Survivorship bias – Did Big Data help win the World Cup?

Two interesting Big Data stories emerged from tech providers in the days after the four-yearly football jamboree had come to a close in Rio. The first, that Microsoft’s Bing and Cortana products correctly predicted 15 out of 16 matches in the knockout stages, and the second that SAP data products helped the German team to win the tournament. There we go, then. That’s the case for Big Data successfully closed.

Of course it’s not quite like that – and what we see is another example of the flaw in human thinking that’s known as suvivorship bias. Rather than looking at the whole case, we focus on the things that have successfully made it through a process. So, no press releases from all of the Big Data providers supporting the 31 teams who didn’t win the World Cup (I can’t believe that SAP were the only ones in Brazil). No trumpeting from the digital personal assistants who loused up in their futurology (although modern-day psychic Nate Silver’s come in for a bit of a kicking following his failed predictions).

Herein lies the big problem: we look to technologies to reduce our fear of uncertainty in the future. We focus on the successful examples, the best practice. Strangely no-one is using Big Data analytics to select their Big Data implementations, but rather we are making decisions based on faith and flawed data. And it’s exactly those big decisions, based on faith and flawed data, that are the ones that have the chance of a transformative impact on an organisation. Decisions based on pouring over past data tend to be the ones that lead to mere incremental improvement.

(Photo by Martin Rose/Getty Images)