Numberwang on the web

The Guardian has a Data Blog with the subtitle Facts are sacred..

Tempting as it is to wander round the epistemological byways here (What’s a “fact,” for a start?) I’ll spare you that. Instead I’ll express bafflement about the factualness of some specific facts on government websites posted yesterday.

A decorative graphic has blobs to represent the costs of UK government websites. There’s a clickthrough button to get the data from Manyeyes.

The data is shocking, on first view. For instance, it seems to have cost the UK taxpayer £154 for every visit to the Cabinet Office site. Cue horror. Except that these appear to be made-up numbers, with the calculations done on a solar powered calculator sitting at the bottom of a deep well.

Here are a few columns that I took from the data, applied simple maths to and lo! most sites cost a few pence per visit.

I’ve left the original numbers so you can work out costs per visit yourself. I’ve shown my calculations on a blue background.

Details from the govt webcosts spreadsheet

Details from the govt webcosts spreadsheet

I fear I’m doing the post a disservice when I scroll down and find a link to Google docs that offers the full data. Download the full list as spreadsheet doesn’t work for me but, when I look at the data on this sheet, it makes a lot more sense. Costs per visit are down to pence rather than tens of pounds. Not that I can replicate this either. But I get a lot closer.
UK govt web costs, as held in google docs

Ah ha: At the end of the Guardian post there’s a caveat:

UPDATE: an error in the cost per visitor column has now been corrected

Sorry, not when I looked at it it. Which, surely must be after it’s been corrected, or else I wouldn’t be able to see the correction note.

In any case, the original costs are dubious. Some of these sites must be new and have not got a full year’s data. The basis on which any given department has costed its sites might be wholly individual. I am forced to conclude that these are pretty well just made up numbers from start to finish. Although that doesn’t excuse the failure for anyone to notice that a simple division has been borked.

Any reader would gasp at the idea of sites that squander 3 week’s unemployment benefit on each visitor. That “information” sticks in the mind. Few readers would be finicky enough to look at the detail.

It says Facts are Sacred in bold letters at the head of the post. You could be forgiven for absent-mindedly taking it that someone had made sure that these were sacred facts.

A paranoid person might easily assume that the release of such spurious data was part of a propaganda offensive to convince UK voters that the public sector is so bloated and wasteful that the planned public sector cuts will not affect anything important. They’ll just involve stopping stupid and wasteful spending….

Like the imaginary huge (6-year-salary!) redundancy payments to inherently idle public sector workers that another Guardian blogger treats as a representative of the true position. Then characterises the head of a public sector union as a rabid militant for opposing the destruction of his member’s jobs and employment conditions.

I initially, and charitably, assumed this was a random Comment is Free post, maybe by a freelancing Daily Mail journalist.

But it appears that the writer is a senior figure in the Guardian. He was the Guardian’s political editor for 16 years. From his profile:

Michael White is assistant editor and has been writing for the Guardian for over 30 years

I see I have been spending £300 every ten minutes on buying the Guardian and it’s only ever worth it on the increasingly-rare days when Charlie Brooker is side-splittingly funny and Marina Hyde isn’t engaged in electoral enthusiasms for former presenters of shows about amusing vegetables. I begin to think a 40% cut in my Guardian spending is long overdue.