Did leaked Facebook data swing the 2016 vote?

[Last updated 3/22/2018]

Video added 3/19/2018

The Guardian published an article about how a company called Cambridge Analytica used unauthorized data obtained from Facebook to help swing the 2016 election to Donald Trump.

The Facebook “likes” and other data were used to draw psychological profiles of individual voters, who were then targeted with messages based on those profiles.

A year or so ago, I made a post, based on an earlier article in The Guardian and an expose by the Real News Network, about how Steve Bannon and the Trump campaign used Cambridge Analytica to identify idealistic liberals, young women and African-Americans in key states, and feed them information to discourage them from voting for Hillary Clinton.

Many people question whether such manipulation was possible on a significant scale.  I am not qualified to say.

The thing is, targeted messages don’t have to work every time, or even most of the time—just enough times to tip the balance.   And the technology is being constantly improved, so even if they didn’t make a difference in 2016, they may affect the next election and the one after that.

I don’t have good ideas as to what to do about this.   It is not unethical to send accurate information to someone you think will respond to it.  Does it become unethical when the information and its target are chosen by an artificial intelligence program?  At the very least, we the people ought to be able to know where the messages come from.

Afterthought [3/20/2018]

After thinking this over for a couple of days,  I’m of two minds about Cambridge Analytica and similar companies.

How artificial intelligence elected Trump

Hedge fund billionaire Robert Mercer bailed out the Trump campaign last summer when it hit its low point, but that was not the most important thing he did.

The most important thing was to teach Steve Bannon, Jared Kushner and Jason Miller how to use computer algorithms, artificial intelligence and cyber-bots to target individual voters and shape public opinion.

The Guardian reported that Mercer’s company, Cambridge Analytica, claims to have psychological profiles on 220 million American voters based on 5,000 separate pieces of data.  [Correction: The actual claim was 220 million Americans, not American voters.]

Michal Kosinski, lead scientist for Cambridge University’s Psychometric Centre in England, said that knowing 150 Facebook likes, he can know a person’s personality better than their spouse; with 300 likes, better than the person knows themselves.

Advertisers have long used information from social media to target individuals with messages that push their psychological buttons.

I suppose I shouldn’t be shocked or surprised that political campaigners are doing the same thing.

Bloomberg reported how the Trump campaign targeted idealistic liberals, young women and African-Americans in key states, identified through social media, and fed them negative information about Hillary Clinton in order to persuade them to stay home.

This probably was what gave Trump his narrow margin of victory in Wisconsin, Michigan and Pennsylvania.

The other way artificial intelligence was used to elect Trump was the creation of robotic Twitter accounts that automatically linked to Breitbart News and other right-wing news sites.

This gave them a high-ranking on Google and created the illusion—or maybe self-fulfilling prophecy—that they represent a consensus.

Flawed algorithms mark people for death

The National Security Administration’s Skynet system marks people for death based on algorithms and metadata—the same technology that Amazon uses to guess what books you’ll probably like.

I find that chilling.  I find the precedent it sets even more chilling.

Now an expert has come along who says the Skynet program is inherently flawed and has likely resulted in the deaths of thousands of innocent people.

Documents leaked to The Intercept indicate that the Skynet program collects data on people in Pakistan by monitoring their phone calls.  Supposedly terrorists can be identified within a certain margin of error by characteristics that, on average, differentiate them from non-terrorists.

Patrick Ball, director of research for the Human Rights Data Analysis Group and a frequent expert witness before human rights tribunals, told Ars Technica that the problem with this is that the terrorist sample is based on a very small number—seven individuals—and the innocent sample is based on a random sample of 100,000 people.

Since there is usually no independent way of verifying that the victim really was a terrorist, that means that there is no “learning” process, as would be the case with a commercial algorithm, such as Amazon’s, which is based on commercial responses.

One of the variables in setting the algorithms is that the fewer false negative (real terrorists who are not detected by the system), the more false positives there will be (innocent people who are marked as terrorists).

Bell said that if the algorithm is set at 50 percent false negatives, that means thousands of innocent people will be killed for every real terrorist who dies.

[Added later]  Martin Robbins wrote in The Guardian that Skynet is used to identify likely Al Qaeda couriers, who are not killed but tracked so as to reveal the locations of Al Qaeda camps and safe houses.   It is a fact that computer algorithms are used to target people for killing, but Skynet isn’t as clear an example as I originally thought.

China tests using credit scores for social control

Source: China Daily.

Chinese authorities are experimenting with a new method of social control.

It is a credit score generated by Big Data methods that evaluates not only a person’s financial record, but everything that could throw light on their moral character, including associations and lifestyle choices.

A good credit score would give a person not only certain privileges, but prestige.  Conformity would be induced not through threats and punishments, but through positive reinforcement.

Last year the Chinese government announced it is working on something called a “social credit system” to enhance “sincerity discipline” in government, commerce and society in general, which is scheduled to be launched in 2020.

More recently a Chinese credit card company started testing a credit rating system that will use social media to gather information not only on people’s finances, but their hobbies, shopping habits, overall lifestyle and interactions with friends.

Based on that, the person will be given a rating of between 350 to 950 that not only determine their access to credit, but other privileges as well.

Some analysts think the two systems will come together to produce a system of total Orwellian surveillance, a kind of incentive-based totalitarianism.

Every aspect of a Chinese person’s life, including political opinions and friendships, would be fed into the system, which would produce a numerical score based on an algorithm.   That score in turn would be the basis for rewards and punishments that would shape the person’s whole life.

Now this is speculative.   I don’t know that the Chinese government actually has this in mind.

Technology and its discontents: Links 2/17/15

Socialize the Data Centers! an interview of Evgeny Morozov by New Left Review.

Knowledge really is power.   Information available on the Internet enables big organizations to know—or think they know—everything important about you.  Evgeny Morozov, a technology writer and critic, believes Big Data should be subject to democratic control and privacy safeguards, not monopolized by private companies such as Google.

One American City Enjoys a Hyperfast Internet—Any Surprise Corporations Don’t Control It? by Thom Hartmann for AlterNet.

Chattanooga, Tennessee’s publicly-owned fiber-optic Internet utility operates at a speed of 1,000 gigabits per second—about 50 times faster than in the average American city where Internet service is provided by for-profit companies.

New High-Tech Farm Equipment Is a Nightmare for Farmers by Kyle Wiens for Wired.

Tractlor manufacturers such as John Deere make it virtually impossible and maybe illegal for farmers to repair and reprogram their own tractors.

The invisible network that keeps the world running by Tim Maugham for the BBC.

Containerized shipping enables the global supply chain to function.  It requires complex coordination that can be done only by computer networks.  The author speculates that someday the process of sorting, loading and unloading cargo may be completely automated, with no human beings in the loop.  What, I ask ironically, could possibly go wrong?

South Korean woman’s hair ‘eaten’ by robot vacuum cleaner as she slept by Justin McCurry for The Guardian.

Technology is an extremely useful servant, but, as any rich person can tell you, people with servants need to keep an eye on them.

U.S. Chamber and high-tech dirty tricks

The U.S. Chamber of Commerce is a leading—maybe the leading—lobbyist for big business and conduit for big business campaign funds.  Now a new high-techninitiative, code-named Team Themis, has begun to use disinformation and NSA-type surveillance to discredit the Chamber’s critics.

Lee Fang of The Baffler has the story.

Following the 2010 midterm elections, a group of defense contractors, including Palantir Technologies, a Silicon Valley data analysis startup, exchanged hundreds of emails discussing how to customize their wares for an exciting new prospective client.  The contractors were developing a state-of-the-art surveillance system and had been in direct conversations with the Chamber and its law firm, Hunton and Williams … … .

The spying operation would gather massive amounts of personal information, some from meta-data scraped off social media accounts like Facebook, LinkedIn, etc. and some stolen through illicit “custom malware” attacks.

The group, nicknamed “Team Themis” after the Greek goddess of law and order … … would keep tabs on an array of journalists, activist groups, and labor unions.

As one of the people crafting the proposal explained, Team Themis would resemble the “fusion cell” used by the Joint Special Operations Command—the elite military unit that hunted down Osama bin Laden.

For these contractors, the opportunity seemed like a natural application for their technology.

After all, Palantir, a Big Data firm founded in part with an investment from the Central Intelligence Agency’s venture capital arm In-Q-Tel, has won contracts from several U.S. intelligence agencies, the Marines, and the Army. And Palantir’s proprietary software scans through immense quantities of information, searching for patterns. It’s tracked Taliban insurgents and Somali pirates, and it’s gained traction within the private sector, including a well-publicized contract to help JPMorgan Chase detect fraud.

The two other firms that make up Team Themis, HBGary Federal and Berico Technologies, employ a roster of executives with extensive backgrounds in clandestine cyber-security work.

The president of HBGary Federal’s sister company, HBGary, is a legendary developer of “rootkits”—undetectable software that can be planted on a target computer for malicious purposes.

For a modest charge of $60,000, HBGary offered a rootkit designed in partnership with General Dynamics that could monitor keystrokes, delete files, and crash a computer infected with its proprietary code.  HBGary Federal—which, as the name suggests, is the government-sector wing of the company—attempted to sell contracts to the U.S. Air Force, among other clients.

There is no evidence that Team Themis is being paid directly by the Chamber.  It is working with the Chamber’s law firm, Huntoon & Willaims, not with the Chamber itself.  But its targers are critics of the U.S. Chamber of Commerce, such as U.S. Chamber Watch, which may have little influence as yet, but are seen by the Chamber as a major threat.

‘Seeing Like a State,’ the NSA and Big Data

 

I’ve long admired James C. Scott’s Seeing Like a State, which describes the history of the modern world as a history of governments collecting more and more information about the people and communities they ruled, and of how they mistake information for understanding, often with disastrous results.

Ancient and medieval kings and emperors collected tribute from the people they ruled, but they often knew little about them.  In order to more efficiently collect taxes, draft people into armies, mobilize economic resources and also carry out reforms, it was necessary for rulers to identify their subjects and collect basic information.

It is for that reason that there is a record of my name and address, my age and birthplace, the size and value of my house, the boundaries of my property, what kind of automobile I own, the amount and sources of my income and much else.  This has advantages in that this knowledge enables governments and corporations to provide me services that could not have been available in an earlier age, and provide them more efficiently.

As Scott pointed out, the problem is that the picture that governments have about their subjects (or, for that matter, corporations have about their employees and customers) represents a simplification of reality, and, when they act on that simplified information, trouble results.

The culminations of this process are the Surveillance State and corporate Big Data.  Government intelligence agencies will have information not only on what I own, where I go, what I earn and how I earn it, but details of my personal life from which inferences can be drawn about my tastes, thoughts and feelings.  Some of these inferences will be drawn by computer algorithms, like the one used select targets for flying killer drones in remote areas of Afghanistan, Pakistan and Yemen.

The power of intelligence agencies to gather information about individuals is greater than ever, and yet this information has not prevented the defeat of the U.S. military nor the growing appeal of Al Qaeda.  The date gathered by U.S. corporations about customers and employees is more extensive than ever, and yet this does not (so far as I can see) result in excellent customer service or excellent employee relations.  Misunderstandings about.  People are put on “no fly” lists for no apparent reason.  Banks foreclose on people with paid-up mortgages.

Knowledge is power and power corrupts.  But the worst corruption is the exercise of absolute power based on the illusion of knowledge.  What is needed is to reverse the polarity of surveillance—to make the inner workings of government and corporations at least as legible to the citizens as the other way around.

The world of surveillance, private and public

Senator Jay Rockefeller is rightly indignant that somebody has compiled lists of rape victims, which are sold to marketing companies for who-knows-what purpose.

But the fact is that we all provide information to private businesses that, when shared, enables them to know all about us. Short of never using a store discount card, never buying anything over the Internet and never using a credit card, there is no realistic way to get around it.

What I worry about is not so much what people in these companies know, or think they know, as what they do with the information. If the information is used by marketing companies to guess what products I might buy, this may be annoying, but it does me no great harm.

If it is turned over to lenders or employers and affects my chances of getting credit or a job, this would be a serious problem.  If it is turned over to government agencies to determine whether I am a potential terrorist or even a troublemaker, this would be an even more serious problem.

Knowledge is power, and there is a lack of balance of power. These people know, or think they know, a lot about me. I ought to be able to know who they are and what they know, or think they know, about me. If my life is an open book to them, I ought to be able to read that book.

So long as the information that companies and agencies have about me is secret, there is no penalty for wrongful derogatory information about me, and no incentive to double-check to make sure it is correct. All the incentives are to err on the side of suspicion.

The right to privacy only extends to individuals. Organizations and institutions should be transparent.