Nassim Nicholas Taleb achieved notoriety with several books written before the housing crisis, criticizing the financial industry for putting so much faith in its predictions. He argued in Fooled by Randomness
that there are problems with our attempts to understand the past, and even larger issues when we use it to predict the future. Those criticisms turned out to be justified.
"Big data" is allowing more industries to try their hand at fortunetelling. With social media and portable devices, we can watch society just like traders watch the stock market. People can be measured, quantified, modeled. As we enter this brave new world, it’s worth considering some of Taleb’s points and seeing how they might apply to big data.
"The more data we have, the more likely we are to drown in it."
Information today is widely available, and easily accessible. Wall Street mines this information for stock patterns, and economists search it for clues on the economy. The UN digs through social media for signs of humanitarian crises
, the NSA uses our posts and updates to hunt for terrorists, and the media uses Facebook
(NASDAQ:FB) and Twitter to predict box office results. LinkedIn
(NYSE:LNKD) and eHarmony
(NASDAQ:IACI) have hung their hats on the idea that better statistics make for better relationships, while nearly every industry on the map hopes that today’s traffic in personal data will allow them to better tailor their advertisements and products.
However, as Mark Twain was known to observe, statistics can lie. If you were judging the impact of Hurricane Sandy by watching Twitter, you would have assumed
that Manhattan was hardest hit; it was better represented because it still had power. Online dating profiles often list a passion for "long walks on the beach,",but few of the people who make those statements really mean it. Much of what we call "big data" was never meant to be accurate, and because so much of it is shared, any inaccuracies can be hard to keep track of.
Even unbiased data can point to trends that no longer exist, or paradigms that are about to change. With his Lucas critique, American economist Robert Lucas rocked finance and economics in the '70s by codifying what investors already knew: Past returns are not indicative of future results. The larger our pool of information, the more chance patterns we’re likely to find in it, and the easier it becomes to “prove” whatever we like.
It’s become a simple matter for businesses to amass data, and warehouse it cheaply at Amazon
(NASDAQ:AMZN) or Teradata
(NYSE:TDC). They can take advantage of new forms of databases like NoSQL and Hadoop, which don’t require any fixed ideas about what the data means. Visualization products from Tableau Software
(NYSE:DATA) and Splunk Inc.
(NASDAQ:SPLK) allow businesses to sort through this information hoard, and condense it into elegantly simplified charts. In short, data mining is becoming so easy that anyone can do it, and the danger is that more industries will fall victim to the sort of statistical “insights” that wreak havoc in financial markets.
"We are trained to take advantage of the information that is lying in front of our eyes, ignoring the information that we do not see."
Data isn’t just getting bigger – it’s also getting less avoidable. The beauty of a PC was that you could – and frequently had to – walk away from it. Smartphones brought the flow of information out of the office, allowing it to follow us into the dining room, the bedroom, and (if we’re being honest) the bathroom. Google
(NASDAQ:GOOG) Glass promises to beam information straight to your eyeball. Our time is increasingly occupied by data, and that data is increasingly trivial.
Consider that Fidelity Investments recently launched a stock ticker
for Google’s headset. About investing, Taleb points out, “Over a short time increment, one observes the variability of the portfolio, not the returns.” He offers the example that an investor with a 93% chance of making money this year – pretty good odds – would only have a 50.02% chance of making money in the next second. In the short term, we see the noise rather than the signal, and the trees rather than the forest. Using a Glass app for your portfolio might only amplify that tendency.
By the same token, we now spend more time responding and less time thinking
. In finance, strategies have shifted toward quantitative analysis and high-frequency trading. In advertising, blitzkrieg social media campaigns like Oreo’s "dunk in the dark"
have captured the industry’s imagination. The focus is on momentary tactics, rather than fundamentals or broader strategy. There’s no denying that speed is important when you’re chasing a market, but when you’re trying to get in front of it – that is, to innovate – haste can make waste.
(NASDAQ:AAPL) has profited from its ability to simplify and offer products that do fewer things but do them better. This requires a certain level of introspection. Tony Fadell was a part of the team that designed the iPod, and as he puts it
, “Great products come from a strong point of view... You say no to most of the features that data says you’ll need.”
Fadell does point out something that big data does well: It "shows how people use your product in ways you hadn’t expected." These unexpected uses aren’t uncommon in business; Tumblr took off on the back of erotica and fan fiction, and medications are frequently more popular for their off-label uses. Data doesn’t allow us to predict who will buy a product, or what they’ll use it for, but it does help us to judge the results.
And that’s the point: The success of big data will depend upon how it’s approached. If businesses and governments see it as a tool for self-measurement, they’ll find it useful. If they see it as a way to “crack the code,” or quantify human nature, or predict the unpredictable, they’re probably fooling themselves.
As Facebook Tests Mobile Shopping Feature, Does PayPal Face a New Threat?
Is Apple's iPhone Finally Coming to China Mobile?
Apple Inc. Would Be in Trouble With or Without Steve Jobs