The Economist • 27 April 2013
Get real. Economist Daniel McFadden takes issue with the concept of a pleasure-seeking, risk-evaluating "homo economicus," arguing that the events directly preceding a decision have a profound influence on how choices are made. Read on for more on why manipulating brain chemistry and triggering social network interactions could help improve public policy outcomes.
The idea that people aren't the purely rational, self-interested decision-making machines that have become something of a caricature of economists' thinking has been around for a while. The theories popularized in the recent book Nudge by Richard Thaler and Cass Sunstein are but one example. Economics can certainly benefit from the insights about human behavior and decision-making uncovered in other fields such as psychology and even biology. Findings from psychological research, for example, help us understand the implications of relaxing the assumption that people take into account all available information to make choices in their best interest. However, it is worth pointing out that "orthodox" economic theory has yielded vast insight into economic behavior that has stood the test of time; incorporation of new assumptions regarding decision-making and choice will likely enhance, rather than supplant, previous work. (Lavoie)
The New Yorker • 22 April 2013
Smarter A.I. Most of the current work in artificial intelligence focuses on a single problem, like playing chess or Jeopardy!, but some researchers are looking at how to create a machine with the flexibility to go beyond one task. Blogger Gary Marcus takes a look at some of the promising approaches under development in labs around the country.
"Could anyone, ever, build an A.I. system that might match a human being in the sheer flexibility of learning new tasks?" asks Mr. Marcus—perhaps rhetorically, since he's dubious about the whole thing. Well. Serendipitously, I learned just this morning in a Los Angeles Times story that the human brain has 100 billion neurons, and the Human Connectome Project (sort of a neurologist's version of the Human Genome Project) seeks to trace the trillions of connections between them. Sounds like a lot of programming to emulate that! I hope we never achieve that holy grail of A.I. in my lifetime: I know all about Skynet. (And, by the way: that LA Times article is about USC—a private university—stealing a huge neurology lab of about 100 scientists from UCLA—a public university. I'll not exceed my remit by whining about the plight of the public university . . . ) (Dooley)
Intelligent Life • May/June 2013
Prediction power. Algorithms are driving more and more corporate strategic decision-making, from Hollywood studios to Formula 1 racing. The results are fascinating—read on to learn which three A-list actors actually wield star power and how Democrats fine-tuned their grassroots campaign efforts.
I would like to see the algorithm that calculates the expected financial return to the investors in a film. Seems a shame to keep that under wraps. Its calculated estimate might be a helpful addition to the views of film critics and advertising. That reminded me of the comedian John Mulaney, commenting on a film that supposedly cost over 100 million dollars: "You didn’t have to make the movie, I would have paid just to see the 100 million dollars." (Washburn)
Smithsonian • 25 April 2013
Thinking ahead. It doesn't take a statistician to figure out a spike in searches on terms like "debt" and "short selling" might signal trouble ahead for the economy, but what about words like "color" or "restaurant"? As Google hones its ability to predict global disasters and personal setbacks simply by tracking search trends, a new study by British researchers harnesses Google Trend data to predict stock market action based on 98 commonly used words.
Ah, the power of aggregation and Big Data! This article describes a study to correlate the use of Google search terms to stock market behavior. Not surprisingly, aggregate enough search terms and enough stock transactions and you can find correlations. That falls apart with smaller samples of users or for individual stocks, but aggregate enough of both and magic happens. Can we do something similar for libraries? Aggregate enough behavior at libraries and compare to Google search terms? With WMS, can we get enough circulation data to be able to start predicting circulation trends? The technology keeps making this easier, all we need is the data. (LeVan)
ReadWrite • 25 April 2013
The "knowledgization" of data. Search is being reengineered for the small screen and the result will be "proactive, geo-fenced answers that will pop up before you even frame the question," says techno-pundit Brian Proffitt. Read on for more on contextual search and why companies are gearing up to deliver knowledge rather than information.
Traffic reports and search results about sports scores appear willy-nilly at the top on my mobile devices? It's nice to see network-level "search in context," but I always refuse to allow "push" on my apps. Personally, I don't like being scrutinized by algorithms and then being manipulated when they send unsolicited messages. Don't get me wrong: I loved doing SDI (Selective Dissemination of Information) when I was a reference librarian back in the 80s. Now I'm a delighted beneficiary on the other side of the desk: our OCLC Library wizards regularly send me new materials for my current work that they glean using their expertise and cool tools. Maybe I'm too old school, however, because I resent that the network—mobile or not—monetizes "information labor" that I unwittingly do for free. (Schaffner)
Above the Fold Quiz
According to an item in this week's News and Views section, what is wikidata and how can it help the 285 wikipedias collaborate on VIAF Unique Identifiers?
Get the answer.