Melanie McCartney delves into the murky world of big data surveillance methods and the ramifications for our privacy.
WE LIVE IN A TIME where often it’s only a handful of seats that can win an election.
instead of election campaigns focusing on ideas and policies, the focus is now on influencing voters to win marginal seats.
Data mining and surveillance tools that were created for war zones before being picked up by the advertising industry are now in use against civilians.
Privacy laws and regulations haven’t kept up with the pace of technology and too many people are willing to exploit these loopholes. Business models that involve data mining like Cambridge Analytica and i360 can’t exist without tech giants such as Google, Facebook, and Palantir.
Everyone is scraping, selling and analysing our data; it’s now about influencing our behaviour, predicting what we will do next. The co-founder of Palantir, Peter Thiel, wasn’t joking when he named his company after the Lord of The Rings stone that is "far-seeing". Technocrats take advantage of the fact that most of us aren’t technologically minded, especially the politicians to which they pitch their technology. This is the main-game now to try to see into the future with predictive technology, including predicting crime.
Zuckerberg bragged at the time about how he was able to collect data willingly from people — he called them “dumb f***s”.
In this series, I will provide background about the tech giants and how they adopt each other's technologies, how data is being used against marginal voters with the debut of i360 software in Australia and what looms on the horizon.
Firstly, we will take a look at Palantir and its ties and similarities to Facebook.
What is Palantir and how is it connected to Facebook?
Peter Thiel co-founded PayPal and sold it to eBay in 2002. He took advantage of a post-9/11 world and created a company that used PayPal’s fraud-recognition software to stop terrorist attacks. This company became Palantir and was founded in 2003. Thiel provided the seed money for Facebook in 2004 and joined Facebook’s board; he is still on the board today. Mark Zuckerberg was only 19 years old at the time. Thiel soon became a mentor and friend. Zuckerberg bragged at the time about how he was able to collect data willingly from people — he called them “dumb f***s”.
Initially, the only investor of Palantir was Thiel but in 2005, the CIA’s venture arm In-Q-Tel also became an investor. The CIA was the only customer of Palantir for the next three years. Palantir had the best engineers testing and evaluating data mining software to predict terror attacks. Word soon got around and before long it was getting contracts from intelligence agencies, military units and eventually, police departments. Police departments in America were interested in using the technology for its ability to analyse big data to predict crime.
Predictive crime technology goes mainstream, including in Australia
Palantir secretly began using the population of New Orleans to test its predictive crime technology in 2012 by offering its services for free. In return, Palantir was given free data relating to public records, court filings, licenses, addresses, phone numbers, social media, and non-criminal data to train its software to forecast crime. Palantir’s prediction model also used an intelligence technique called social network analysis (SNA), which looks for connections between people, places, cars, weapons, addresses, social media posts and any data stored in databases.
This information is analysed and used to predict if people are most likely to be the victim of a crime or to commit a crime based on their connection to known victims or criminals.
It’s frighteningly close to the TV series Person of Interest, there are so many things that can go wrong here:
- a culture in law enforcement of guilt by association;
- the profiling of citizens being analysed by an algorithm with no human oversight; and
- privacy issues.
There is no clear evidence of predictive technology reducing crime in New Orleans. Despite this, it led to more contracts for Palantir, including foreign ones and with the Australian Government.
There are twenty-four patents relating to Palantir in Australia. The latest tender, signed last year, is worth $7.5 million and ends in June 2021. The Australian Criminal Intelligence Commission (ACIC) and different partner agencies use Palantir’s fusion software. ACICs different partner agencies, include the Australian Federal Police, the Australian Tax Office, the Department of Immigration and Border Protection and the Department of Human Services.
Home Affairs becomes an intelligence hub
Two bills were introduced in February this year, the Identity-matching Services Bill and the Australian Passports Amendment (Identity-matching Services) Bill. The first Bill authorises the Minister for Home Affairs Peter Dutton and his department to match photos against identities of citizens in various agencies without a warrant. Media reports about it say that the Government was basing this bill on an FBI model, but where did the Government get its model from? Palantir.
Home Affairs is now a huge data collection hub akin to Palantir and it will continue to grow as it collects information, each time a user makes a request for their identity-matching services. For example, say a bank supplies CCTV footage and data with a request, the footage and all of the data associated with the request will be hoovered up into the hub.
The reasons for Home Affairs using this information are very broad and include criminal intelligence gathering and profiling, community safety (an example is a person acting suspiciously in a crowded, public space), road safety and the policing of activist communities and protests. The second Bill allows Foreign Minister Julie Bishop to direct the automation of sharing passport data for the purposes of national security — meaning no human oversight, just algorithms talking to each other.
Palantir patents "crime forecasting" tech
Palantir is renowned for its secrecy — the only way to tease out what it's up to is to keep tabs on their patents.
The most recent patent that was granted was in December last year and it relates to "crime forecasting" technology, to help police 'to know when and where crimes are most likely to occur in the future'. Another pending patent involves the data mining and analysis of billions and trillions – or petabytes – of records. For context, one petabyte equates to 58,292 movies or 13.3 years of HDTV content. These could be bank transaction logs, call-data records, computer network access logs, email messages of corporations or other high-volume data.
The last patent pending involves household behaviour predictive software that analyses past behaviour in an effort to predict future behaviour. Examples detailed in the patent include obtaining data about household incomes, the number of cars belonging to a household and your household bills. There are similarities here with Facebook’s "Loyalty Prediction" tool and their "Household Audience" tool. These tools allow you to target entire families in the household.
The Intercept (14 April) obtained a confidential document that claims that Facebook is promoting their "Loyalty Prediction" ad service tool. It’s meant to assist advertisers chasing users that are on the verge of disengaging with brands. The Facebook business model has always been about data collection, but now it’s moving into territory where it is feeding data into a machine learning program called "FBLearner Flow". Algorithms predicting behaviour aren’t foolproof and, because companies are paying for these predictions, there’s a financial incentive to make sure that the predictions are correct. There is concern that Facebook could engineer results.
Facebook has been experimenting on its users for years, their mood experiment in 2014 proved that it could change people’s emotions dependent on the content they are shown. It would be tempting to employ this method to change people’s minds about disengaging with brands. GQ reported that Google has an application for a patent in the same realm regarding an algorithm to determine a user’s mood from a “plurality of data sources” — or big data.
This technology is basically social engineering and it could be used to sway voters. Next, we will take a quick look at the Cambridge Analytica debacle and then on to the debut of i360 software in the South Australian election.
The Cambridge Analytica debacle wasn’t really a data breach
In 2014, when Cambridge Analytica data mined Facebook, there was a way that anyone could do it using one of Facebook’s own tools. Facebook had a "reverse search tool", which enabled users to search for people on Facebook just by using a phone number or email address to find their profile. The feature was "opt out" for users and could only be turned off in the privacy settings. You could potentially feed the tool a list of phone numbers or email addresses that could have been taken from data breaches, online leaks, or even electoral rolls. It was only this month, after being warned for years by developers of the potential for using the tool to data mine, that it was finally shut down.
' ... and you have to wonder what else the ABS and the Government have been secretly up to'
When Facebook learned about what Aleksandr Kogan was doing for Cambridge Analytica in 2015, it actually paid him to do consultancy work. He was also asked to explain his technique for Cambridge Analytica and to give talks to Facebook staff about behavioural psychology. Another method that app developers like Kogan use to scrape Facebook data is by writing code inside their apps to capture your data.
It was a Palantir employee that gave Cambridge Analytica the idea to use an app to data mine Facebook. Many are extremely sceptical of Zuckerberg telling Congress recently that he is unaware of what Palantir does and if Palantir itself has ever scraped data from Facebook.
i360 makes its debut for the South Australian election
The South Australian (SA) Liberal Party lost the election in 2014 because it didn’t win enough seats to form government. Leader of the SA Liberal Party Steven Marshall and SA Liberal State Director Sascha Meldrum looked to America for solutions. In 2016, they purchased a product licence for i360 software; it costs around $25,000 per month.
They worked together with i360 to customise the software to include Australia’s compulsory and preferential voting system. They began using i360 in 2017, giving them a headstart in targeting marginal seats for the March 2018 election. Dozens of staff and volunteers including Victorian Opposition Leader Matthew Guy had a dry run using i360 in the field at the SA election. i360 has been credited for the SA win. The Victorian Liberals have been using it for a while, Queensland is about to sign up and the NSW Liberals and the Federal Government is thinking about it.
What is i360?
i360 was created by Michael Palmer in 2009 with a team of nine data scientists with PhDs behind him. It started out as a database to help Republicans catch up with Obama’s successful use of data during his 2008 campaign.
In 2011, the Koch brothers helped out by merging i360 with their non-profit organisation, Freedom Partners. Millions and millions have been pumped into it by the brothers and their rich, conservative allies ever since. Making it non-profit means that they can hide who the donors are as well as the money trail. i360 is no longer just a database — it offers a suite of cutting-edge tools, including analysis and predictive technology for political campaigning.
i360 is primarily funded by the Koch brothers. If you haven’t heard of the Koch brothers, they’re billionaires that fly under the political radar and have made a fossil fuel fortune. Their company, Koch Industries, owns and operates a massive network of oil and gas pipelines and it makes a wide range of products including chemicals, jet fuel, plastics and synthetic fertilisers. They have been funding climate change denialism to protect their interests through foundations, institutes and front groups for years, including an institute in Australia — the Institute of Public Affairs (IPA). (More about that in this link and this link.)
Next, we will look at how i360 works — their technology will sound a little familiar.
How does i360 work?
i360 has a multi-pronged approach that involves apps, technology, analytics and predictive analysis, as well as data science, digital marketing and advertising. Over the years, it has amassed trillions of data points on hundreds of millions of Americans. It uses thousands of unique pieces of data and combines data, analytics and predictive modelling for election campaigning. Facebook and Google Adwords work alongside i360 and are its featured digital partners.
i360 predictive technology
Advanced predictive modelling enables i360 software to test the effectiveness of thousands of political ads before using them. It uses segment models that go above and beyond demographics. It includes issues such as whether you want to "cut immigration numbers" or are in favour of "raising the minimum wage". These models can also help to predict which issue people care about the most. Knowing these details about people enables organisations to tailor their ads specifically for you; they personalise the style and tone of the ad, even the aesthetics and colours. These models get updated with new data and refreshed every night. This business model has much in common with Cambridge Analytica.
i360 tracks your online movements
According to the i360 website, it has partnered with a number of "mobile ID matching" services instead of using traditional cookies to track your online movements. Cookies are a small piece of data sent from websites that you browse back to your computer to help identify you so that it can advertise to you. This works for desktop and laptop computers but isn’t good for tracking your movements on mobile devices and for tracking your activities within apps that you visit.
i360 claims to use "mobile device IDs" which are used to identify you via your mobile devices and the apps that you visit. And then there is "direct matching", which matches up apps that you visit while being signed into Facebook or Google, enabling you to be found on multiple devices. According to its website, this service is 100 per cent accurate — no doubt because i360 is partners with Facebook and Google. In fact, the technology just sounds like a jargon-laden version of the ad-tracking tools that Facebook already offers. (More on this below.)
Facebook and Google tracks you too
The Facebook "Like" button has a small piece of code that tracks you as well as the Facebook share button when sharing online content. And then there’s Facebook Pixel. You only need to visit a page that has one of these buttons or Pixel code attached to it for it to collect data about you. What is Facebook Pixel, you may be asking?
Pixel allows you to track user movements offline with:
'A piece of code for your website that lets you measure, optimise and build audiences for your ad campaigns.'
It tracks your activities and reports it back to Facebook, and the code doesn’t expire like traditional cookies do. Advertisers can track what users are doing offline even if they don’t have a Facebook account — it’s called "shadow profiling". Google does this too, even if you don’t have a Gmail account — you only need to communicate with a Gmail address for one to be created.
Asher Wolf has recently written about how the Australian Bureau of Statistics (ABS), with an unnamed telco, did an experiment in 2016 to track people by using their mobile devices. It turned out that the telco was Telstra; the ABS said that their reasoning for the experiment was to estimate temporary populations and to assist policymakers. It’s not too far of a stretch to assume that this data has been entered into Palantir’s fusion software — and you have to wonder what else the ABS and the Government have been secretly up to in the last couple of years.
The ABS promptly released a statement claiming that the data that it used was non-identifiable — this argument is wrong. Technology has advanced far enough that it can re-identify anonymous people’s data.
i360 uses set-top box data to tailor ads
i360 has a TV targeting service which offers personalised TV ads that can be shown on any channel or program being watched. It has partnered with D2 Media (the image above is their business model) and is available in America through Dish Satellite TV or Direct TV.
Addressable Television business model (Source D2 Media).
These services and set-top boxes are also available in Australia. In 2015, Dish TV partnered up with Freeview as its manufacturer and it’s of note that Foxtel is currently in the process of transferring its cable service over to satellite. Are there plans afoot for this in Australia in the near future, or are they already in place?
Political parties, volunteers, contractors and representatives, exempt from Privacy Act
The last time that there was a Privacy Act review was ten years ago. It lasted more than two years and recommended hundreds of reforms. Many Australians don’t realise that registered political parties are specifically excluded from the definition of "organisation", so they’re exempt from the Privacy Act. Also exempt from it are political representatives (MPs and local government councillors), volunteers for political parties, and contractors and subcontractors of political parties. One of the recommendations was for these exemptions to be removed. Both major parties ignored the recommendation.
In light of recent events with Cambridge Analytica and Facebook, I think it’s time to remove the exemption. If political campaigning is no longer a public process and is done in private, being exempt from the Privacy Act is no longer acceptable.
Facebook's Mark Zuckerberg says he's sorry ... but is his apology enough? pic.twitter.com/15FXmzY8Wy
— AJ+ (@ajplus) April 12, 2018
The second part of this two-part study on data mining in Australia will explore the connection between i360, the Liberal Party and elections. Look out for it tomorrow.
You can read more from Melanie McCartney on her blog or you can follow her on Twitter @CartwheelPrint.
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Australia License
Support independent journalism Subscribe to IA.
