The Breaking Bad of Predictions: Learning from Failed Forecasts

by Mark Ridinger

 

Predicting the long range and far reaching effects of disruptive, emerging technologies is the focus of many organizations, including CReST. Different phases can be identified, that run the gambit from improving efficiencies within an established industry to disrupting that industry entirely, or creating entirely new, previously unimagined industries. The PC and its word processing “killer app” at first augmented the efficiency of secretaries, for example, but ultimately essentially ended that profession all together, shifting writing documents and memos to the executive or manager. The World Wide Web has been even more disruptive. Ripe for the Internet “reaper” has been “the middleman”, for example. Cut him or her out, and save the customer and seller time and efficiency. Case in point: the travel agent, driven to near extinction by Expedia et al. Successful and accurate prognostications are exciting, but studying—and learning from—examples of projected Internet produced chaos and disruption that has failed to materialize is equally invaluable.

 

Case in point: the real estate agent. By all accounts, this middleman industry should have been essentially eliminated by the Web, and was widely predicted to be so by savvy investors and pundits alike. A 6% commission on a very large sum of money—indeed the largest purchase or investment most folks will ever make– is a lot of money to part with. Add to that the bursting of the real estate bubble in 2006 and the financial crises and ensuing Great Recession of 2008-9, not to mention substantial venture capital backing of startups trying to take over this huge industry, it was all but inconceivable that the demise of the broker did not occur. With so many aligned financial incentives, it seemed to be a slam-dunk prediction. It was a perfect storm.

 

But it wasn’t. In fact, real estate agents are thriving. Bloomberg reports that only 9% of homes were sold without a broker in 2012, down from 13% in 2008.

 

So what happened? What went wrong? And how did so many get it wrong? At CReST one of the books we are reading is Radical Evolution, by Joel Garreau. In it, the author addresses this point at a high level. For example, some categories to which bad or failed technological predications fall into: underestimating complexity, inadequate cost/benefit ratio, the emergence of an even newer/more disruptive technology, prior bad experiences with similar technology, and a fundamental misunderstanding of human behavior. The last category explains why the real estate agent demise prediction went wrong.

 

The successful Internet real estate startups—now substantial companies—recognized this; people wanted their hand held, and were willing to pay for it. Far from displacing real estate agents, Zillow, Trulia and Realtor.com [the main players] have become essentially advertising companies for brokers. Many of the things brokers had to do for clients—show pictures, comparable sales, neighborhood and school information to name a few—are done by these internet firms for free. What is left, namely title search and legal closing documents primarily (which admittedly require expertise), could be outsourced to an attorney for a fraction of what a home seller is paying in commissions. Yet that hasn’t occurred. Redfin, the startup that set out to eliminate brokers and their commissions, has been on death’s door as a company for years, but has finally switched its model as well.

 

Technology does not advance merely for the sake of technology, nor change for the sake of change. The missing link is often the human element. In Social Physics, another book we are reading, the author, Big Data guru Sandy Pentland, contends that “people prefer trusted and personalized relationships,” likely an evolutionary remnant, and one that remains very much intact even in the era of social media. How Big Data and the Internet might be used to exploit those relationships is in part one of the focuses of the emerging field of Cognitive Security.

 

We need to look at—and learn from—failed predictions as much as successful ones, in order to improve our ability to make successful science and technology forecasts, and ultimately policy recommendations. Often, when we get things wrong, it is because we fail to accurately account for human behavior and desires. In short, we need to understand ourselves better to become better forecasters.

 

 

How Technology Changes Everything

How Technology Changes Everything

by Jen Buss

Technology is constantly improving our lives. Each year dozens of new technologies and products are invented that change the way we live day to day. New apps are developed, new sensors to measure our biofeedback, and new safety features for our cars, and so on. Occasionally, a technology is developed that is not just evolutionary but revolutionary and drastically changes society.

The impact that technology has on society occurs in roughly three phases. Initially, technology helps us do things better. The first impact of most technologies is to make existing process work faster or better. These are what we define as phase I impacts.  Later, a new technology often inspires entirely new processes that would not be possible without the technology. We call these phase II impacts. Even later new technologies begin to change entire systems, industries, or even governments. We develop whole new platforms, culture and/or society shifts, or the market does things completely unexpected. We call these phase III impacts.

There are several examples of phase I, II, and III impacts in the past decades.

Consider the multi-phase impacts of the computer on business operations and processes.  During phase I, computer word processors made the existing process of business communication, secretaries and typing pools, faster.  During phase II, business communication changed from memos to emails, removing the need for secretaries and typing pools, thus creating a new set of business communications processes. During phase III, businesses and industries restructured and reorganized.  Two decades ago one needed a very expensive international infrastructure to market and sell globally.  Today, anyone with a computer, a Fedex, Amazon, or Google, account can market and sell from one’s living room.

Another example of multi-phase impacts can be seen in the printing industry.  Phase I impacts are that most, maybe all, former print media, newspapers, magazines, etc. are now delivered and mostly read online; faster delivery of formal media via electronic communication. The phase II impacts are the fact that formal publishers are a struggling business.  Many authors self publish today through blogs, online journals, social media, etc. The emergence of new processes is the hallmark of phase II impacts.  Phase III impacts sustain a large restructuring of the way society obtains their news and entertainment.  The major newspapers, TV networks, and publishers are no longer the primary source of information for many people.  A growing majority receives their news and information directly from those participating in the news via video, tweets, blogs, and real-time connectivity.

As technology continues to impact, influence, and change our society at an ever increasing rate, we should expect its effects to be seen and felt from phase I through phase III in ever increasing and interesting ways.

Is Digital Propaganda All Around Us?

By Ewelina Czapla

Propaganda, communication intended to influence an individual’s or community’s attitude, has arguably existed throughout the entirety of history. Books, pamphlets, posters and paintings have been used to covertly and overtly send political messages intended to manipulate individuals to support one point of view or another. While the “propaganda” often conjures up images of communist sickles and hammers or war campaigns it is also prevalent in stable democratic societies like our own.

In recent decades propaganda has found a new media, the Internet. The Internet has served as a birthplace for many social movements using platforms like Twitter to spread propaganda. For example, the success of the Arab Spring has been attributed to the use of social media which proved so important that government efforts were made to limit access to messages. We are regularly subjected to political campaigns in online advertisements and our inboxes. However, propaganda on the Internet and government response need not be so overt in order achieve shifts in attitude.

Although it can be quite easy to determine if a publication or news network tends to present a biased message it is much more difficult to determine if websites are doing the same. With the breadth of data available for each user it is possible for any company to determine an individuals personal preferences and to tailor that individuals online experience. This tailoring may not necessarily be done to improve the users experience or in the name of marketing but instead to covertly further attitudes that are beneficial to the company’s success. While one of the most fundamental principles attributed to the success of the United States is the free press should the same liberties be granted to private industry? Are we now subject to industry’s propaganda as well as the political worlds? Have they become one in the same? These questions must be addressed as society becomes more integrated in an online world.

Cognitive Liberty: Looking beyond the Freedom of Speech and Press

By Ewelina Czapla

Cognitive liberty is defined as sovereign control over one’s own consciousness. As the ability to manipulate and monitor the human mind grows with the development of neuroscience and analytical tools the question of one’s right to remain cognitively sovereign will need to be answered. The issue of cognitive liberty raises questions about privacy and freedom of thought as well as the 4^th and 5^th Amendments to the Constitution. Is cognitive liberty is a basic right? Should an individual simply be protected from cognitive manipulation by others or also be free to engage in the use of mind-altering technologies to improve their own cognition?

Technological developments, such as the fMRI among others, allow for the scanning and analysis of the brain’s activity. FMRI scans can determine whether or not an individual is responding honestly to questions or if they approach situations with an unconscious bias. Additionally, brain scans can identify areas of abnormality in the brain such as a spot or cyst that suggest the propensity to commit a particular kind of crime. These technologies have the ability to contribute to decisions regarding a defendant’s guilt, a prisoner’s sentence or date of parole. Should this technology alter criminal and civil procedure? What are the future civilian implications of technologies like fMRI? Could they be used as a pre-screening measure for employment, similar to a drug test?

If amendments to our rights were made to ensure ‘cognitive liberty’, there would be profound repercussions on commercial and advertising firms, campaigns, and others who may in the future seek to influence cognition. Arguably, should ‘cognitive liberty’ be granted, the balance of power could shift in favor of the individual from corporations or government. Whether people have the right to use the ‘fourth estate’ free of manipulation is a pressing governance question with broad future ramifications.

The Fiction of Pyschohistory has become a Promethian Reality

by Jenny McArdle

In the year 12,069 (or Foundation Era -79) Hari Seldon predicted the fall of the Galactic Empire, the ensuing turbulence of the interregnum years, and the rise of the ‘Foundation’, a group of scientific pseudo-religious and merchant rulers modeling Plato’s own ‘philosopher kings’. Seldon did this through the science of psychohistory, a science that used statistics, history, and sociology to predictively model human behavior. However, Seldon did more than just foresee the imminent collapse of the Galactic Empire, he used psychohistory to mold the future to his liking, setting in motion the futures that would support the rise of these scientific ‘philosopher kings’ and eventually the installation of the Second Empire for the good of humanity.

While Hari Seldon’s psychohistory was a literary thread that Issac Asimov used to bridge his science fiction short stories in the Foundation trilogy, the scientific discipline he fictionally created between 1942 and 1950 now seems strangely clairvoyant.

Indeed, the convergence of big data, psychology, and behavior science (i.e. cognitive security) is making psychohistory a reality. Big data has allowed scientists to study billions of human interactions at the individual level. The MIT’s Human Dynamics Laboratory under the guidance of Alex ‘Sandy’ Pentland has discovered that by using computers to analyze mathematical patterns of human interactions, they can explain and predict phenomena—political upsets, flu pandemics, human productive output, financial crashes, among others. While Pentland views cognitive security as a force for future good, he notes that the ability to track, predict, and potentially control human behavior can also be exploited.

Prometheus has long been a symbol of the human quest for scientific knowledge. Prometheus ensured human progress through the gift of fire, but was sentenced to eternal torment by the Olympian Gods for his transgression. Science does at times have overreaching and unintended consequences. Cognitive security can be used as a force for good, but in the wrong hands it can also be egregiously misused. Are we to assume that all our future ‘psychohistorians’ will be motivated for the good of humanity, like Hari Seldon? A brief glance through the history of mankind would beg to differ.

 

Are we entering an era of the ‘informationized power broker’?

by Jennifer McArdle

 

Are we entering an era of ‘informationized power brokers?’ In dystopian science fiction futures, books like Red Mars have presented futures where transnational corporations wield as much, if not more, power than states. The convergence of big data, psychology, and neuroscience (i.e. cognitive security) may help enable that reality, creating new ‘informationized power brokers’. Internet sites and social media platforms like Google, YouTube, and Facebook, have amassed immense amounts of data on individual users, in some cases up to 1,200 pages of data on a single person. This data, when combined with behavioral science and analytics can effectively model human behavior; providing these new ‘informationized power brokers’ the ability to ‘social engineer human behavior.’

Corporations have long used advertising to influence consumer behavior. The advent of the printing press helped spawn weekly print advertisements in newspapers and periodicals in the 17^th and 18^th centuries. However, what is fundamentally different now, is that corporations can uniquely cater their message to an individual based on their data profile. In a much-publicized 2012 media story, Target was able to identify a pregnant teenage-girl before her father, simply based on her consumer Internet search history.  Internet search data when combined with the power of behavioral science can reveal very unique things about individuals, even life-changing events, like pregnancy. But what power, besides more influential advertising, does this really give corporations? Why does this make them potential ‘power brokers’?

Internet corporations, particularly platforms like Facebook and Twitter have content visibility and data sharing methods that are based on private algorithms and policies. These algorithms and polices are often opaque or inaccessible to the public, yet have immense influence. They control what the public does or does not see. What happens when one of these platforms is biased? A Nature study noted that during the 2012 election, people voted in statistically higher numbers after seeing the civic “go vote” message on Facebook. Could a corporate social media platform use such a tool to selectively target individuals whose views would advance their corporation? Zeynep Tufekci speculates that it is possible, and more importantly, that this would go on largely undetected by the public and government. If corporations are able to ‘social engineer’ the public for corporate benefit, will we be entering an era where the real power brokers are not states but corporations?

Pre-Crime

Pre-crime.  Part 2 of the Modeling and Profiling Blog

by Mike Swetnam

Many have noted that the greatest threat to the human race in the 21st century is ultra-powerful technology (like nuclear, bio weapons, or nano tech) in the hands of madmen.

Clearly there are technologies like nuclear and bio that can destroy all or most of us.  Also there is no shortage of crazy leaders and despots who would not hesitate to use these weapons if they could.

There is also a rising number of people who just lose it and go ‘postal’.  People that are so disenfranchised, so disconnected, and so nuts that they grab a gun and start shooting.  We have had far too many of these incidents this past decade.

Combine the rising number of ‘postal’ cases with the increasing availability of bad technology, like bio and how long will it be before some crazy person cooks up a super virus instead of crabbing a gun?

This is a really bad concept!

A crazy person with a gun might kill 20 innocent children, but a crazy person with advanced biotech can kill millions!

We can not let that happen!

How do we prevent this Marriage of Mayhem: destructive technology married to a crazy fatalistic personality?

We certainly don’t want to wait for it to happen, and then just prosecute the crazy person.  It would be far better to find, identify, and deal with the crazy before he did the deed that killed millions.

Fortunately, we are developing the behavior modeling technology to do this.  Today, industry models your buying behavior so well that they can predict what you will buy.

All of us have joined these frequent buyer programs that track what we buy, how often, and what we buy at the same time.  This data has been used to profile us.  These profiles of our behavior help industry market to us in very targeted ways.

They also help industry identify and deal with fraud.  When someone steals a credit card and attempts to use it, the profile of that fraudulent use is different then the profile of the card owner.  Computer programs note the difference and alarms go off.  These alarms are used to stop fraud and miss-use early.  This happens when you travel to new places and use your card in strange and new ways that cause the card company to want you to call in and verify your identity.

This same technology can be used to identify the behavior of someone getting ready to go postal.  It turns out that the behavior of those approaching the breaking point is fairly identifiable.

As this technology develops and is proven accurate, with low false alarms, how will we use it?

The US Constitution says that one is innocent until proven guilty.  Can we legislate the legitimacy of scientific models as proof of potential future guilt?

These are not esoteric questions.  Our very survival could be threatened by madmen with access to bad and destructive technology.  It is absolutely clear that we cannot wait for them to attempt our destruction to act.  It is just as clear that our basic beliefs in innocence until proven guilty will be challenged by these realities.

It’s time for a Constitutional discussion that we have not had for 237 years.