Track Talk W3

How to Prevent your Next Software Release Ending in Disaster

Andrew Brown

11:00-11:45 CEST Wednesday 14th June

You have a major release scheduled. You have prepared a comprehensive plan, including contingencies, back-out routes, and go/no-go decision points. Yet, despite all your preparations, it all ends in disaster. You could even see the disaster approaching but were unable to prevent it. Instead, you watched in fascinated horror as unfolding events sucked you and your release into calamity.

Release disasters are mostly avoidable. Decision-makers usually have ample information available prior to the go/no-go decision point to show their current plan is no longer viable. They also possess a workable back-out plan and the means to implement it. Yet despite this, decision-makers doggedly pursue their planned course of action until beyond the point of no return, where failure becomes inevitable.

This behaviour is caused by a cognitive bias known as PLAN CONTINUATION BIAS, which is a leading cause of software release failures. It is also a leading cause of aviation incidents, particularly Approach and Landing Accidents (ALA). Within medicine, it is a prime cause of failed surgeries. It is also responsible for environmental disasters such as the Torrey Canyon oil tanker catastrophe.

In this session, we explore examples of aviation, maritime, and software release disasters to show how, when and why humans become vulnerable to Plan Continuation Bias. We uncover the human and situational factors present when it occurs, then explain why they drive plan continuation, rather than re-evaluation or other behaviours. We unearth the warning signs that will help us spot when we are at risk, how to safely warn a team of an unfolding disaster, plus we also learn techniques we can apply and safeguards we can implement to avoid our future software releases ending like the Torrey Canyon oil tanker.