It’s time for third-party data brokers to emerge from the shadows

From The Conversation.

Facebook announced last week it would discontinue the partner programs that allow advertisers to use third-party data from companies such as Acxiom, Experian and Quantium to target users.

Graham Mudd, Facebook’s product marketing director, said in a statement:

We want to let advertisers know that we will be shutting down Partner Categories. This product enables third party data providers to offer their targeting directly on Facebook. While this is common industry practice, we believe this step, winding down over the next six months, will help improve people’s privacy on Facebook.

Few people seemed to notice, and that’s hardly surprising. These data brokers operate largely in the background.

The invisible industry worth billions

In 2014, one researcher described the entire industry as “largely invisible”. That’s no mean feat, given how much money is being made. Personal data has been dubbed the “new oil”, and data brokers are very efficient miners. In the 2018 fiscal year, Acxiom expects annual revenue of approximately US$945 million.

The data broker business model involves accumulating information about internet users (and non-users) and then selling it. As such, data brokers have highly detailed profiles on billions of individuals, comprising age, race, sex, weight, height, marital status, education level, politics, shopping habits, health issues, holiday plans, and more.

These profiles come not just from data you’ve shared, but from data shared by others, and from data that’s been inferred. In its 2014 report into the industry, the US Federal Trade Commission (FTC) showed how a single data broker had 3,000 “data segments” for nearly every US consumer.

Based on the interests inferred from this data, consumers are then placed in categories such as “dog owner” or “winter activity enthusiast”. However, some categories are potentially sensitive, including “expectant parent”, “diabetes interest” and “cholesterol focus”, or involve ethnicity, income and age. The FTC’s Jon Leibowitz described data brokers as the “unseen cyberazzi who collect information on all of us”.

In Australia, Facebook launched the Partner Categories program in 2015. Its aim was to “reach people based on what they do and buy offline”. This includes demographic and behavioural data, such as purchase history and home ownership status, which might come from public records, loyalty card programs or surveys. In other words, Partner Categories enables advertisers to use data brokers to reach specific audiences. This is particularly useful for companies that don’t have their own customer databases.

A growing concern

Third party access to personal data is causing increasing concern. This week, Grindr was shown to be revealing its users’ HIV status to third parties. Such news is unsettling, as if there are corporate eavesdroppers on even our most intimate online engagements.

The recent Cambridge Analytica furore stemmed from third parties. Indeed, apps created by third parties have proved particularly problematic for Facebook. From 2007 to 2014, Facebook encouraged external developers to create apps for users to add content, play games, share photos, and so on.

Facebook then gave the app developers wide-ranging access to user data, and to users’ friends’ data. The data shared might include details of schooling, favourite books and movies, or political and religious affiliations.

As one group of privacy researchers noted in 2011, this process, “which nearly invisibly shares not just a user’s, but a user’s friends’ information with third parties, clearly violates standard norms of information flow”.

With the Partner Categories program, the buying, selling and aggregation of user data may be largely hidden, but is it unethical? The fact that Facebook has moved to stop the arrangement suggests that it might be.

More transparency and more respect for users

To date, there has been insufficient transparency, insufficient fairness and insufficient respect for user consent. This applies to Facebook, but also to app developers, and to Acxiom, Experian, Quantium and other data brokers.

Users might have clicked “agree” to terms and conditions that contained a clause ostensibly authorising such sharing of data. However, it’s hard to construe this type of consent as morally justifying.

In Australia, new laws are needed. Data flows in complex and unpredictable ways online, and legislation ought to provide, under threat of significant penalties, that companies (and others) must abide by reasonable principles of fairness and transparency when they deal with personal information. Further, such legislation can help specify what sort of consent is required, and in which contexts. Currently, the Privacy Act doesn’t go far enough, and is too rarely invoked.

In its 2014 report, the US Federal Trade Commission called for laws that enabled consumers to learn about the existence and activities of data brokers. That should be a starting point for Australia too: consumers ought to have reasonable access to information held by these entities.

Time to regulate

Having resisted regulation since 2004, Mark Zuckerberg has finally conceded that Facebook should be regulated – and advocated for laws mandating transparency for online advertising.

Historically, Facebook has made a point of dedicating itself to openness, but Facebook itself has often operated with a distinct lack of openness and transparency. Data brokers have been even worse.

Facebook’s motto used to be “Move fast and break things”. Now Facebook, data brokers and other third parties need to work with lawmakers to move fast and fix things.

Author: Sacha Molitorisz, Postdoctoral Research Fellow, Centre for Media Transition, Faculty of Law, University of Technology Sydney

The Battle Of The Mobile Wallet

Juniper Research has just released a report “NFC Vs QR Codes ~ Which Wallet Wins?”

They estimate that, by 2019, nearly 2.1 billion consumers worldwide will use a mobile wallet to make a payment or send money, up by nearly 30% on the 1.6 billion recorded at the end of 2017.  The emergence of several high profile mobile payment services, including Apple Pay, Samsung Pay and Google Pay, has provided the sector with fresh impetus.

Furthermore, the accounts (or wallets) used to store consumer credentials are now have an integration of offline and online payments, enabling users to access them both for remote purchases and instore.

But there are significant regional variations in the mechanisms to make contactless mobile payments. In some countries mobile wallets win out, whereas elsewhere the NFC payment card wins. In addition Host card emulation (HCE) is on the rise, the software architecture that provides exact virtual representation of various electronic identity (access, transit and banking) cards using only software. Prior to the HCE architecture, NFC transactions were mainly carried out using secure elements, such as the chip on a card or other means.

HCE enables mobile applications running on supported operating systems to offer payment card and access card solutions independently of third parties while leveraging cryptographic processes traditionally used by hardware-based secure elements without the need for a physical secure element. This technology enables the merchants to offer payment cards solutions more easily through mobile closed-loop contactless payment solutions, offers real-time distribution of payment cards and, more tactically, allows for an easy deployment scenario that does not require changes to the software inside payment terminals.

When we compare the relative share of contactless cards and wallets in key markets outside the US (Europe, Canada and Australia), we see that, typically, cards account for well over 90% of transactions by value (rising to 98% in Spain and Canada). In the US, the positions are reversed, with mobile wallets accounting for 87% of the total.

While many markets focus on enabling instore mobile payments via NFC (which uses the same infrastructure and technology as contactless cards), a small number have embraced QR code-based instore payments. While precise mechanisms vary, typically the consumer is presented with a printed QR code, after which he/she launches the payment app and scans the code with the smartphone camera. This directs them to a payment page, where the transaction amount is entered and the transaction is made.

By far the most successful deployments of QR code-based payments have come in China, where these have already surpassed cash and cards in both instore transaction volume and values. Deployments elsewhere are sporadic, but the mechanism has been a mainstay of Scandinavian wallets for several years and is also gaining traction in India.

However, a study by researchers at the System Security Lab at the Chinese University of Hong Kong’s Department of Information found that it was possible to gain access to the phone’s camera to record an image of a QR code.

Furthermore, as QR codes can contain any kind of data (not just payment/transaction details), it is possibly to create codes containing links to malware or phishing sites.

As a result, the People’s Bank of China confirmed in December 2017 that it would be introducing plans to regulate payments by QR codes and other scannable codes. The new regulations, which come into effect in April 2018, will include a payments cap of RMB500 ($79) for basic payments, rising to RMB5,000 ($790) if additional security procedures are implemented, such as tokenisation, risk monitoring and anti-counterfeit measures.

Outside China, NFC has long been the proximity payment mechanism of choice by mobile wallet providers, although the initial model whereby the SE was based on the SIM has largely been jettisoned in favour of alternatives, where the SE is either embedded in the handset or else virtualised using HCE.

The evolution of offline payment in the US has lagged behind that in other developed markets, with EMV only mandated from October 2015. After that point, if merchants had not introduced processing systems to facilitate chip-based payments, then liability for fraud would pass from the card providers to those merchants.

Even with the onset of EMV, banks were reluctant to move to Chip & PIN, apparently concerned that their customers would be unable to remember a 4-digit PIN. Hence, US customers now use Chip & signature instead of the more secure alternative.

This means that Apple Pay and the wallets that followed in its wake, have the opportunity to establish themselves as the contactless mechanisms of choice.

The challenge facing Apple and its rivals is to ensure that the infrastructure is in place for consumers to make instore payments. According to Head of Apple Pay Jennifer Bailey, when Apple Pay first launched in September 2014, it was supported by just 3% of retailers, a figure that had risen only marginally by the end of that year. However, by the end of 2017, half of US retailers supported the mechanism, indicative of the progress that contactless has made in that market.

Nevertheless, although a majority of the remaining US retailers are now believed to own POS terminals capable of fulfilling contactless transactions, a significant number have not yet activated the technology. Furthermore, in some stores only a minority of terminals accept the technology: Juniper estimates that just under 30% of all POS terminals in the US were capable of processing contactless transactions by the end of 2017.

Purely from a payments and convenience perspective, it will be difficult for mobile wallet providers to gain market share from contactless cards. It is therefore incumbent upon them to deliver services through which the mobile wallet will become the default payment mechanism.

We would argue that there are at least 2 means by which this could potentially be achieved:

  • Offering an integrated wallet which can be used on both offline and online environments;
  • Offering services based around loyalty.

HCE threatens the central role of the network operator in NFC’s value chain, it strengthens that of the bank and makes handset-based contactless payment a more attractive proposition.

Banks have increasingly understood this. By the end of 2014, Juniper Research estimates that just 7 banks had introduced commercial services based on HCE. By mid January 2016, that number had increased to 55; by the end of 2017, Juniper Research estimates that well over 200 banks had introduced such services. Those launching in 2017 included Belfius (Belgium), Citi (US), Credit Agricole (France), Deutsche Bank (Germany), Rabobank (Netherlands) and SBI (India).

A number of banking collectives have also sought to implement HCE. In June 2016, the Danish banking collective, the BOKIS partnership, launched an HCE wallet utising a solution provided by Nordic digital payments specialist, Nets. The BOKIS partnership includes 62 banks that form the small to mid-sized banks segment of the Association of Local Banks, Savings Banks and Cooperative Banks in Denmark, together with 5 Danish regional banks: Jyske Bank, Sydbank, Spar Nord Bank, Arbejdernes Landsbank and Nykredit Bank. Meanwhile, In October 2016, 27 Spanish banks teamed up to launch a new mobile payment platform called Bizum whicih utilises HCE.

However, despite this plethora of bank launches, adoption has been relatively modest: many services have only a few tens of thousands of users, with none yet reporting that they have achieved more than a million. The scale of the challenge facing the banks is largely tied to that facing NFC in general: in Western Europe; banks’ own contactless services are up against both contactless cards and the OEM-Pays, making it extremely difficult to gain a foothold.

Charting The Financial Services Revolution

I caught up with Glenn Hodgeman, the brains behind the upcoming AltFi Australasia Summit  2018 to be held in Sydney on the 16th April at Doltone House Jones Bay Wharf.

This is the third annual event and is designed to bring various industry players, private equity, venture capital, innovators and regulators together to share insights at the inflection point of the fintech revolution as it moves “from marginal into the mainstream”.

The revolution underway is partly being driven by new innovative players and platform providers who can move quickly, without legacy, whilst larger more established players wrestle with legacy systems and culture, yet some are now beginning to see the potential. The potential opportunity is significant, not just paving the cowpaths, but to create totally new business models and new customer value propositions.

Glenn believes the large incumbents will increasing be focussing on “big corporate” borrowers, which creates space for small fleet of foot players to address in particular lending in the consumer and small business sectors.  Of course there are also a myriad of cashed up investors seeking to get footholds into the opportunity stack

AltFi have strong connections with London, and they believe Australia is currently perhaps 4-5 years behind the leading edge there. This creates opportunity to learn from events overseas, as well as from New Zealand, Israel and local success stories.

Glenn was keen to underscore the fact that the conference is not a “scatter gun” of concepts, from the alphabet soup which is Fintech, but rather he wants to drill into a small number of high potential critical areas, from lending, payments and robo advice.

Topics scheduled include global case studies in alternative finance, the thought leaders in the Australian Banking and Finance Industry, Digital Mortgage lending, Microfinance, alternative SME lending and point of sale credit.

This is rich menu, and the event is likely to be well frequented.

You can get 20% off the conference price by using this link, and the promotional code DigitalFinanceAnalytics.

I get nothing from this, but it does offer some additional benefit to DFA Blog readers! I may see you there.

Courting co-owners to buy a house online may be riskier than it looks

From The Conversation.

Digital platforms are offering people who couldn’t afford a house on their own, the opportunity to divvy up the costs with others. But co-ownership of real estate can be a risky and potentially costly business.

In an environment of high residential prices where families are becoming a smaller proportion of households, and permanent relationships are giving way to transient and more distant connections, digital platforms for co-ownership are filling an emerging need.

Co-ownership permits a whole range of sharing options. For example, it allows an occupant to be a part owner in their own home even if they cannot afford to buy the whole thing. It also permits an investor to take a smaller position as a promise of a better relationship with the tenant.

However, buying or selling a property involves legal, financial, statutory and agency costs that mean that even moving across the road can cost about half a year’s income. This means that you need to be sure of what you are doing and reasonably confident that you will not be changing your mind about your investment too quickly.

Digital platforms like Kohab are using the legal relationship known as “tenancy in common” to facilitate co-ownership. It permits the separate parties to have a defined share of the house and to transfer their interests independently.

But this still presents considerable practical risks. Someone wanting to sell their share of a house is likely to find a limited market of other people willing to take over the part ownership, and they are likely to have a weaker negotiating position in selling.

The remaining co-owners of a house also have no control over who the incoming partner will be. This may limit their preferences in relation to how the property should be managed. It can make remaining with the investment uncomfortable and lead to even more turnover of ownership and the prospects of sale at a discount.

How co-ownership has changed over the years

Shared real estate ownership has been evolving for some time. The strata title system was introduced over 50 years ago when it became necessary for single buildings to be owned by multiple people.

Company title co-ownership was unpopular because the co-owners did not always agree on how to manage their property. It was also unpopular because sale of part-interests was difficult and often settled at a discount to true value.

However, it overcame the problems of management and resale that dogged the earlier company title. This approach used company shares to split ownership of a property between several investors.

Australia was later an innovator in the development of property trusts that applied the company model, but with someone to manage the the shared ownership of complex properties.

These property trusts have blossomed as an investment and are now commonly known by their US name – real estate investment trusts. These trusts usually focus on commercial buildings where they provide a vehicle for small investors to access property investment in major real estate assets.

Where digital platforms come in

Online applications such as DomaCom or BrickX have brought the trust model online and applied it to smaller properties that are not usually the target of these traditional trusts. BrickX for example divides its selected investment properties into 10,000 “bricks” and allows investors to buy bricks.

This permits up to 10,000 owners for an individual property. It also allows small investors to spread their funds across multiple properties to control their risks. DomaCom follows a different strategy to achieve the same goal of allowing a large number of investors to be involved in individual managed investment properties.

Then there’s new apps like Kohab. Its point of difference is that it operates on a smaller and perhaps more intimate scale.

It does not rely on a crowdfunding approach, but uses its online platform to connect owner/occupiers and investors for the purpose of co-ownership. It does not produce real estate investment trusts, but does facilitate the co-ownership of individual dwellings by more than one owner.

In capital cities it is getting harder for individuals and families to afford properties. Co-ownership with others, either as shared occupants, or distant investors, is one way to cross the rent/buy gap. But it’s not without risk

Author: Garrick Small, Associate Professor, CQUniversity Australia

Would A Loan Comparison Tool Compete With Brokers?

From The Adviser.

A loan comparison tool proposed by the Productivity Commission could compete with the broker channel, according to six of Australia’s largest non-major banks.

In a joint submission to the Productivity Commission (PC), AMP, the Bank of Queensland, Suncorp, Bendigo Bank, MyState and ME Bank warned that an online loan comparison tool could undermine the broking industry.

In its draft report, the PC called for the Australian Prudential Regulation Authority (APRA) to collect interest rate and fee data and use it to determine a median rate that would be published via an online tool.

The PC claimed that such a reform could help increase transparency for customers and enhance competition.

The non-majors claimed that a proposed comparison tool with the “authority of a government agency” could undermine the broker channel.

“[The] online tool would, in some respects, compete with the broker channel, particularly given the proposal is for the comparison tool to have the authority of a government agency standing behind it,” the banks stated.

“Such an approach could potentially undermine the broker industry and eventually favour the banks with larger bricks and mortar networks,” the banks added.

Further, the lenders argued that the publication of a median interest rate could “mislead customers”.

“While this has the potential to improve competitive pressure from the demand side of the market, it may also involve considerable practical difficulties,” the submission read.

“More importantly, it may mislead customers as to the true cost of a product. The main problem with such tools is that they have a tendency to lead to ‘gaming’, whereby suppliers develop products that rate well on the tool but have shortcomings in other areas.

“For example, comparison tools have difficulty capturing the full benefits of a ‘bundle’ of services offered by a financial institution.”

The banks claimed that the tool could also create an incentive for some lenders to “shift costs” to products and services outside the tool’s scope.

“They also provide an incentive for suppliers to increase costs for services outside the scope of required disclosures. For example, in the case of mortgages, suppliers could shift costs to account closing or switching fees,” the submission said.

Additionally, the banks claimed that the tool could instigate a “race to the bottom”, with lenders creating products that “fall short of expectations”, potentially requiring regulatory intervention.

The lenders said: “[Some] financial institutions may respond by choosing not to offer services outside what the tool requires, and consumers could end up with products that fall short of expectations.

“Such an approach could see suppliers in a race to the bottom, offering only the most basic and feature-free products in order to present the most attractive median interest rates to the comparison tool.

“This would then inevitably result in additional regulatory interventions as governments attempt to patch over the shortcomings of the tool.”

Broker remuneration

Moreover, the banks advised against changes to the broker remuneration model, claiming that “consumers have a strong tendency to resist paying for services”.

The lenders added that “disruption” to the broking industry’s remuneration model could have a “material” impact on market competition.

“A significant disruption to the economic viability of the broker industry would be a material competitive neutrality issue for smaller banks.”

“Disclosure of mortgage broker ownership is a priority”

In their submission, the banks also expressed support for the PC’s call for increased disclosure for mortgage brokers.

The non-majors noted that they believe customers should “know the identity of the broker’s owner”, and they claimed that the level of business activity directed to an aggregator’s owner or associated company should also be published.

“[We] believe it is important to ensure that the customers of mortgage brokers know the identity of the broker’s owner so they can factor this information into their decision-making process.

“In addition to ownership disclosure, [we] recommend that broker networks and aggregators publish information showing the amount of business directed towards their owners or associated companies, relative to the proportion directed elsewhere.”

Australia’s digital divide is not going away

From The Conversation.

Despite large investments in the National Broadband Network, the “digital divide” in Australia remains largely unchanged, according to a new report from the Australian Bureau of Statistics.

The Australian Household Use of Information Technology report says we are doing more online, and we are using an increasing number of connected devices. Our homes are more connected.

However, the number of people using the internet is not growing, and the basic parameters of digital inequality in Australia – age, geography, education and income – continue to define access to and uses of online resources.

Almost 2.6 million Australians, according to these ABS figures, do not use the internet. Nearly 1.3 million households are not connected. So what is going on? The ABS data points to the complexity of the social and economic issues involved, but it also helps us identify the key areas of concern.

Who’s missing out

Age is a critical factor. While more than nine in ten people aged between 15 to 54 are internet users, the number drops to eight in ten of those aged 55-64 years, and to under six in ten of those over 65 years.

Most people with jobs (95.1%) are online, compared to just 72.5% of those not employed. Migrants from non-English speaking countries are less connected (81.6%) than those Australian born (87.6%). Those already at a disadvantage – the very people who have the most to gain from all the extraordinary resources of the internet – are missing out.

This is not to say that it is only individuals that will benefit from greater digital inclusion. Raising the level of digital inclusion yields direct benefits for the community, government and business. There are, for instance, clear efficiency gains for government moving services online.

Raising the level of online health engagement for those over 65 years of age (the heaviest users of health care) would provide such a benefit. Currently, just over one in five people in this age category access online health services, substantially below the national average of two in five.

But nor should we focus only on the economic and efficiency gains of inclusion: the social benefits of connection and access to entertainment and information are considerable for most internet users, and especially so for those who are isolated and lonely, as older people may be.

Income and affordability matter

Australians with higher incomes are substantially more likely to have internet access at home than those with lower incomes – 96.9% of the highest quintile (bracket representing one fifth of the sample) income households have access, whereas only 67.4% of the lowest quintile have access.

And better-off Australians appear to be doing more online. Compared to the general population their uses of online banking and shopping, education and health services are higher. They are connected to the internet with multiple devices, with an average of 7.2 devices at home, compared to 4.4 in the lowest income quintile.

The gap between the major cities and the bush has not narrowed over time – 87.9% of those living in major cities have internet access at home, 82.7% in inner regional, 80.7% in outer regional and 77.1% in remote areas. It’s important to note that this survey did not include remote Indigenous communities, where the evidence suggests that internet access is usually very poor.

Among those who are connected, geographical differences in the means of access and modes of engagement with online services suggest a further gap among those who are already disadvantaged. People in remote areas use the internet much less for entertainment and formal education compared to their urban counterparts, which are services that require more bandwidth and better quality connections.

Unfortunately, the ABS did not ask why households do not have home internet access, as it did in 2014-15. That data revealed cost was a factor keeping 198,600 households offline. Unsurprisingly, 148,200 of these households were from the two lowest income quintiles. Cost was the major factor in keeping more than 30,000 of the 76,000 family households (with children under 15) offline.

Given the increasingly central role of the internet in educational activities, the fact that the number of family households without access has not fallen since 2014-15 is concerning.

Affordability will continue to be a problem as more data-intensive services are offered online and the demand for data increases, and as mobile services become increasingly important.

However, cost was not the only reason people gave for non-use. Around 200,000 of the two lowest income households lacked knowledge or confidence to use the internet. Digital ability, and our readiness to make use of the internet, are clearly areas for continuing attention. We know that interventions there can make a difference.

The final survey on household use of IT

This ABS survey is the last of its kind. We hope the Bureau will be able to undertake further surveys in this area. The end of this data series does not signal its lack of relevance, at a time when digital inclusion is more important than ever. On the contrary, it points to a pressing new challenge for governments, the community, and business.

As our service economy increasingly moves online — in education, health, work, and government services — we need to ensure that all Australians, particularly those already disadvantaged, have affordable access to the online world. A reliable evidence base to inform our work in this area is essential.

But the information we have should be enough to spark action in some critical areas. The affordability of broadband is clearly one of these. When we consider, for example, the situation of families with children — where cost is clearly an issue for a significant number of them — we need to recognise that existing policy settings and market mechanisms are not working.

The digital divide is likely to grow

The ABS findings correspond to other recent work in the area. Australian policy has long had the aim of making communications widely accessible across our huge country and dispersed, fairly small population.

But the Australian Digital Inclusion Index has highlighted the problem of affordability and unequal access across economic, social and spatial lines. Australia’s performance also compares poorly to other countries.

The Inclusive Internet Index, produced by The Economist’s Intelligence Unit, rates Australia at 25 out of 86 countries, behind Russia and Hungary.

So despite the egalitarian aspirations embodied in the policy language of the National Broadband Network, the evidence suggests that the Australian internet remains unusually unequal in terms of access and affordability.

Instead of a digital economy designed for everyone, we appear to have created a highly stratified internet, where the distribution of resources and opportunities online reflects Australia’s larger social and economic inequalities. The risk is that over time the digital divide will amplify these. Unfortunately there is little indication in the ABS data that any of the key indicators will change soon.

Authors: Julian Thomas, Director, Social Change Enabling Capability Platform, RMIT University; Chris K Wilson, Research Fellow, Technology, Communication and Policy Lab – Digital Ethnography Research Centre, RMIT University; Sora Park, Director, News & Media Research Centre, University of Canberra

How Cambridge Analytica’s Facebook targeting model really worked

From The Conversation.

The researcher whose work is at the center of the Facebook-Cambridge Analytica data analysis and political advertising uproar has revealed that his method worked much like the one Netflix uses to recommend movies.

In an email to me, Cambridge University scholar Aleksandr Kogan explained how his statistical model processed Facebook data for Cambridge Analytica. The accuracy he claims suggests it works about as well as established voter-targeting methods based on demographics like race, age and gender.

If confirmed, Kogan’s account would mean the digital modeling Cambridge Analytica used was hardly the virtual crystal balla few have claimed. Yet the numbers Kogan provides also show what is – and isn’t – actually possible by combining personal datawith machine learning for political ends.

Regarding one key public concern, though, Kogan’s numbers suggest that information on users’ personalities or “psychographics” was just a modest part of how the model targeted citizens. It was not a personality model strictly speaking, but rather one that boiled down demographics, social influences, personality and everything else into a big correlated lump. This soak-up-all-the-correlation-and-call-it-personality approach seems to have created a valuable campaign tool, even if the product being sold wasn’t quite as it was billed.

The promise of personality targeting

In the wake of the revelations that Trump campaign consultants Cambridge Analytica used data from 50 million Facebook users to target digital political advertising during the 2016 U.S. presidential election, Facebook has lost billions in stock market value, governments on both sides of the Atlantic have opened investigations, and a nascent social movement is calling on users to #DeleteFacebook.

But a key question has remained unanswered: Was Cambridge Analytica really able to effectively target campaign messages to citizens based on their personality characteristics – or even their “inner demons,” as a company whistleblower alleged?

If anyone would know what Cambridge Analytica did with its massive trove of Facebook data, it would be Aleksandr Kogan and Joseph Chancellor. It was their startup Global Science Research that collected profile information from 270,000 Facebook users and tens of millions of their friends using a personality test app called “thisisyourdigitallife.”

Part of my own research focuses on understanding machine learning methods, and my forthcoming book discusses how digital firms use recommendation models to build audiences. I had a hunch about how Kogan and Chancellor’s model worked.

So I emailed Kogan to ask. Kogan is still a researcher at Cambridge University; his collaborator Chancellor now works at Facebook. In a remarkable display of academic courtesy, Kogan answered.

His response requires some unpacking, and some background.

From the Netflix Prize to “psychometrics”

Back in 2006, when it was still a DVD-by-mail company, Netflix offered a reward of $1 million to anyone who developed a better way to make predictions about users’ movie rankings than the company already had. A surprise top competitor was an independent software developer using the pseudonym Simon Funk, whose basic approach was ultimately incorporated into all the top teams’ entries. Funk adapted a technique called “singular value decomposition,” condensing users’ ratings of movies into a series of factors or components – essentially a set of inferred categories, ranked by importance. As Funk explained in a blog post,

“So, for instance, a category might represent action movies, with movies with a lot of action at the top, and slow movies at the bottom, and correspondingly users who like action movies at the top, and those who prefer slow movies at the bottom.”

Factors are artificial categories, which are not always like the kind of categories humans would come up with. The most important factor in Funk’s early Netflix model was defined by users who loved films like “Pearl Harbor” and “The Wedding Planner” while also hating movies like “Lost in Translation” or “Eternal Sunshine of the Spotless Mind.” His model showed how machine learning can find correlations among groups of people, and groups of movies, that humans themselves would never spot.

Funk’s general approach used the 50 or 100 most important factors for both users and movies to make a decent guess at how every user would rate every movie. This method, often called dimensionality reduction or matrix factorization, was not new. Political science researchers had shown that similar techniques using roll-call vote data could predict the votes of members of Congress with 90 percent accuracy. In psychology the “Big Five” model had also been used to predict behavior by clustering together personality questions that tended to be answered similarly.

Still, Funk’s model was a big advance: It allowed the technique to work well with huge data sets, even those with lots of missing data – like the Netflix dataset, where a typical user rated only few dozen films out of the thousands in the company’s library. More than a decade after the Netflix Prize contest ended, SVD-based methods, or related models for implicit data, are still the tool of choice for many websites to predict what users will read, watch, or buy.

These models can predict other things, too.

Facebook knows if you are a Republican

In 2013, Cambridge University researchers Michal Kosinski, David Stillwell and Thore Graepel published an article on the predictive power of Facebook data, using information gathered through an online personality test. Their initial analysis was nearly identical to that used on the Netflix Prize, using SVD to categorize both users and things they “liked” into the top 100 factors.

The paper showed that a factor model made with users’ Facebook “likes” alone was 95 percent accurate at distinguishing between black and white respondents, 93 percent accurate at distinguishing men from women, and 88 percent accurate at distinguishing people who identified as gay men from men who identified as straight. It could even correctly distinguish Republicans from Democrats 85 percent of the time. It was also useful, though not as accurate, for predicting users’ scores on the “Big Five” personality test.

There was public outcryin response; within weeks Facebook had made users’ likes private by default.

Kogan and Chancellor, also Cambridge University researchers at the time, were starting to use Facebook data for election targeting as part of a collaboration with Cambridge Analytica’s parent firm SCL. Kogan invited Kosinski and Stillwell to join his project, but it didn’t work out. Kosinski reportedly suspected Kogan and Chancellor might have reverse-engineered the Facebook “likes” model for Cambridge Analytica. Kogan denied this, saying his project “built all our models using our own data, collected using our own software.”

What did Kogan and Chancellor actually do?

As I followed the developments in the story, it became clear Kogan and Chancellor had indeed collected plenty of their own data through the thisisyourdigitallife app. They certainly could have built a predictive SVD model like that featured in Kosinski and Stillwell’s published research.

So I emailed Kogan to ask if that was what he had done. Somewhat to my surprise, he wrote back.

“We didn’t exactly use SVD,” he wrote, noting that SVD can struggle when some users have many more “likes” than others. Instead, Kogan explained, “The technique was something we actually developed ourselves … It’s not something that is in the public domain.” Without going into details, Kogan described their method as “a multi-step co-occurrence approach.”

However, his message went on to confirm that his approach was indeed similar to SVD or other matrix factorization methods, like in the Netflix Prize competition, and the Kosinki-Stillwell-Graepel Facebook model. Dimensionality reduction of Facebook data was the core of his model.

How accurate was it?

Kogan suggested the exact model used doesn’t matter much, though – what matters is the accuracy of its predictions. According to Kogan, the “correlation between predicted and actual scores … was around [30 percent] for all the personality dimensions.” By comparison, a person’s previous Big Five scores are about 70 to 80 percent accurate in predicting their scores when they retake the test.

Kogan’s accuracy claims cannot be independently verified, of course. And anyone in the midst of such a high-profile scandal might have incentive to understate his or her contribution. In his appearance on CNN, Kogan explained to a increasingly incredulous Anderson Cooper that, in fact, the models had actually not worked very well.

In fact, the accuracy Kogan claims seems a bit low, but plausible. Kosinski, Stillwell and Graepel reported comparable or slightly better results, as have several other academic studies using digital footprints to predict personality (though some of those studies had more data than just Facebook “likes”). It is surprising that Kogan and Chancellor would go to the trouble of designing their own proprietary model if off-the-shelf solutions would seem to be just as accurate.

Importantly, though, the model’s accuracy on personality scores allows comparisons of Kogan’s results with other research. Published models with equivalent accuracy in predicting personality are all much more accurate at guessing demographics and political variables.

For instance, the similar Kosinski-Stillwell-Graepel SVD model was 85 percent accurate in guessing party affiliation, even without using any profile information other than likes. Kogan’s model had similar or better accuracy. Adding even a small amount of information about friends or users’ demographics would likely boost this accuracy above 90 percent. Guesses about gender, race, sexual orientation and other characteristics would probably be more than 90 percent accurate too.

Critically, these guesses would be especially good for the most active Facebook users – the people the model was primarily used to target. Users with less activity to analyze are likely not on Facebook much anyway.

When psychographics is mostly demographics

Knowing how the model is built helps explain Cambridge Analytica’s apparently contradictory statements about the role – or lack thereof – that personality profiling and psychographics played in its modeling. They’re all technically consistent with what Kogan describes.

A model like Kogan’s would give estimates for every variable available on any group of users. That means it would automatically estimate the Big Five personality scores for every voter. But these personality scores are the output of the model, not the input. All the model knows is that certain Facebook likes, and certain users, tend to be grouped together.

With this model, Cambridge Analytica could say that it was identifying people with low openness to experience and high neuroticism. But the same model, with the exact same predictions for every user, could just as accurately claim to be identifying less educated older Republican men.

Kogan’s information also helps clarify the confusion about whether Cambridge Analytica actually deleted its trove of Facebook data, when models built from the data seem to still be circulating, and even being developed further.

The whole point of a dimension reduction model is to mathematically represent the data in simpler form. It’s as if Cambridge Analytica took a very high-resolution photograph, resized it to be smaller, and then deleted the original. The photo still exists – and as long as Cambridge Analytica’s models exist, the data effectively does too.

Author: Matthew Hindman, Associate Professor of Media and Public Affairs, George Washington University

Why it’s so hard to Delete Facebook: Constant psychological boosts keep you hooked

From The Conversation.

Here we go again: another Facebook controversy, yet again violating our sense of privacy by letting others harvest our personal information. This flareup is a big one to be sure, leading some people to consider leaving Facebook altogether, but the company and most of its over 2 billion users will reconcile. The vast majority will return to Facebook, just like they did the last time and the many times before that. As in all abusive relationships, users have a psychological dependence that keeps them hooked despite knowing that, at some level, it’s not good for them.

Decades of research has shown that our relationship with all media, whether movies, television or radio, is symbiotic: People like them because of the gratifications they get from consuming them – benefits like escapism, relaxation and companionship. The more people use them, the more gratifications they seek and obtain.

With online media, however, a consumer’s use provides data to media companies so they can serve up exactly what would gratify her most, as they mine her behavior patterns to tailor her online experiences and appeal to her individual psychological needs.

Aside from providing content for our consumption, Facebook, Twitter, Google – indeed all interactive media – provide us with new possibilities for interaction on the platform that can satisfy some of our innate human cravings.

Interactive tools in Facebook provide simplified ways to engage your curiosity, broadcast your thoughts, promote your image, maintain relationships and fulfill the yearning for external validation. Social media take advantage of common psychological traits and tendencies to keep you clicking – and revealing more of yourself. Here’s why it’s so hard, as a social network user, to pull the plug once and for all.

Buoying your ‘friend’ships

The more you click, the stronger your online relationships. Hitting the ‘Like’ button, commenting on photos of friends, sending birthday wishes and tagging others are just some of the ways in which Facebook allows you to engage in “social grooming.” All these tiny, fleeting contacts help users maintain relationships with large numbers of people with relative ease.

Molding the image you want to project

The more you reveal, the greater your chances of successful self-presentation. Studies have shown that strategic self-presentation is a key feature of Facebook use. Users shape their online identity by revealing which concert they went to and with whom, which causes they support, which rallies they attend and so on. In this way, you can curate your online self and manage others’ impressions of you, something that would be impossible to do in real life with such regularity and precision. Online, you get to project the ideal version of yourself all the time.

Snooping through an open window

The more you click, the more you can keep an eye on others. This kind of social searching and surveillance are among the most important gratifications obtained from Facebook. Most people take pleasure in looking up others on social media, often surreptitiously. The psychological need to monitor your environment is deep-rooted and drives you to keep up with news of the day – and fall victim to FOMO, the fear of missing out. Even privacy-minded senior citizens, loathe to reveal too much about themselves, are known to use Facebook to snoop on others.

Enhancing your social resources

The more you reveal, the greater your social net worth. Being more forthcoming can get you a job via LinkedIn. It can also help an old classmate find you and reconnect. Studies have shown that active use of Facebook can enhance your social capital, whether you’re a college student or a senior citizen wanting to bond with family members or rekindle ties with long-lost friends. Being active on social media is associated with increases in self-esteem and subjective well-being.

Enlarging your tribe

The more you click, the bigger and better the bandwagon. When you click to share a news story on social media or express approval of a product or service, you’re contributing to the creation of a bandwagon of support. Metrics conveying strong bandwagon support, just like five stars for a product on Amazon, are quite persuasive, in part because they represent a consensus among many opinions. In this way, you get to be a part of online communities that form around ideas, events, movements, stories and products – which can ultimately enhance your sense of belonging.

Expressing yourself and being validated

The more you reveal, the greater your agency. Whether it’s a tweet, a status update or a detailed blog post, you get to express yourself and help shape the discourse on social media. This self-expression by itself can be quite empowering. And metrics indicating bandwagon support for your posts – all those “likes” and smiley faces – can profoundly enhance your sense of self worth by appealing to your ingrained psychological need for external validation.

In all these ways, social media’s features provide us too many important gratifications to forego easily. If you think most users will give all this up in the off chance that illegally obtained data from their Facebook profiles and activities may be used to influence their votes, think again.

Algorithms that never let you go

While most people may be squeamish about algorithms mining their personal information, there’s an implicit understanding that sharing personal data is a necessary evil that helps enhance their experience. The algorithms that collect your information are also the algorithms that nudge you to be social, based on your interests, behaviors and networks of friends. Without Facebook egging you on, you probably wouldn’t be quite as social. Facebook is a major social lubricant of our time, often recommending friends to add to your circle and notifying you when a friend has said or done something potentially of interest.

A Facebook ‘nudge’ can push you to attend a local event. Facebook screenshot, CC BY-SA

 

Consider how many notifications Facebook sends about events alone. When presented with a nudge about an event, you may at least consider going, probably even visit the event page, maybe indicate that you’re “Interested” and even decide to attend the event. None of these decisions would be possible without first receiving the nudge.

What if Facebook never nudged you? What if algorithms never gave you recommendations or suggestions? Would you still perform those actions? According to nudge theory, you’d be far less likely to take action if you’re not encouraged to do so. If Facebook never nudged you to attend events, add friends, view others’ posts or wish friends Happy Birthday, it’s unlikely you would do it, thereby diminishing your social life and social circles.

Are you willing to say goodbye? Facebook screenshot, CC BY-ND

Facebook knows this very well. Just try deleting your Facebook account and you will be made to realize what a massive repository it is of your private and public memory. When one of us tried deactivating her account, she was told how huge the loss would be – profile disabled, all the memories evaporating, losing touch with over 500 friends. On the top of the page were profile photos of five friends, including the lead author of this article, with the line “S. Shyam will miss you.”

This is like asking if you would like to purposely and permanently cut off ties with all your friends. Now, who would want to do that?

Authors: S. Shyam Sundar, Distinguished Professor of Communication & Co-Director of the Media Effects Research Laboratory, Pennsylvania State University; Bingjie Liu, Ph.D. Student in Mass Communications, Pennsylvania State University; Carlina DiRusso, Ph.D. Student in Mass Communications, Pennsylvania State University; Michael Krieger, Ph.D. Student in Mass Communications, Pennsylvania State University

Sydney Angels funds QPay $570k to steal millennial students from banks

Australia’s first ever student marketplace app, QPay, has raised $570,000 from a series of high profile investors, including Sydney Angels and the Sydney Angels Sidecar Fund 2, to break into student banking through the release of a student-targeted QPay MasterCard.

QPay aims to use the QPay MasterCard to capture the largest cluster of millennial consumers at the point when they’re most likely to begin making serious financial decisions – when enrolled in tertiary education.

“University is the time when life decisions start to become quite future-focussed, especially regarding our finances,” said Andrew Clapham, Co-Founder of QPay.

“We might be weighing up the amount of student debt we can responsibly accrue, and what return we might expect to receive career-wise. We could be trying to save for a deposit on a property, and wondering the best place to deposit our cash. Or, we might simply be getting a handle on our first experience with budgeting outside of Mum and Dad’s house.

“Whatever the case, university is a crucial turning point for financial decision-making. And the thousands of student transactions occurring on our app each month have allowed us to develop a financial product that perfectly suits the financial behaviour of this group.

“Given that universities arguably comprise the largest cluster of millennials anywhere in Australia, we see this as our first step towards becoming the next challenger bank for millennials,” said Andrew Clapham.

QPay is already used by more than 150,000 students across all major Australian and UK universities, including the University of Sydney, Melbourne, and Queensland, and the University of Oxford and Cambridge. The QPay MasterCard will build on the financial behaviour of these students by uniquely tailoring the rewards it offers every time the card is used for a purchase.

“If you’re a frequent coffee drinker, expect a free coffee from your local coffee shop, or if you regularly shop from a certain store, your next purchase may come with a 50% discount,” continued Andrew Clapham.

“Students are always looking for affordability and convenience – the best deal for the least amount of effort – which is why QPay is being so strongly embraced across all of these universities,” concluded Andrew Clapham.

QPay is backed by a Corporate Authorised Representative with an AFSL license, and the waiting list for the MasterCard has already grown to 4,305 students, far exceeding the initial goal of 2,000 cards.

This proof of concept was a key attraction for QPay’s prestigious investors, which include the head of Royal Bank of Scotland’s Australian arm, Andrew Chick, world-renowned leadership consultant, Charles Carnegie, and prominent angel investor, Rayn Ong.

“QPay’s viral acquisition strategies have created a high level of adoption and engagement even at this early stage,” said Rayn Ong, lead investor and non-executive director of QPay. “It makes sense to take it one step further by bundling relevant deals into the MasterCard offering.”.

QPay received $400,000 from Sydney Angels in 2016 in its first funding round, and has since performed over $11 million transactions for university students – a number which is projected to double by the end of 2018.

The original idea came when the co-founders were students, and realised there was no single access point for student needs such as second-hand textbooks, timetabling, accommodation, student organisations, and campus events.

NAB Adds Samsung Pay

NAB customers can now use Samsung Pay to conveniently make contactless payments.

Samsung Pay is a secure and easy-to-use mobile payment service that allows users to add credit and debit cards from participating financial institutions, and loyalty cards from participating merchants.

NAB Executive General Manager of Consumer Lending, Angus Gilfillan, said Samsung Pay complements the bank’s mobile payments service, NAB Pay, which customers can already use on compatible Samsung and other Android devices.

“We are continuing to invest in giving our customers the best digital payments experience,” Mr Gilfillan said.

“We know Australians increasingly want to pay for their purchases quickly and conveniently. The growth in ‘tap and pay’, and take up of NAB Pay since we launched it two years ago, has been remarkable.

“By adding Samsung Pay, we’re giving our customers more choice in digital wallets.”

NAB customers can also use Samsung, Fitbit, and Garmin smartwatches to make contactless payments, and NAB will also be adding Google Pay to its suite of mobile payment services very soon.

“We see the highest number of NAB Pay transactions being made at supermarkets, restaurants, and at takeaway venues – which is exactly when you want to be able to make quick and easy payments,” Mr Gilfillan said.

Samsung Electronics Australia’s Head of Products and Services, Mark Hodgson, says Samsung is thrilled to be able to provide the Samsung Pay experience to even more Australians.

“Our partnership with NAB builds on our commitment to providing a simple and secure digital wallet experience to every Australian using a Samsung smartphone or wearable. We believe our collaboration with partners like NAB will help further enhance our mobile experience, and we look forward to evolving the portfolio further over the upcoming year.”

NAB also announced last year that it is working with the Commonwealth Bank of Australia and Westpac to build Beem It, a free app enabling anyone to make an instant payment using their smartphone, and to request payment from someone who owes them money or to split a bill.

“We are continually looking at all options to provide our customers with access to safe and secure ways to make digital payments.”

“From our own mobile banking app and NAB Pay, to a range of other payment platforms and services, we’re investing in solutions to help our customers use and manage their money. We’re very pleased to be launching Samsung Pay to our customers now,” Mr Gilfillan said.

NAB’s Mobile Banking App includes a range of world-leading features to help customers have more control over their cards. Features include the ability to switch on or off online transactions, overseas card usage, and ATM withdrawals, and to block, unblock, and replace a card that may have been lost.

The app also includes real-time alerts and merchant information, providing customers more information about their transactions straight away, and a range of options to help them know more about and manage their repayments and accounts.