Reporter's Notebook: 7 Big Data Themes Emerging from Nonbanks

Big Data. The business buzzword evokes dual emotions in bankers: they're jazzed up about new revenue possibilities and stumped by its many meanings and current limitations. Indeed, finding ways to extract value from the ever-increasing treasure troves of data sources plagues companies worldwide, including banks.

Presenters from the likes of the CIA and health insurance provider Aetna came together to speak to the opportunities and hurdles of Big Data during GigaOm's Structure: Data 2013 event held this week. The drawbacks execs cited are varied, but include inherent flaws in modeling, biased assumptions, bad data, overflow of new information sources, and data obesity — when companies horde data indiscriminately.

Of the utmost importance to society, argued speakers, are the ethical issues arising from computer algorithms to which consumers are continually outsourcing their decisions, for instance by relying on Pandora for music selections. Algorithms are meant to increase human intelligence, but that intention has yet to be realized in the present day. Eric Berlow, founder of Vibrant Data Labs, articulated the problem clearly during his presentation:

"In theory, algorithms augment our intelligence and then machines learn from us and this goes round and round and we all get smarter and superhuman and everything is great. But the problem is if we don't pay attention, [algorithms] can make us dumber," said Berlow.

Berlow pointed out that news media websites' algorithms, such as "most emailed" lists, herd people like sheep to the most popular content (often sex scandals and personal health articles), leaving the reader wondering, "Where's the news?" While in theory, humans have a growing relationship with analytics to see the big picture that they couldn't otherwise see, in reality, people are getting bogged down in the minutia of what's trending now, which encourages "dumb learning," he said.

To do better, Berlow suggested augmenting algorithms with human input. One of his projects, WeTheData.org, is designed to do that. The initiative gathers input from human beings to map out the complexity of personal data challenges. The group illustrated how its mapping approach works at the 2013 World Economic Model in Davos.

Beyond the need for humans to input insights into algorithms to achieve more meaningful outputs, speakers identified a number of big data themes to watch out for in the coming months and years. Below are seven:

1. Algorithms must be used to solve bigger issues. "We are using [big data] to solve trivialities," said Sean Gourley, co-founder and chief technology officer of Quid, a Big Data software provider to enterprises like banks. Companies are using data to build models that determine where the cereal should sit on a shelf to yield more sales. One of the top predictors of intelligence is individuals who like curly fries on Facebook, he said. Indeed, recent research from the University of Cambridge shows that among Facebook users, "likes" of "curly fries," "thunderstorm," and "science" were some of the best predictors of high intelligence.

Gourley advocates for a reimagining of data science that would attack larger problems, such as what putting 30,000 more troops in Iraq would do. "Big data must solve big problems," he said. To that end, he prefers the term "data intelligence" to "data science."

2. Data can show what customers are saying, but it needs to show why they're saying it. Sentiment analysis, for example, tells companies what people are feeling about their brands but fails to clue enterprises into why people are upset. That will change. Enterprises will be answering the "why" question in the next two to three years, said Timothy Estes, founder and chief executive of Digital Reasoning, a CIA-backed analytics company.

3. Smart machines and sensors on mobile devices are becoming more meaningful sources of data for organizations, including the CIA. The CIA's CTO, Ira "Gus" Hunt, took to the stage to make this point. Take Fitbit, for example. The technology has 100% accuracy in identifying people by their gaits. Fitbit also tracks calories burned, steps taken, distance traveled and sleep quality. Hunt reminded the audience that everyone can be tracked by their phones, even when they are turned off.

Sensor data inputs are expected to play a big role for retailers. Speakers said retailers will use location data in their physical stores to help people find what they want to buy.

4. An app's user interface must be simple. The CIA's Hunt also asserted that emerging tools must come with intuitive user interfaces so that people can use them just as easily as they can use Excel. "The power of big data can only be fully realized when it's in the hands of the average user," Hunt said.

5. Data is the new currency. Enterprises looking to get customers to give up more of their data must incentivize them. Banks, for example, are trying to send contextual offers to consumers based on their purchasing histories.

6. Amazon is a Big Data role model for other businesses. The online retailer got the most shout outs from varied speakers. An executive from Aetna suggested that people should be able to logon to their health websites and receive recommendations on ways to improve their well-being the way they get product pitches from Amazon's recommendation engine. Other speakers pointed out how data-driven companies like Amazon are creating new forms of competition for traditional companies bogged down by legacy systems. But it's not the technology alone that's distinguishing these newer threats. "What's disruptive isn't the technology, it's the business model," said Paul Maritz, chief strategist at storage hardware provider EMC.

7. Collaboration of humans and machines is a must. Humans are better at some things like intuition and designing the use case for an algorithm to model out, while machines can scale information better. That's why both parties have to work together.