EuSpRIG, the European Spreadsheet Risks Interest Group, is holding its 2023 conference in London on Thursday 6 July.
It’s the first gathering since 2019 until covid rudely interrupted them. These conferences had been held every year since 2000, making them the longest running of the gatherings devoted to Excel, spreadsheets and modelling.
When it started, such things were rare. Now there are many of them, but Eusprig is the one I take trouble to attend regularly. Its focus is not on the latest whizz-bang features of Excel, which wow less jaded members of the audience and induce yawns from those already abreast of them. Rather, it concerns itself with the propensity of humans to make mistakes using spreadsheets among many other tools.
And yes, these lapses come from basic properties of the way brains work, not from deficiencies in the design of spreadsheet software, much as many find it convenient to blame Excel for their own blunders.
Jumping to conclusions before being entirely certain that they are well-founded can be a life saver. When a relevant conclusion was that the rustling in the grass might be an approaching tiger, people who acted on that thought got to live, and to breed. More hesitant humans left less of an impression on the gene pool.
An evolutionary adaptation that is handy with charismatic megafauna is less helpful with spreadsheet models. It leads us to over-confidence in our analysis. As a representative of one of the leading firms providing formal audits of project finance models, this is a subject of the deepest interest to me.
Eusprig features contributions from academics researching how these instincts menace the engineering of software of all kinds, spreadsheets being just a special case of a more general principle. It combines them with support from practitioners in finance and industry who grapple with analytical risk on a commercial scale.
This year’s keynote is by Andrew Gordon and Jack Williams of Microsoft Research, covering the AI tools that Microsoft is integrating with its products. Microsoft is in a unique position to deploy this technology as the largest early backer of OpenAi, the foundation behind Chat GPT.
At their current stage of development, large language models have a tendency to hallucinate, confidently fabricating factually incorrect responses. Simon Thorne, of Cardiff Metropolitan University, will explore the risks that spreadsheets generated by these AI tools will be flawed.
Also presenting will be the always engaging Tom Grossman, Professor and Chair of Business Analytics & Information Systems of the University of San Francisco’s School of Management.
Eusprig’s 2023 Conference takes place at The Foundling Museum, 40 Brunswick Square London WC1N 1AZ, on Thursday 6 July.
I’ll look forward to seeing, for the first time in years, many regular attenders who have become old friends, as well as, I hope, very welcome first timers.
