Spreadorist

Interested to read in yesterday's Guardian about the erring spreadsheet. It seems that a major plank in our Chancellor's evidence based policy making has got a touch warped. Or maybe a touch of dry rot. It seems that Professors Rheinegold and Bogoff late of the IMF and other illustrious institutions got a bit overconfident and published the spreadsheet containing the evidence on which some important policy advice was founded. A rival gang took the spreadsheet apart and found various errors, errors which did the evidence no good at all.

But I do have some sympathy for the careless professors. Excel is a very powerful tool and one can use it to build more or less arbitrarily complicated spreadsheets. The catch being that all the fun is in the building and it is all too easy to ease up on the boring old business of testing. Even those academic drudges who used to be called research assistants but who probably now have a grander title might baulk at it. Not part of my job description guv..

At about the time I packed up work, the drill was that you hired testers whose only business was to test and so test they did. It almost got to the point where you paid them by results: £A times the number of severity 1 errors plus £B times the number of severity 2 errors and so on and so forth. And if you did not keep them under control, they could get you under control, more or less taking over the management of the development cycle for the fattening of their budgets and pay cheques. But I doubt if an economic professor would care to burn up his budget in this way - it is not as if he has any penalty clauses hanging over his work. Maybe he just relies on good old academic rivalry to push any errors up to the surface, one gang being only too pleased to be able to trash the work of the other gang, although if they have good manners they try not to let it show too much what a bang they get out of said trashing. Much wailing and many crocodile tears instead.

There was also mention in the Guardian of a gang called the 'European Spreadsheet Risk Group' (see http://www.eusprig.org/) which made me wonder whether any thought was given to the risk of a disgruntled Microsoft employee or contractor putting time bombs into the Excel code itself. Something which would sit there, inactive, until triggered by something or other, when it would start corrupting execution in some reasonably subtle way. So that big spreadsheets slowly degraded, putting out all kinds of slightly duff stuff in the meanwhile. I would have thought that such code could be quite hard to detect; it would not need to be bulky and it would not need to do odd things with core data. Let us hope that our Security Service is on the case.