Air force pilots in WWII got shot like crazy and suffered farcical
levels of fatalities; in an effort to save airmen, the Allies used
statistical analysis to determine where the planes that limped home had
taken flak and armored up those sections – which totally failed to
work. That’s because the planes that made it home had suffered
non-critical damage, so shoring up the sections where they’d been hit
had virtually no effect on the rate at which flak to critical sections
of the aircraft caused it to be shot out of the sky. In other words, by
looking at survivors rather than the dead, they were protecting the least important parts of the planes.
In a transcript of a speech given at Voxxed Days Belgrad, Daniel G
Siegel describes how this “bullet hole misconception” traps technology
designers – their survivor bias causes them to solve the problems that
are sufficiently unserious that they can survive until someone can
notice them. It’s like a customer service department that only hears
from the people whose experience is sufficiently bearable that they buy
their purchases and then complain – but they don’t hear anything from
the vastly larger numbers who never make it to the cash register because
of some process failure.
This is particularly exacerbated by the professionalization and
financialization of the tech sector. When there was no defined pipeline
into computing, people came from lots of different disciplines; when the
cash rewards for tech sector success were modest, people came because
of their passion, not their dreams of riches. The homogenization of tech
to people with engineering degrees who want to get really rich puts the
focus on a very narrow class of production and innovation techniques.
