To err is human, to forgive divine, but to correct and learn from error is the route to improve quality.
Philosophers, psychologists, and ergonomists have spent lifetimes classifying and analysing human error; if you want a sound and readable introduction, I can recommend any of James Reason’s excellent books. Broadcast media revel in errors, from the first news bulletin of the day to late-night satire. Hardly a play or movie reaches stage or screen without a plot that relies at some point on human error. A considerable proportion of our national output is consumed by errors or their correction. Some of us have the misfortune, perhaps an error of circumstance, to have to spend much of our working day detecting or correcting errors.
I am the first to admit that I have made mistakes, in answering questions here, and previously in MacUser. Thankfully you are usually quick to correct those making it to print or online, and I try to make those corrections as quickly as I can.
The most dangerous errors of course are those that remain undetected for too long – so long that correction becomes tough, costly, or even impossible. The most expensive and embarrassing examples have been NASA’s space calamities, from exploding Shuttles to planetary probes that misjudged distance. Every bit as complex, some of the long-standing security flaws in Windows could prove far more damaging in the long run, with their continuing toll of botnets, identity theft, and organised crime.
One example of protracted detection was the extraordinary bug in Microsoft Excel 2007 that displayed 100000 as the result of multiplying 850 by 77.1, 20.4 by 3212.5, and others – simple tasks that should return 65535 (or in other cases 65536). In case you are worried, I have just checked the current releases of Excel 2011, Numbers, and the pre-release version of Excel 2015, and none has reverted to this former behaviour. Yet.
This was the more damaging as it was not picked up during Microsoft’s testing of Excel prior to release in 2006, but reported on 22 September 2007 when a user noticed the error ‘in the wild’, as the gurus might say of a security issue. Although heartening to see that Microsoft professes that it takes “calculation in Excel very seriously” – as if calculation was an optional plug-in, not its raison d’être – the patch to repair this pervasive flaw was not released until 9 October 2007.
More sinister still are errors that are detected but not admitted to, perhaps in the vain hope that they will remain hidden until those responsible have moved on, away from the resulting blamestorm.
A culture of blame or litigation cannot support honesty and transparency about error.
A society that has no concern for improving standards, that does not encourage the admission of error, cannot foster quality management. If I fear disgrace or dismissal when my error is detected, then I will naturally do my utmost to conceal it, or maybe try to frame someone else to carry the can. And when the cock-ups become clear, if we cannot investigate thoroughly why they happened, we cannot learn to prevent their recurrence. Punishing error thus perpetuates error, lining up more victims for the future, instead of improving.
Moving away from a culture of blame does not absolve anyone from ownership, though.
It seems all too common for those who have mismanaged situations to remain in office long after those that they have mismanaged have lost all confidence in them. Managers, executives, even ministers often act not as if the buck stops there, but to deny the very existence of a buck. Whether in healthcare services or mortgage lenders, we end up with organisations that are rotting at the core, with their officers devalued and their values debased.
This article looks at some of the ways in which you can reduce the errors occurring in spreadsheets, databases, web pages, and elsewhere. Remarkably few important worksheets or forms contain even the most cursory of checks on data that are entered, and most of us appear content to trust essentially untested formulae for business-critical purposes.
Hopefully some of those tips will save you from expensive errors.
Updated from the original, which was first published in MacUser volume 24 issue 01, 2008.