E-Day result

Well, it was a bit disappointing perhaps, but E-Day made back some ground to finish just 0.1% above the daily average of energy consumption.

Or did they?

Here's the screenshot I took of their energy consumption graph at about 7:30am this morning. Note how the nice green E-Day line is generally above the nasty red average consumption line, showing that E-Day began with above-average consumption:



And here's the equivalent screenshot from just after 6:00pm when the 24 hour period had finished:



See how they have clawed back ground from 1.5% above the daily average to just 0.1% above - not a roaring success, but not such a humiliating failure as it seemed it might be earlier today.

But look at the two graphs. On the second one, the difference between E-Day and the average has disappeared for most of the graph.

I noted on my earlier post that although the current figure ran at 2% to 2.5% most of the afternoon, the total difference shrank from 1.5% to 0.1%. How could that be? And how could the difference in consumption in the first 12 hours disappear by 6:00pm this evening.

I can't see any explanation other than a massaging of data. But I'm mailing the organisers to ask how they account for this:

Hi,

I took screenshots of the E-Day site at about 7:30am this morning and then just after 6:00pm this evening. The first graph showed a clear difference between E-Day consumption and the average. On the second one, this difference, during the first twelve hours, had vanished.

How could that be?

I have blogged this here (link), and would be glad to add your explanation to the post.

Under the circumstances, I think it's fair to regard this correspondence as open and I have pasted the text of this email into the blog post.

Regards,

Peter Risdon.

UPDATE: Incidentally, this does ring a bell

UPDATE 2: Thanks to anonymous in the comments for the news that the E-Day site now carries the following explanation:

E-Day did not succeed in cutting the UK's electricity demand. The drop in temperature between Wed 27 Feb and Thurs 28 Feb days probably caused this, as a result of more lights and heating being left on than were originally predicted. The National Grid refined their assessments, based on actual weather data, during Thursday afternoon but I am afraid that E-Day did not achieve the scale of public awareness or participation needed to have a measurable effect. I will do my best to learn the relevant lessons for next time. Thank you to everyone who helped me or left something off specially as their contribution to E-Day, and this Leave It Off experiment. Please enjoy E-Day's solution, video and science sections which all worked well. Warmest regards, Matt

This is unintentionally funny. A site based on the proposition that it's possible to model accurately a chaotic and half-understood system like the weather to the degree of accuracy necessary to attribute temperature changes of fractions of a degree over decades to a specific, human generated, cause had to adjust a line on a graph during the afternoon because the predictions of temperature, based on these models, for a very short and immediate time frame - one day, predicted a day in advance - were inaccurate.

It's also gibberish. To paraphrase: we didn't succeed because it was colder than we expected and people used more power. And we adjusted our analysis to compensate for this change in temperature and so make it irrelevant.

I'm trying to track down data for the temperatures yesterday and today (the Met Office don't have directly comparable figures that I can find yet) and for the effect of temperature on power consumption. More later. Probably.