“Do you expect me to talk?”

I’ve now been retired for a little over a year, but that doesn’t mean that I’ve lost interest in testing and its many ramifications for builders and users of systems. Of course, testing still remains in the news, in the form of ongoing cases involving IT systems. Although other cases have arisen over the past year – the most recent is the overpayment of Carers Allowance by the British Department of Works and Pensions (DWP), and their insistence that people struggling with caring responsibilities for family members and working hard to bring in some small earnings nonetheless repay overpaid benefits that they may have received despite complying with all the DWP’s requests for information – the Horizon case is still making headlines as the public enquiry into the scandal continues to bring key players from within the Post Office into the spotlight.

A recent LinkedIn post from forensic psychologist Ian Ross looked at the cycle of groupthink that seems to have infected the Post Office. It started either with Fujitsu people not wanting to own up to their own testing or their system coding (or both) being inadequate, or with Post Office people not wanting to admit that they’d spent big money on a system that wasn’t fit for purpose. Setting the story running that this could only be down to crimes snowballed with a toxic mix of groupthink, misplaced belief in technocratic infallibility and backside covering as the Post Office got further and further down the rabbit hole.

Thinking of the testing implications, I’ve now coined a new heuristic, which I call the “Goldfinger heuristic”. At the beginning of the novel of the same name, Ian Fleming wrote “Once is happenstance, twice is coincidence, but three times must be enemy action.”

In this case, the enemy is the real world. By the time you get to three independent reports of a problem or unexpected outcome from a system, it’s time you at least considered that it might just be the system that’s at fault, not the user.