Report oozes bias but still 18% never turned off times 118 idle hours times power use times carbon (or money) times number of workers = a big number

A potentially interesting (and widely quoted) report disappoints with sloppy science and failure to recognise bias.

The UK’s National Energy Foundation and 1E‘s survey of energy use by computing has some big numbers. The crux of the report is that if software systems were introduced to manage computer idle time (ie turn them off), then the UK would save lots:

the use of a product…. to switch off networked PCs centrally could save up to 1.5TWh (1.5 billion kWh) per annum, with a value around £115 million. In carbon dioxide terms this would equate to potential savings of 700,000 tonnes of CO2 (equivalent to almost 0.2MtC). This single action could make a significant contribution to UK national targets.

The report aimed to establish the “percentage of UK workers who fail to properly shut down their computers at night, and the motivations behind this behaviour” and to calculate the impact of this , both for individual business and nationally.

The report first briefly reviews climate change and Britain’s policy responses.

The Energy Review also draws attention to the specific issue of standby losses, noting that in 2004 an estimated 8% of residential electricity consumption was used on standby, and that the proportion was rising.

and

Non-domestic electricity use by ICT equipment has increased by over 70% between 2000 and 2006.

They then propose scenarios to predict future use. One of these, the Earliest Best Practice (EBP) model relies on vigorous implementation power management “simultaneously implemented across the PC population”.

They examine barriers to turning computers off at night. Although 70% of people claim to turn their computers off each night (this is higher than I would have thought but is glossed over), 22.8 leave their computers on at least three nights per week.

Why?

17.5% because it’s a hassle
10.4% say no one else does
9.8% say it’s not important
7.7% say they forget in the rush to leave
3.3% say they’re out of the office and don’t return
1.8% say they worry they’ll lose work

At this point the report seems to lose almost all objectivity. It is quite dismissive of respondents’ quite valid reasons for leaving computers on:

…variations on “It’s too slow logging back on in a morning”, which must, I suppose be linked to the hassle factor. (But just how much hassle is it to switch off a computer).

and

And in another surprising result, only those aged over 35 reckoned that they were told to switch off their computer “every night”

and

Another respondent seemed to believe that anti-virus scans only ran overnight.

These little asides totally distract from the value of what we could learn from people’s perceptions. It is also disappointing that the report relies heavily on self reported behaviour rather than attempting to measure actual behaviour.

In the last section, the report takes the findings from the reported behaviour study and calculates, using a ‘big leaf’ approach the effects of this behaviour on individual PC, company and nation. 18% never turned off times 118 idle hours times power use times carbon (or money) times number of workers = a big number.

With the exception of a paragraph of advertising on the about page (p28), nowhere in the report is there a disclaimer about potential bias. Instead, 1E’s product “Nightwatchman” is referred to as the miracle cure at least eight times. This is unfortunate and distracts from the report, rendering it marketing and giving people who disagree with the premise a very real chink in the armour. This is Techworld’s take:

The report is part of a 1E marketing effort for its NightWatchman software product, a quasi-sales ‘brochure’ in effect. The product reduces energy costs by automatically shutting down PCs according to a centrally-controlled schedule. Also, by reporting on company-wide PC energy consumption, NightWatchman enables facilities departments to understand, control and hopefully reduce energy needs.

…

Very fortunate for 1E as the favoured tool is its own NightWatchman product.