Saturday 28 August 2010

I'm off...

...in a few hours. We're setting off for a week in Devon.

Which means that over the last few days, I've had to sort out my year-end billing, clear all my deadlines over the next week and a bit, and run negotiations over a potentially major new project (on which my lips must be sealed for a while). Oh, and fought off a nasty cold - which, as I am male, means that I have been suffering pain and discomfort greater than any other known to mankind (except possibly childbirth).

So I'm sorry for the comments left unanswered!

But now, the sky is bright, the rain has stopped, I have no work to do for a week, and my cold seems to be lifting. See you all in a week!

Saturday 21 August 2010

Confused...

Today was productive - I cleared out a corner of the garden that had been lost to weeds, junk, old discarded toys, and general rubbish. It is now clean, tidy and (more importantly) navigable on foot without protective clothing. We also had a nice crop of blackberries for lunch, as they seemed to good to waste.

Talking of waste, the rubbish fell into several categories. The overgrown plants are now on the compost heap, and in three years will be doing good elsewhere in the garden. The brambles and other weeds are safely in the green bin - I want them out of the garden for good.

Then there is the general waste. It is currently occupying the (entire) rear load space of the landrover, and will be heading for the local tip tomorrow.

Now, can someone from British local government please hurry along to explain to me the following. I am going to fire up the landrover tomorrow, consume diesel, and emit CO2 in order to take this domestic rubbish to the tip. A few short days later, a lorry is going to go past my house and collect other rubbish and take it (ultimately) to the same tip.

The lorry is not allowed to take the rubbish from today's clearout, because my allocated quota of rubbish is insufficient to allow this. This is because the rubbish quota is a fixed size per household, regardless of the size of the household, the house, or the council tax bill that applies to it. Therefore, the lorry that will be coming past anyway is not allowed to take the rubbish, and it must be conveyed by an additional journey.

Now, all of that I can understand. What I do not understand is, how did you get that through on environmental grounds??

Thursday 19 August 2010

Lies, Damn Lies, and Speed Cameras

I seem to remember writing a post a few short days ago bemoaning the willingness of the BBC to over-interpret some completely unreliable data from Oxfordshire's Safety Camera Partnership.

I may have made a silly mistake. I assumed that the data, although obviously unreliable as it was based on a hopelessly small sample, was true. I assumed that the Safety Camera Partnership was staffed by honest civil servants doing what they genuinely believed to be the best course of action for road safety. I assumed that the SCP staff were not lying spin doctors peddling untruths in order to whip up public sentiment in a desperate attempt to regain their budget and line their pockets at the expense of taxpayers.

My assumptions may have been right. However, consider this. When the local rag asked for the data underlying the claims of rising numbers of speeding drivers, the SCP said that
"those figures were not readily to hand"
which is, of course, odd given that a few days earlier the SCP had issued a press release that relied on them.

The reporters did manage to get hold of them, though, which is even odder when you consider that the figures were "not to hand". In the case of the camera at which the SCP said speeding offences had risen 18%,
speed offences actually fell by four per cent [...] during five days of monitoring since the switch-off on August 1, compared to offences committed between 2008 and 2009.
Very odd indeed.

One for Albert

The UK is an Anglican country, no? It must be, surely* - by definition?

Maybe not:
A survey of 37,000 churches, shows the number of people going to Sunday Mass in England averaged 861,000, compared with 852,000 Anglicans
Ooops.

-----------------------------------------------------------------------
*All together now: "It isn't, and don't call me Shirley...."

Tuesday 17 August 2010

Is Global Warming Dead?

A statistical study on temperature proxy reconstructions has been submitted to the Annals of Applied Statistics.

I'll admit, that's not the catchiest of introductions to a blogpost. But bear with me - this is important. Seriously.

The Annals of Applied Statistics is apparently one of the “top statistical journals in the world, although I can't vouch for that personally. The article in question is listed to be published in the next issue. From that, we can infer that the article is a good one. If you can't wait for the next issue, of if (like me) you don't subscribe to that particular journal, a copy can be downloaded from here. I encourage you to do so; it is very readable, and not just by the standards of statistical journals.

The article looks at the statistical quality of the correlation between long-term temperature proxy data and actual historic temperatures. That still sounds fairly dull, even by the standards of statisticians (who, generally, are the mathematicians who were too detached from reality to become accountants). However, this is a crucial subject. It is, I think, about time that I explained why...

We all know that Global Warming (GW) is a huge issue, and is being used to justify many and varied laws and policies. Politicians neatly side-step questions of whether GW is real, by pointing to an impressive array of scientists who tell them that GW is real. That enables the politicians to treat GW as fact and not enquire any more deeply. So the justification for the political influence of GW rests on the scientific foundations of GW.

So, how does a theory gain scientific acceptance? Simple - by use of the "scientific method", a term that many have heard of but surprisingly few understand. I say "surprisingly", because it really is very easy to understand if anyone actually makes the effort to explain it. In short, the scientific method requires that you think up an idea about how things might be (you can call this a "hypothesis", if you like impressing people). Then you go and see if things really are like that (again, amongst those that like to use fancy words, this bit is called an "experiment"). Note the order of these two steps; first you develop the theory, then you go and test it.

Testing is quite a hard step, in fact. You need to design your test so that it is only testing your idea, and is not influenced by other possible factors that might be varying at the same time. For example, if my theory was that readership of a blog increases with the rate of posting, I could do that by simply writing more posts - but I'd have to make sure that all of the new posts were of a similar length, quality, readability and so on compared to before. I would need, in short, to control all the other possible factors so that if I saw a change, I could safely attribute the change to the factor that I was deliberately varying. Common sense, really.

So the basic principle is that you develop a theory, and then you go and see if it is true by way of a test. In the case of GW, the theory arose from computer programs that tried to model the way in which the Earth's atmosphere worked. They suggested that increasing levels of CO2 in the atmosphere led to an increase in global temperatures that could become uncontrollable. That is obviously not a good idea, if it is true. But is is only a theory at that stage. We need to test it, to see if it is true. Until then, we cannot claim that it is a scientific truth. All we know if what some equations running on a computer tell us - we can't have any confidence that this says anything about the real world until we test the theory out.

The easy way to test it would be to pump lots of CO2 into the atmosphere and see what happens. However, as the point of validating the theory is to find out whether we need to stop emitting CO2, that approach is problematic.

An alternative would be to use a different planet - pump that full of CO2 and see what happens. Or, extract all the CO2 from its atmosphere instead. However, we don't have one within handy experimental reach.

So, the only test of GW is to look at the past, instead. We know that CO2 emissions have increased hugely over the past centuries as industrialisation has progressed*. Therefore, GW predicts that global temperatures should have been steady for centuries until the Industrial Revolution, following which they should have increased steadily up to today. All we need to do is look at the varying temperatures over that period.

There is a snag, though. We have only had (a) decent thermometers and (b) people who obsessively record temperatures since about 1850, whereas to prove the point we have to compare with the temperatures before then. So, we have to use a "proxy", i.e. something else that we can measure, which correlates with temperature.

There are, gratifyingly, many and varied proxies. Between them, they cover the last thousand years or so. Sadly, they all give us different sorts of measurements - tree ring data gives us thickness measurements, ice core data gives us isotopic ratios, and so on. We need a way to convert these to temperature data in units we understand. Fortunately, there is a way to do this, because there is a period of overlap in the 150 years from 1850, where we have both hard thermometer data and proxy data. So what you have to do is plot the proxy data for this period against the actual temperatures and use that to work out what the relationship is. Obviously, there won't be a perfect one-to-one correlation, due to experimental error and outside factors. So you need to apply statistical methods in order to work out the way in which they are related. A computer can do this, provided that you program it correctly. Then, you apply the program to the old proxy data, and it gives you temperatures going back a thousand years.

Then, you have the temperature data that you need in order to validate (or refute) the prediction made by the atmospheric models. If that data validates the model, you have a scientific basis for saying that the model accurately reflects the influence of atmospheric CO2 on the Earth's climate. With that basis, you can predict that unless CO2 emissions are curtailed, disaster will follow.

On the other hand, if the temperature data does not show a steady temperature in pre-industrial times followed by a dramatic rise since we started to emit CO2, then you have a problem. Specifically, your atmospheric model has been shown to be wrong. I'm emphasising that word - wrong - because there are plenty of weasel words that can be used, but in the end they all boil down to the model being wrong. If the model is wrong, its predictions are also wrong (the GIGO principle), and the basis for years of climate hysteria suddenly vanishes.

Now, if you were alert, you will have noticed a small proviso: I said that a computer can do the necessary correlation work, provided that you program it correctly. Up to now, that programming work has been done by climate scientists. The significance of the paper that I linked to way up there at the start of this post is that a pair of statisticians have pointed out that this is a statistical question, not a climate science question, so surely it should be done by statisticians? Not only that, they have also re-done the work, using the proxy data provided by climate scientists**.

Their findings?

First, they have a number of criticisms of the manner in which the climate scientists approached the statistical issues in question.

Second, when the maths is done properly, the temperature data does not support the GW hypothesis.

I'll say that again: the temperature data does not support the GW hypothesis.

Their results show that the theory of Anthropogenic Global Warming is wrong.

Now, this post is titled "Is Global Warming Dead?", not "Global Warming Is Dead", for a reason. That is, there are several comments in the paper which are carefully worded so as to remain scientific in nature, but which could be seen as somewhat barbed. I am concerned that the authors may have approached the issues with a pre-existing opinion on GW - in which case the risk of confirmation bias applies to them (and me) just as strongly as it applies to many climate scientists and GW activists.

But whichever way you read it, this paper is a huge problem for GW.

-------------------------------------------------------------------------

*Or do we?

**this, of course, shows why the CRU's refusal to release its data was so utterly shameful

(Hat Tip to the
Watts Up With That? blog)

Wednesday 11 August 2010

Speeding to a conclusion

I suppose I should really say something about the mass switch-off of safety cameras.

I regard this as a really exciting thing, in fact. Not because I'm less likely to be prosecuted (there are still plenty of speed traps in Oxford), but because it will provide an excellent experiment. For years, those of us who opposed cameras claimed that the statistical evidence which purportedly supported them was illusory, while those whose livelihoods depended on them claimed that they were responsible for saving countless lives. Now, they have been suddenly and abruptly switched off in one part of the country, while remaining in use in others. We will, in time, be able to compare, directly, the change in accident and fatality rates in Oxfordshire before and after 31 July 2010, and put this alongside any change in the same rates (measured in the same way) in other counties.

This is a statistical gift to the nation and to the cause of road safety. If done properly, it could actually answer the question of whether speed cameras are a good idea or not. Swindon gives us a hint as to what might happen, but (even there) it is too early to say.

Which is why I am very annoyed by this tweet and the news story behind it. I did in fact hear the full story on the radio yesterday, and it concerned a reported who went with a BBC reporter to the site of 2 speed cameras in Oxfordshire. It seems that although the cameras are not active (they are not putting film in them) they are still switched on and therefore record the number of times they would have fired had they been loaded with film. From this, the camera operator was able to tell the BBC reporter that one site showed an increase of 88% and the other showed an increase of 18%. Asked to comment on this, the operator heavily qualified his reply by saying that if this was representative, then more people could be expected to die in accidents.

This then becomes the headline:
The number of drivers speeding past Oxfordshire's deactivated speed cameras has increased by up to 88%, a road safety partnership claim.
First, they do not claim this. They report that one camera showed an increase of 88%. Note also that they report that the other showed an 18% increase. Second, this is after 5 days, at 2 sites out of 72.

This is, therefore, hardly representative. A hopelessly short period, a pathetically small sample, and an overblown interpretation give, as ever, a lurid headline with which to bash those horrible nasty people who use cars. I hope that the research that is eventually done is reported more responsibly, more competently.

Now, as it happens, I know both camera sites. The 88% site is one which falls into the category that might politely be described as having a somewhat lower speed limit than might be expected given the conditions. The 18% site is in a 30 zone that is 100% justified. If the results are at all meaningful then it is, of course, rather interesting that drivers seem mainly to be ignoring the less-obviously-correct limit.

Perhaps we will not see the slaughter of the innocents that the safety camera partnerships predict. I hope not, for several reasons.

Update - Here's a much better suggestion...

Wednesday 4 August 2010

Healthy debate

The Raccoon has been thinking about healthcare. Worth a few minutes of your time, I would suggest.

Oh, and she's spotted that smoking is good for you*. That should set the cat amongst the pigeons. Unless they just decide to ignore the medical evidence, of course.

(*in some respects, and within certain boundaries...)

Tuesday 3 August 2010

Introduction to Socialist Business Methods, Part 1

Call me callous, but I find this just hilarious:
Northern Rock Asset Management (NRAM), the so-called "bad bank" part of the old Northern Rock business, has reported a return to profit.

NRAM holds most of the rescued bank's old mortgages and unsecured loans.

Pre-tax profit came in at £349.7m for the first six months of the year, compared with a loss of £724.2m in the same period last year.

However, Northern Rock PLC, the "good bank" holding savers' deposits and new loans, made a pre-tax loss of £142.6m.
Right, let's get this clear. New Labour split the Rock into a good bit and a bad bit. The good bit is the viable part of the business, which can operate safely and will be attractive to investors. The bad bit is full of all those dodgy loans, and would stay with the Government who would bravely shoulder the burden for all of us and do other good things that only The State can do because The State is a wonderful and bountiful thing that will save us all from ever having to make a decision or take a risk. Or something.

So now, the bad bit is making thumping great profits and the good bit is making a whopping loss. It strikes me that there are two possible explanations.

The first explanation is that New Labour really do have no idea whatsoever as to what makes a good business. They seem to have decided that a business which holds lots of depositor's money on which it has to pay interest, but has very few loans on which it can earn interest, is a viable and tempting investment and will make a profit. Robert Peston calls this a "paradox", which tells us all we need to know about at least one of his economic skills and his political leanings. This leads us to the conclusion that New Labour believe that economic safety lies in having lots of spending liabilities and no clear & sufficient source of income. This is, it has to be admitted, rather in line with other evidence.

However, there is another explanation. This one is simpler, and Occam's Razor suggests that we should therefore prefer it. It is that New Labour are a bunch of incompetent idiots who couldn't organise an (ahem) party in a brewery.

Chose whichever one you like. Or both. I don't mind.