Wednesday, 25 December 2013

Happy Spreadsheet Christmas

Cartoon Description

Santa and his reindeer are silhouetted against the moon, overlooking the Sphinx and two pyramids, one with a palm tree in front of it. The nearest reindeer is saying, "Er, Santa? Was it wise to program the route-finder using a spreadsheet?".

Saturday, 21 December 2013

Spreadsheet Evolution, Flared Trousers, and Colour in the Depths of Winter

This post wasn't going to be about spreadsheets, and then I realised that it would be. This is the first time I've created a blog in Blogger, so I was pleased to remember that there's a Blogger blog that I like, and could use as a guide to designing my own blog and arranging its layout "gadgets".

The blog in question is called Saaibestrijding, written by Paul Tieman from Maastricht. His blog's name is a Dutch coinage meaning "struggle against boredom", and his blog is mainly photos showing unusual and colourful ways for men to dress. I like it because I like colour, especially in the depths of a British winter when everyone is wearing brown, beige, grey, and black. I also like comfort, and some of his clothes look very comfortable.

So Paul's blog helped me lay out my own, but surely it has nothing to do with spreadsheets. Then I remembered that I once cited it in an essay called "Dress Code", which I wrote for the Dr Dobbs computer magazine after the 2009 European Spreadsheet Risks Interest Group conference in Paris. It had been very hot, and yet some of the people at the conference didn't like me wearing shorts, because it didn't convey a professional image. In  "Dress Code", I was commenting on ergonomics versus business.

And I was also writing about an earlier EuSpRIG conference, the 2002 one in Cardiff. Now, the first modern spreadsheet was created in 1979. Since then, technology has made spreadsheets look much nicer, but it is still hard to build reliable software with them. As you can tell from this blog and my Indiegogo campaign, that's one of my research interests, and it's also the problem that EuSpRIG was founded to attack.

So in the 2002 conference, to emphasise how bad spreadsheets still are, I wandered around Cardiff's arcades until I found a retro clothing shop, bought a pair of extremely flared 1970s bellbottoms, and with marker pens and fluorescent yellow card given me by the shop's owner, made a lapel badge reading Spreadsheets have not evolved since flares were last in fashion. The "last" was hastily inserted when I looked around and realised that flares were now in fashion for the second time.

I wore this badge to the rest of the conference. And I could still wear it today, because although Microsoft have made Excel bigger, they still haven't equipped it with the features that programmers of other languages expect to use for building reliable programs.

Launching my Campaign

I've started this blog to accompany my Indiegogo campaign to fund a project on spreadsheet safety. I want to replicate a famous economic-modelling spreadsheet, the "Reinhart-Rogoff spreadsheet", using tools I've developed for writing modular, easy-to-read, spreadsheets. This will deliver (a) a modular economic model, and (b) a demonstration of how to write spreadsheets that are modular, easy to read, and thereby safer than spreadsheets built using only Excel.

Why do I want to do this? The spreadsheet I'm interested in was written by economists Carmen Reinhart and Kenneth Rogoff for their 2010 paper "Growth in a Time of Debt". They examined the relation between debt and economic growth in forty-four countries spanning 200 years, and concluded that in countries where debt exceeds 90% of gross domestic product, economic growth is notably lower.

But early in 2013, Thomas Herndon at the University of Massachusetts tried to replicate the spreadsheet. He couldn't, and eventually discovered a programming error. Reinhart and Rogoff had accidentally omitted five countries when calculating an average. Because their paper's conclusion is so relevant to governments' austerity programmes, journalists pounced, writing features with titles such as "How not to Excel at economics" and "Grad Student Who Shook Global Austerity Movement". Critics said the austerity programmes were founded on a blunder. Reinhart and Rogoff even received threatening email blaming them for tax increases and public-sector cuts, as they describe in "Debt, Growth and the Austerity Debate", a New York Times article for April 16th 2013 in which they explain their paper and the error.

Actually, things are never so simple. The programming error may have caused only a small, not significant, change to the conclusions. And Herndon and his colleagues also disagreed with Reinhart and Rogoff's choice of methods for selecting data and weighting results, things that probably have much more effect.

However, I suspect that using my tools, it would have been harder to omit these countries. That is, harder to make the error. Moreover, it would have been possible to make the spreadsheet formulae — the equations of the economic model — easier to read, and easier to add explanatory commentary to. Had this been possible with Excel, the extra clarity could have averted much of the ill-informed and hurtful media criticism hurled at Reinhart and Rogoff. That's one of the things that I want to demonstrate with this project.

[Update January 24th 2014] The Christmas and New Year break delayed my publicity. So that I can make it coincide with a launch date that's safely out of the way of the holidays, I've relaunched my campaign at the new URL given at the start of this post.