Everybody Lies: What the Internet Can Tell Us About Who We Really Are by Seth Stephens-Davidowitz
[Bloombsbury, 2017 978-1-4088-9471-2]
"This book is about a whole new way of studying the mind. Big data from internet searches and other online responses ... offer an unprecedented peek into people's psyches." Stephen Pinker.
The author is an internet data expert, tracking the digital trails people leave behind as they make their way across the web. He tries to understand what we really want, what we will really do, and who we really are. The book is a fascinating read, and I've just picked out some of his findings.
He accesses a variety of sources, including Google Trends, which tells users how frequently any word or phrase has been searched in different locations at different times. It allows you to compare the relative frequency of different searches, but does not report the absolute number of any search. It also will only provide data when lots of people make the same search. (2021: Google Trends requires a modern browser such as Chrome.) Other sources analysed were Wikipedia, Facebook profiles, Stormfront, PornHub and other Big Data sources. And it turns out that people lie to friends, lovers, doctors, surveys and themselves - but on Google they share embarrassing information.
Racial prejudice in the USA. Barack Obama won the 2008 US presidential race rather easily and the polls suggested that race was not a factor in how people voted. But Google Trends showed that on election night, searches and signups for Stormfront (a white nationalist site) were ten times higher than normal. And surprisingly, places with the highest search rates included upstate New York, western Pennsylvania, eastern Ohio, industrial Michigan and rural Illinois alongside West Virginia, southern Louisiana and Mississippi. This suggested that the true divide was not South versus North, but East versus West and that plenty of Democrats have racist attitudes.
The cause of racism is typically thought to be economic insecurity and vulnerability, but neither racist searches nor membership of Stormfront (a white nationalist site) rises when unemployment does.
Many people tell pollsters they intend to vote, but don't actually do so. Google searches for 'how to vote' or 'where to vote' more accurately reflected actual turnout. Seemingly neutral searches that include the names of both candidates can give a clue to voting intention, with the preferred candidate usually the first named. And the secret racism was a likely factor in Trump's election.
Surprisingly searches reflecting anxiety don't rise after terrorist attacks. Searches for jokes are lowest on Mondays, the day when most people report they are unhappy, they are lowest on cloudy or rainy days, and plummet after major tragedy.
While larger datasets can be useful in some situations, in others it is more important to have the right data. What symptoms best predict pancreatic cancer? Researchers found that searches for back pain and yellowing skin, or indigestion and abdominal pain were most likely to be made for those later diagnosed with the cancer.
There is a basic human fascination with the dramatic. In a survey people ranked death by tornado as more common than death from asthma, but asthma causes about seventy times more deaths than tornadoes. But tornado deaths get in the news,while asthma deaths don't.
What are the chances of becoming a basketball player? It is commonly held that most NBA players grow up in poverty, combining a range of data says not. In fact you are more likely to succeed if you live in a wealthy county of the US, with a comfortable family life and have a common given name - plus, of course, being tall.
Buying wine? A big percentage of the quality of a wine is down to the weather during the growing season. Warm weather and early irrigation are needed for the best grapes. So if thinking of buying a 2016 bottle, you need to look at the weather that year where the grapes were grown.
Democrat politicians use terms such as estate tax, privatize social security, worker's rights and poor people while Republicans refer to death tax, reform social security, private property rights and government spending. Sometimes it is using different terms for the same concept. Republicans described the federal inheritance tax as a death tax, Democrats as an estate tax. if an area is generally liberal, the dominant newspaper will tend to be liberal and the market determines a newspaper's slant more often than its owner's views. And newspaper readership tilts a bit left on average.
And it turns out that people lie a lot, often the little white lie. A 1950 election survey in Denver found that far more people said they were registered to vote, voted last presidential election, voted last mayoral election, gave to a recent Community Chest charitable drive than actually happened taken from official counts. Giving the wrong information made people look good to others, even in a situation where they were not identified, while some people even lie to mess up surveys.
Surprisingly the internet news industry is dominated by a few massive sites. Therefore they tend to cover a wide range of subjects and views to appeal to the most people.
Data reveals that child abuse does not lessen in hard economic times, it just gets reported less. Women search more often for self-induced abortion in areas where it is hard to get a legal abortion.
On social media there is no incentive to tell the truth; on the contrary there is incentive to make yourself look good to friends, family members, colleagues and strangers. Family life seems perfect, and young adults are always at cool parties on Saturday nights.
In 2006 Facebook launched News Feed which gives users updates on activities of all their friends. While some saw this as stalker activity and tried to get it removed, Facebook saw that people were staying far longer on the platform. News Feed is still there.
Google searches show that you are not alone. The questions you don't raise in person as they sound too silly but they can alert us to people who are suffering. Analyzing these anonymous searches can assist agencies working in specific concerns. E.g. Child protection agencies could target resources to specific areas. And searching can lead us from problems to solutions. 'When we lecture angry people, search date implies that fury can grow. But subtly provoking people's curiosity, giving new information and giving new images of the group that is stoking their rage may turn their thoughts in different, more positive directions.'
What baseball team is doing well when a US boy is around 8 is likely to remain their lifetime team to support. For female baseball supporters, it is more likely to at age 22. Our political preferences are also linked to specific periods. Between the ages of 14 and 24, many Americans will form their views based on the popularity of the current president; the views will, on average, remain constant a lifetime.
There is a popular view that watching violent films encourages people to commit violent actions. It turns our that if watching such films at a cinema, where alcohol is not available, actually reduces crimes.
Doppelganger searches for someone who is as near as possible your exact match is used in some form by Amazon to suggest other books you might like, and Pandora to suggest songs you might want to listen too. But it can also help in medicine by finding others with similar symptoms; this can narrow down diagnoses, especially where the condition is rare.
Many people play a lottery, but few win. But does winning make you happy? Not in the short term it seems, but longer term generally yes. But if someone wins and starts buying items such as an expensive car, neighbours are significantly more likely to go bankrupt attempting to buy likewise.
Money lenders want to assess the likelihood of borrowers repaying the loan. A research study showed that terms used by people researching lenders can indicate the most likely to repay (debt-free, after-tax, graduate, lower interest rate and minimum payment) or to default (God, promise, will pay, hospital, thank you). But this raise some ethical questions - is it acceptable?
The ability to zoom in on gambling customers it potentially dangerous. Casinos know that every gambler has a 'pain point' when the level of losses frightens the player such that they stop for an extended period. So they try and make sure that the gambler gets close to the pain point but not quite there, for instance, by offering free meals. They use doppelganger methods to establish the likely pain points.
Education research. Anonymized children's online behaviour can reveal how they are learning and developing. How is their spelling - does it indicate dyslexia? Do they have friends. Are they becoming more mature. Education software companies: with EDU STAR students log in to a computer and are randomly exposed to different lesson plans, then take a short test to see how well they learned the material. Schools can therefore learn what software works best. This can yield surprising results as students who were taught fractions via a game tested worse than those who learned in a more traditional way.
END
