Thursday, 2 September 2010

Would You Trust Your Life To Excel?

Initially, I wanted to post this here, but, whilst ranting, I realised I had (much) more to say than my self-imposed format allows. So, in all its righteousness, here's my rant on the press release touting spreadsheets as diagnostic tools.

Let me first comment on the very title of the press release: "Microsoft Excel-based algorithm predicts cancer prognosis". Now, I am sure the scientists in question really used Microsoft Excel, but does it mean that the same algorithm could not be implemented in a different tool? I'd like to know if Microsoft paid for this advertisement... sorry, research, or are now just smiling into their lattes.

Secondly, and probably more importantly: at least in the IT industry the potential pitfalls of using spreadsheets for any serious calculations are too well known. And, while messing up big time in business and finance is bad enough, messing up when it comes to life-or-death decisions is incomparably worse.

It's fine to develop your initial algorithm in any way, and using any tool that's handy, and you know your way around (even if I suspect the scientist in question may not be as wizardly in Excel as they imagine). Then, once you have it, it's time to develop an application that can be properly designed and tested using best practices of software engineering. In case of medical software, I'd even argue that the same safety rules as apply to X-ray machines and other "dangerous" medical instruments should apply even to such "soft" diagnostic tools.

Finally, to answer my own question from the title: NO.

I just hope I get a choice when my time comes...

This is all not to say I'd trust my life on ANY spreadsheet. 
But I'd still have a particular issue if it were Excel... ;)