.

The original post is sensationalistic and massively overstates what's currently possible when it comes to micro-targeting.

From a satirical article published on Halloween last year:

> [Campaign] has also hired a firm specializing in big data and advanced intrapsychologic modeling. [Data firm] then takes data from Cookie Monster and analyzes it using their own proprietary Artificial Intelligence-powered (AI) algorithm, which allows the campaign to not only identify key voters, but to also identify key parts of their brains that are activated by certain messages.

> “Most campaigns only look at individual voters. We take it a step further and dig down into key parts of the voter’s subconscious. That way, we can say, ‘This meme penetrated a voter’s volitional association area of their prefrontal cortex — let’s double down on this message.’”


I'm sure Cambridge Analytica would love us to think they had an unprecedented impact. I haven't seen any actual evidence though.

There are several elements of this article that are refuted by Dominic Cummings, who ran the Vote Leave campaign (a different organization than Leave.EU).

You can read a more detailed description by an insider for the software Vote Leave used at his personal blog, here: https://dominiccummings.wordpress.com/2016/10/29/on-the-refe...

The impression I got when I originally read this article was that:

1) Trump outsmarted Clinton (and the presumed technology advantage she inherited from the Obama campaign apparatus) with psychometric local targeted propaganda / communication

2) Some of that communication may have been deliberately targeted at discouraging democratic voters by putting negative articles about Clinton in their social media feed.

This is interesting in the context that Trump only won by a 70K voter advantage split over three states.

Clinton was predicted to win which really hurt her turnout. She was also relatively unpopular as her numbers would often drop after a personal apearance.

While she won the popular vote by 3 or so million people, trump's message was very targeted to areas that mattered. A republican with the slogan 'your fired' ended up being the pro worker candidate in many areas. That takes very skillful image manipulation and a gullible electorate.

PS: Don't forget he was predicted to have around a 40% win chance. That's far from negligible despite what people where thinking.

> Clinton was predicted to win which really hurt her turnout.

I'm not sure this claim can be backed up by data.


Is some support for turnout being a problem. "Wisconsin tells the same numbers story, even more dramatically. Trump got no new votes. He received exactly the same number of votes in America’s Dairyland as Romney did in 2012. Both received 1,409,000 votes. But Clinton again could not spark many Obama voters to turn out for her: she tallied 230,000 votes less than Obama did in 2012. This is how a 200,000-vote victory margin for Obama in the Badger State became a 30,000-vote defeat for Clinton."

My thoughts where people would have been more willing to hold their noses and vote for her if they thought it was going to be a very close election. Few democrats in Michigan really thought the state was up for grabs at something like 78.9% vs 21.1% https://projects.fivethirtyeight.com/2016-election-forecast/...

More support for this is new-hampshire (70-30) was predicted as a closer state than Pennsylvania (77-23), but ended-up very comfortably with Clinton. Remember, Pennsylvania (20) + Wisconsin (10) + Michigan (16) + 227 = 273 and a win.

Now, I am not saying the effect was huge, but in tight elections it does not take much. Further, both the Senate +2 and House +6 shifted toward democrats though maintaining republican leaning.

"Democrats close to Bill Clinton said Thursday that one mistake Clinton’s top aides made was not listening to the former president more when he urged the campaign to spend more time focusing on disaffected white, working class voters.

Many in Clinton’s campaign viewed these voters as Trump’s base, people so committed to the Republican nominee that no amount of visits or messaging could sway them. Clinton made no visits to Wisconsin as the Democratic nominee, and only pushed a late charge in Michigan once internal polling showed the race tightening.

Bill Clinton, advisers said, pushed the campaign early on to focus on these voters, many of whom helped elected him twice to the White House. The former president, a Clinton aide said, would regularly call Robby Mook to talk about strategy and offer advice.

But aides said the Clinton campaign’s top strategists largely ignored the former president, instead focusing on consolidating the base of voters that helped elect President Barack Obama to the White House. In the closing days of the campaign, Clinton targeted young people, Hispanics and African-Americans with laser like focus, casting Trump as a racist who only sought the presidency to benefit himself."


Not sure where that says anything about a turn-out issue?

I am not sure if it hurts a candidate more to be labeled "never going to win" vs. "is sure going to win". Trump had to fight against an incredible dominant loser narrative.

He almost was forced to make his voters believe in an alternative reality in order to actually go out and vote.

It depends on the candidates.

If you like or are neutral to candidate X, and dislike Y then voting even if it's pointless still feels good. So, win or lose voting still feels good.

If you dislike X, but dislike Y more then voting feels painful. If you think it's close then you are more willing to go though that pain. But, if you think X will win and you don't want to vote for them then there is little downside to staying home. On the other hand, if you think Y is going to win then you can say to your self it's not a vote for X, it's a vote in protest to Y.

Granted, if you dislike X, and hate Y or have something else you care about, like voting in every election, and are in the booth anyway then that's another story.

Now, this is probably not a huge effect but, in states that where within 1% it still matters.

I wonder if this isn't some kind of mass denial happening in some subset of the world. I've been seeing a lot of theories abot why Trump won, including mass manipulation and Russian influence. But what about the simplest explanation - that he won because people voted for him (usual vote counting shenanigans that happen every election aside)? Is this fact so scary people need to rationalize it away?

Saying "he won the election because he got more votes" doesn't explain anything. The question is who voted for him, and why.

And that's not about Trump specifically, it gets asked after every election.

I think if you are ideologically opposed to Trump then you find it tough to accept something like this. Please also note that HN by nature tends to be more on the liberal spectrum. Like everything else, it is a bit of everything. It's Clinton complacency with Trump's strong messaging + democrats backstabbing their own voters by sabotaging Sanders. Being someone from marketing, it is easy for me to see that "MAGA" is much catchy & better slogan than "Stronger together" (what does it even mean?), it is difficult of any American to argue against MAGA vs. Stronger together since you will be considered "unamerican" & nationalism is currently a theme globally. Trump kept things simple and easy to grasp. If you read Daniel Kahneman & Amos Tversky's work, he basically appealed to our system 1 minds which prefers shortcuts and jumps to conclusion. (which does not mean that people who does this are somehow bigots because we all do this all time, in fact living would be really difficult we were ultra rational and thoughtful about everything)

That's not how the electoral system works more often than not people do not vote for, they vote against.

Things would be different if we actually voted for policies instead of oligarch strawman.

What if CA could not, in fact, influence the outcome of the election, but instead predict it, and then they just made the bet to approach the winning party and got $$$?

And then they just keep doing this at each election or referendum and claim each time to have been influential?

What if?

So they received 15 million for being able to influence people on a massive scale (at least that's the claim). The article makes it sound like they can micro-target a solid percentage of the US population into action or non-action. If this were true, 15 million is a ridiculously low amount of money they got for that ability. I mean the money they should be raking in from advertising should be...tons. If they can change the election behavior of millions of people surely they can get millions of people to buy an extra soft drink here and there...

So maybe this is an advertorial for CA, or maybe it's Chicken Little clickbait. Or maybe both. But even Kosinski's work alone is frightening. This is one way that AIs will pwn us. Indeed, how AIs are pwning us.

There were several articles on Cambridge Analytics and Trump but none made it to the HN front page, including the English translation of the original German article.

I think this is an (another?) English translation of that Das Magazin article.

To be honest the story is actually super boring. Nothing of substance how it actually works. It is for people who apparently have never noticed that it is possible to show different ads to different target groups...

Is this title just the six most used words in the article? :-D

It's incomprehensible to me - after some time I just started parsing it as "Big Data, Cambridge Analytica, Brexit, Trump". It'd be better if it was just re-titled to use the article headline, I think.

We've updated the title from “Big Data Cambridge Analytica Brexit Trump”, but we're happy to change it again if someone can suggest something less click-baitey than the article's title.

It's words from the URL. I'm guessing they were used because the article headline isn't very descriptive.

This HN title is word salad.

Article's title has been improved. It was baaaad in its first version.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact