Big advertising or big surveillance: you choose

Around August last year, Janet Vertesi began hiding the fact that she was pregnant. Not from her husband, family and friends, but from the internet and every marketer who uses the internet to influence the product choices of new mothers.

Vertesi, an assistant professor of sociology at Princeton University, used her experience as a mother-to-be to examine the way that the bots and cookies that track our online actions can affect our political and social interactions with the world.  She was particularly interested in what happens when offline tracking systems (e.g. supermarket loyalty cards and branded credit cards) come together with online systems (e.g. cookies, bots, Amazon recommendations), leading to interesting combinations like the way her use of Facebook could affect the way that real-world shops in her area would interact with her.

(See Vertesi from 1:30)

To start, Vertesi banned all mentions of her pregnancy on social media, both in text and pictures. To be sure, she phoned all her family and friends with the good news (an old-fashioned action in itself) and asked everyone to avoid mentioning her pregnancy online. Interestingly, this ban didn’t translate well for some of her family members, who were unfriended to make sure they couldn’t spoil the project.*

The other big change Vertesi and her partner made was to make sure all their baby-related browsing and buying was untraceable, which meant they had to avoid all the ways that purchases can be analysed. These included obvious incentives for people to consent to having their data analysed – like company loyalty cards – but also larger buying systems, like credit cards and franchises’ analysis of purchases based on name, location, pattern of internet use and other identifiers that can be used to pinpoint an individual.

To make their purchases completely anonymous, they used cash for everything, from vitamins and maternity wear to a pregnancy cast. For online purchases, they created a new email address, disconnected from their other affairs, which they used to create an equally anonymous Amazon account. They then bought their baby goods using Amazon gift cards they’d bought with cash, having the products delivered to local parcel lockers. Needless to say, it was a lot of extra work.

Browsing for information on pregnancy and raising a baby was equally hard. As Vertesi quipped in the presentation “I’m actually here today to win a ‘most creative use of TOR’ award”. It’s a funny idea, but also really serious, as Vertesi and her partner realised when they went to buy yet another slew of Amazon gift cards and discovered a sign saying the company “reserves the right to limit the amount of pre-paid card purchases and has an obligation to report excessive transactions to the authorities”.

This was just one of the ways that Vertesi realised their behaviour could have raised alarms. Looked at objectively, Vertesi and her partner were making significant cash withdrawals across the city (including one withdrawal of thousands of dollars for a pram), frequently using TOR** and using items like gift cards to obscure their purchasing habits. While in this case Vertesi was hiding her pregnancy, these could also have been read as signs of her involvement in organised crime, political rebellion or any one of a number of illegal activities.

As Vertesi points out, this connection can be drawn only as a result of simplifications that conflate ‘unusual activity’ with ‘suspicious activity’. This has huge implications for people who have legal reasons for their unusual activity such as legal protest against the government, hiding their pregnancy or, in the case of a friend of mine, hiding the fact they were transexual until they were in a position to come out on their own terms.

The implications of data mining for privacy is also worth considering here. When I’m writing this blog, I am very aware that it’s a semi-permanent record which will be available for the rest of my life. Anything I say here, I need to be willing to stand by for the rest of my career (or at least willing to laugh it off as a 20-something’s indiscretions).*** However, I’m very aware of the staying-power of my words because of my study and work around internet issues. In contrast, the public conversation about online accountability has only come as far as the importance of not sending naked selfies and other teen-related indiscretions (though there have been plenty of examples of adults missing that lesson). We haven’t yet had the hard examples of early online actions having long-term implications. These watch-your-actions-early-in-life lessons have been common in some circles for a while (e.g. student politicians’ arguments being re-published decades later {See The Deadly Newt for the original audio}), but most people do not live their lives with the expectation that their every action is up for analysis.

XKCD on embarrassing pictures on the internet

It pains me to disagree with XKCD, but those pictures can be awfully significant.

To most people, the internet is – and should be – a place to talk comfortably and casually, with no expectation that their discussions about gardening, fitness or politics could have wider significance in the next five years, let alone the next 50. However, individuals’ comments now exist as searchable, contextualised information that can be found decades later – after someone has changed their stance on an issue – or used for a purpose that the individual would never have given their permission for.**** I have no solution to suggest to this – opting out of the online sphere has implications, and limiting one’s speech, especially on political subjects – has even bigger implications in the long run, as we lose the benefits of robust political discourse.

Ultimately I’m taking two lessons from this:
1. As an individual, I should think before I speak/type/share data/tick the box on a EULA.

2. As a public servant, I should know the potential consequences of my policies; even those that are a stretch at the time.

Both of these seem self evident, but this research has taught me that they’re awfully important.


*An aunt and an uncle (who lived on different sides of the world) both sent private Facebook messages to Vertesi congratulating her on the good news. When she deleted the messages and explained to them that this counted as ‘mentioning it online’, one commented that she “didn’t realise that a private message wasn’t private”. While it was true that individuals wouldn’t be able to go through that conversation, the aunt hadn’t realised that the information from the conversation could still be used by Facebook for data analysis.

**TOR is free software that people can use to increase their chances of being anonymous on the internet. It’s not a perfect anonymiser, but it’s widely available and very effective.

*** The European ruling about an individual’s right to be forgotten is significant here – if it becomes a widely applicable rule, all bets are off.

**** The Economist’s article about American healthcare companies using social media updates to determine whether someone is too active to be offered health insurance is a perfect example. Among the details, the article mentions the head of an American company that produces insurance underwriting software who has shifted purchasing fast food with cash, so that the choice of junk food can’t be factored into his health insurance premiums.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s