I expected a Model T, but instead I got a loom: Awaiting the second big data revolution by Mark Huberty.
Abstract:
Big data” has been heralded as the agent of a third industrial revolution{one with raw materials measured in bits, rather than tons of steel or barrels of oil. Yet the industrial revolution transformed not just how firms made things, but the fundamental approach to value creation in industrial economies. To date, big data has not achieved this distinction. Instead, today’s successful big data business models largely use data to scale old modes of value creation, rather than invent new ones altogether. Moreover, today’s big data cannot deliver the promised revolution. In this way, today’s big data landscape resembles the early phases of the first industrial revolution, rather than the culmination of the second a century later. Realizing the second big data revolution will require fundamentally di fferent kinds of data, diff erent innovations, and diff erent business models than those seen to date. That fact has profound consequences for the kinds of investments and innovations firms must seek, and the economic, political, and social consequences that those innovations portend.
From the introduction:
Four assumptions need special attention: First, N = all, or the claim that our data allow a clear and unbiased study of humanity; second, that today equals tomorrow, or the claim that understanding online behavior today implies that we will still understand it tomorrow; third, that understanding online behavior off ers a window into offine behavior; and fourth, that complex patterns of social behavior, once understood, will remain stable enough to become the basis of new data-driven, predictive products and services. Each of these has its issues. Taken together, those issues limit the future of a revolution that relies, as today’s does, on the \digital exhaust” of social networks, e-commerce, and other online services. The true revolution must lie elsewhere.
Mark makes a compelling case for most practices with “Big Data” are more of same, writ large, as opposed to something completely different.
Topic mappers can take heart from this passage:
Online behavior is a culmination of culture, language, social norms and other factors that shape both people and how they express their identity. These factors are in constant flux. The controversies and issues of yesterday are not those of tomorrow; the language we used to discuss anger, love, hatred, or envy change. The pathologies that afflict humanity may endure, but the ways we express them do not.
The only place where Mark loses me is in the argument that because our behavior changes, it cannot be predicted. Advertisers have been predicting human behavior long enough that they do miss, still, but they hit more than they miss.
Mark mentions Google but in terms of advertising, Google is the kid with a lemonade stand when compared to traditional advertisers.
One difference between Google advertising and traditional advertising is Google has limited itself to online behavior in constructing a model for its ads. Traditional advertisers measure every aspect of their target audience that is possible to measure.
Not to mention that traditional advertising is non-rational. That is traditional advertising will use whatever images, themes, music, etc., that has been shown to make a difference in sales. How that relates to the product or a rational basis for purchasing, is irrelevant.
If you don’t read any other long papers this week, you need to read this one.
Then ask yourself: What new business, data or technologies are you bringing to the table?
I first saw this in a tweet by Joseph Reisinger.