Tuesday, November 24, 2009

If it looks like a duck, and quacks like a duck...

The CRU is now in full spin trying to limit damage. Apparently what they consider the most damaging is this email, which (presumably) they admit is genuine, given that they're arguing over its context.

The relevant passage is:
"I've just completed Mike's Nature trick of adding in the real temps to each series for the last 20 years (ie from 1981 onwards) amd from 1961 for Keith's to hide the decline"
Now, there are several different meanings of 'trick'. It could mean a deception, which is clearly what some took it as; it could also mean a skillful technique - like a 'trick shot' in snooker. Clearly Prof Jones would like us to believe it's the latter, and has provided two images and an explanation. This image is the one that was published (WMO1999), including the trick, and this one (which looks rather hastily prepared) shows the series separated.

First, looking at that second image, I'm going to point out a few things that seem rather glaring to me. Firstly, the scale is rather large - let's cut that down for now, as data prior to the instrument record is relevant if (and only if) the data for the overlapping years is a good match. Here we go:

The red/green/blue lines are the "reconstructions" - the CRU data - and the black lines are recorded temperatures. Now, it's a pretty close match in the middle - 1900-1950 or so:

But, um... it's not so good in the 50 years or so before that... in fact, it looks a bit like negative correlation:

Still, things were a bit rough in the Victorian era, so maybe we can forgive inaccuracies there? I mean, they didn't even have cars and aircraft, let alone the satellite temperature readings we have today! Clearly the more modern data is important.... Oops:

With accurate global temperature measurements, the divergence is huge. One of the series is going exactly the opposite way to the temperature, the other two aren't showing any warming. I'd say that's a pretty awful fit, really, wouldn't you?

I agree with Prof Jones on one point: "The ‘decline’ in this set of tree-ring data should not be taken to mean that there is any problem with the instrumental temperature data". Indeed it should not. When your proxy data disagrees with your real data, you reject the proxy - and that's just what we need to do. The tree-ring reconstructions here, while superficially promising for early-20th-century data, are clearly a very poor match and must be discarded as irrelevant.

Yes, they go back a thousand years - the bible goes back further, but that doesn't make it an accurate source of climate data!

Hang on, though. What about how this data was published - using "Mike's Nature trick":
The top circle on there is where the instrumented temperature finishes up, and the bottom one is (as you can see in the graphs above) where the data ends up: below the 1961-1990 average in all three cases. Yes, the data runs out before the end of the graph, but that's hardly a reason to add extra points to the series to make it finish well above the baseline. If I had to pick one word for that, it would probably be "fraud".

I'm not a climatologist, I'm not in posession of the full facts, and I'm not a professor, so I'm in no way competent to judge whether any dishonest practices have taken place. That's up to the relevant academic authorities, the journals, and the funding councils, and I look forward to hearing their decisions. But, for me, this whole thing stinks.

- KoW

Labels: , , ,

Sunday, November 22, 2009

The Day Global Warming Died

Friday afternoon saw a huge story break: the Climate Research Unit at the University of East Anglia has been hacked and a large quantity of email, raw data and (proprietary) climate model code has been leaked. The director of CRU, Professor Phil Jones, has confirmed the leaked information as being genuine.

The CRU is one of the leading lights in climatology and, in particular, paleoclimatology - reconstructing historical temperatures from secondary sources such as tree rings and ice core samples. Analysis of some of their data - released only after pressure from the Royal Society - has already shown significant anomolies.

The emails do not paint a flattering picture - in fact, they seem to show a pattern of deliberate fraud and abuse of the peer-review process with conflicts of interest, hiding models/data from review and manipulating editorial boards. These are not trivial charges and, if the emails are shown to be genuine, they should be sufficient to end careers and probably secure convictions for fraud.

As is traditional, this scandal has got a "-gate" suffix: Climategate. But, for once, it might actually be bigger than Watergate.

- KoW

Labels: , , ,

Saturday, November 14, 2009

Free Cheeseburgers!

According to this report on the BBC, the figures of 2000 calories a day for women and 2500 for men are inaccurate. The quotas are too low by about 400kcal - "a cheeseburger" - apparently, as we already take enough exercise to burn the extra calories off.

Naturally, I'm shocked - reality often rounds itself to politically-expedient numbers, after all. That's why we can be sure that CO2 atmospheric concentration targets are Proper Science - they're exact multiples of 50ppm. Next thing you know someone will be claiming that alcohol consumption recommendations shouldn't be an exact number of units (defined as 10ml of pure ethanol, based on the content of a one-ounce shot of 40% spirits, just so it's nice and exact), and the world will go to rack and ruin!

The article mentions a fear that the the Department of Health will "sweep this report under the carpet" - and they should rightly be fearful. As we've learned from Alan Johnson sacking Professor Nutt, we can't have science causing "confusion" with official policy. Such tension has to be resolved, and quickly lest it worry the populace - but, as we all know, the state know's what's best for us, so policy can't be wrong or changed. It must be the science to blame, eh?

I don't know about you, but I'm about to go and eat my free cheeseburger!

- KoW

Labels: , , ,

Mission: Impossible

The BBC reports that the IMechE has said that the 80% let's-go-back-to-the-middle-ages CO2 cuts by 2050 that the government has signed into law are unachievable. No, really. Reducing CO2 emissions to barely more than the population produces by respiration is a bit tricky. Shocking.

The Department for Energy and Climate Change - yeah, they're going to be neutral with a name like that - have said that the IMechE should be thinking positively. Specifically, they accused the Institution of having a "can't do, won't do attitude". Because that's all it takes to achieve massive geo-engineering projects: the right attitude. And this government wonders why it's consistently failed to achieve anything beyond passing new (and mostly unenforced) laws and pissing people off...

I think the IMechE are wrong about the environmental side of this, but I'm 100% behind their sound engineering approach. You can't make offshore wind turbines or nuclear reactors appear by fiat: you have to build them, which means moving materials and equipment on a massive scale. That takes time and money and people, and there are finite limits to all of them.

More worrying is the quoted prat from the Tyndall Centre saying we need carbon rationing. That really is one step too far: the ability to eat, travel, stay warm and travel restricted by the government and sacrificied to "the greater good". Never.

- KoW

Labels: , , ,

Friday, November 13, 2009

Zero Tolerance = Zero Intelligence

Apparently handing a gun in to the police is a crime - with a minimum penalty of five years and no possible legal defence. Zero Tolerance policing at work, folks! Tough on crime, tough on the causes of crime, tough on people trying to be good citizens! Fiat justitia ruat caelum!

(Hat tip to Old Holborn)

Labels: , ,

Bonuses - in MoDeration

So, the MOD has paid £47m in bonuses to civilian staff in the first six months of this year, and nearly £300m since 2003. That's rather a lot compared to the £20m cut in TA funding and it's shameful that they considered their own wellbeing ahead of the troops. What's interesting, though, is that this story is being spun in two different directions.

The BBC story above takes the tack that, as it's split 50,000 ways (out of 85,000 staff), the average payment is less than £1000 - though other reports note that some senior officials are getting £8k and some got bonuses of £17k in 2007-8.

However, Alan Johnson - yes, the home secretary is taking point on a defence story - is pushing the angle that the civil servants are being paid "danger money" and in lieu of overtime for their work in the field. Afghanistan is mentioned and the implication is that the bonuses are going to those people.

Clearly both cannot be true: the number of unique Brits in Afghanistan in 2009 is unlikely to get anywhere near 50k, even including the troops. Main Building and Abbeywood haven't suddenly decamped to there.

The truth of the matter, as I suspect any intelligent reader has figured out for themselves, is that the MOD has decided to make moderate discretionary payments to a bit over half of its staff, and a few larger ones for extraordinary circumstances. Rationally, that's not a big deal - it's a tiny proportion of the wages budget (85k people x £30k salary = £2.55bn), so why the need to lie and spin?

Surely the bigger scandal is why we're paying billions in wages to office workers who have repeatedly failed to deliver.

- KoW

Labels: , , ,

Monday, November 09, 2009

Nonce Sense

NSPCC/ChildLine have released a study today, which made the front page of Metro and the BBC also covers it.

It turns out that there's been a 132% rise in the number of women being implicated in sex-abuse cases since 2004-5. That's hardly surprising, given the systematic bias against men (whom we shall refer to as "potential paedophiles") working with children. It's also symptomatic of an under-reporting of woman-on-boy assault - like most teenage boys, I certainly wouldn't have complained to the authorities if a female teacher got a bit frisky (chance would be a fine thing!)

The statistics bear out what we already knew: the vast majority of child abuse is from natural parents and close family. Most of the rest is from step-family, then professional carers. The "paedo under the bed" has - to all intents and purposes - never existed, despite the hysteria about strangers assaulting children.

The number of calls - 16,094 - seems quite high. No doubt there are a fair number of malicious lies there, and a huge number of multiple-callers but, even if not, that represents around one child in 1000 being assaulted. One in a thousand. We're preparing to (ineffectively) vet over 12m people... TWELVE MILLION... for something that affects 0.1% of the population - and that vetting isn't even required of parents or step-parents, despite being 45% of the male and 67% of the female accused.

Sorry, Alan Johnson, looks like the Home Office is yet again failing at Evidence-Based Policymaking.

- KoW

Labels: , , ,

Sunday, November 08, 2009

ASLEF is an Anagram for "Total and Complete Bastard"

It's obviously not a strike, drivers just haven't turned up for work - in a coordinated action - today. That makes me feel warm and fuzzy to be living on the Hitchin-Cambridge spur of the Great Northern line, so I'm completely unable to travel today - the nearest replacement bus services are about 10 miles away, if I wanted to take a bus to Cambridge and then the Liverpool Street train.

That makes me glad I bought a season ticket.

I don't know if First Capital Connect will call a Void Day, presumably they will as they have no way of running a service today, which means I might get a couple of quid compensation. Woo.

It wouldn't be so bad if the "grievance" wasn't utterly contrived: drivers are objecting to being paid overtime to come in on a Sunday as it's only "voluntary" work. So they've all decided en-masse not to volunteer so, even though there are enough who would want double-time or whatever the pay is on any given Sunday, there is no service to protest this lack of compulsion. We know full well that the union would reject any calls to make Sunday working mandatory, and that would make the system less flexible and worker-friendly than it is currently.

So, what this actually is, is a strike about the lack of grounds for calling a strike.

Thanks, ASLEF. Strikes are only viable if the employer is a non-monopoly: it will lose customers to its competitors, so has an incentive to end the strike. With a monopoly, or a state-run pseudocompany, all it does is victimise customers. This union, like so many others, is playing a zero-sum game against the general public.

This is, it has to be said, a failure of privatisation: awarding of monopoly franchises does not create a market. The airline model is much better - arrival/departure slots bought from the stations, central (state-run) traffic control, and multiple service providers on the same route. This is so tantalisingly close on the King's Cross lines - GNER/NXEC, WAGN/FCC, Hull Trains and now Grand Central all run on the same tracks, and there are multiple providers as far north as Stevenage and Hitchin. Why isn't it a viable model more generally? Why isn't there a choice between easyTrain where I'd be standing for the entire journey, but it'd be cheap, or a First Class-only OpCo with complementary champagne?

- KoW

Labels: , , ,

Thursday, November 05, 2009

Quantitative Easing

So, the Bank of England is to print another £25bn to pump into the economy, but it will be spent over three months (half the previous rate). By my calculations, that means a shortage of about £8bn/month between QE and the government's borrowing - previously they were about the same. Are the global money markets able to pick up the slack

Now, obviously, the BoE cash isn't going directly to public-sector borrowing - that would violate EU law - but it is going indirectly, giving the banks a profit margin from buying new-issued Gilts and then selling them on to the Bank. The net flow is the same, though - newly-printed money is supporting the government deficit, and there'll be less of it in the near future. The further dilution of Sterling is unlikely to appeal to potential investors, though the "small" delta here probably won't make much difference in the short term; likewise, it's probably too small to have a significant effect on exports.

Which all adds up to bad news for the hopes of getting out of the recession this quarter - if the government can't push enough money through the economy, it can't raise GDP, which means no "technical recovery"...

- KoW

Labels: , ,

Wednesday, November 04, 2009

David Cameron on Europe

I thought it was a good speech this afternoon - he made good, pragmatic points in a respectful but forceful way. It really is too late to do anything about Lisbon, and it'll be in force in 26 days' time whether we like it or not. It won't have satisfied the "wets" or the UKIP nutters, but it's the only practical response.

And, yes, it'll be a very hard fight to get those three vetos - we'll need support of our partners in Europe, and undoubtedly some squaring will take place. As is traditional, you'll be able to recognise it because there will be absolutely no connection whatsoever between us supporting their pet policies and them supporting ours.

I started wondering about the Sovereignty Act, though. It seems like a very good idea - basic, straightforward, and in lieu of a constitution. We don't need a constitution, because Common Law is based on the principle that everything is legal unless specifically banned - the opposite of the Napoleonic Code where rights only exist because they're specifically granted.

The most obvious objection is based on the Factortame case, which "confirmed the supremacy of European Union law over national law in the areas where the EU has competence" as granted by the European Communities Act 1972. The ECJ said in 1990 that courts can strike down national laws; Lord Denning suggested that this applies only to accidental contradictions, and that Parliament is still sovereign. There are legal arguments to be made, of course, but there is one fundamental principle that makes them irrelevant: there is nothing the EU can do about it. If we declare our sovereignty, and choose not to follow EU Directives we don't like, what are the ramifications? To quote Sir Humphrey from Yes Minister:
"Well, Minister, in practical terms we have the usual six options: One, do nothing. Two, issue a statement deploring the speech. Three, lodge an official protest. Four, cut off aid. Five, break off diplomatic relations. And six, declare war."
Speeches and protests are, as ever, worthless. We're a net contributor to the EU, so they can't cut off aid. There can be no formal diplomatic relations with the EU before December, so we're hardly going to miss those, and trade sanctions cut both ways - we're a market as well as a supplier. Finally, even with the Lisbon Treaty, the EU doesn't have the power to set military policy or declare war. Basically, if (as a nation) we choose not to follow a Directive - assuming we don't then choose to allow ourselves to be punished for the transgression - there is nothing that can be done about it. Naturally, things wouldn't get that far - as in Italy and Spain and France, we'd just be allowed to get away with it.

I think there are a lot of benefits to close ties with Europe - (almost) free travel and working is good, and integrated continent-wide policies on agriculture/energy/education/healthcare/transport could bring obvious benefits. But that's not the same as having to obey every single law that comes out of Brussels, particularly ones that damage our society or economy. I look forward to the next government rolling back some of the more egregious European laws in the UK.

- KoW

Labels: , , ,


So, lots of things happened yesterday, and I don't have time to post on all of them individually.

Václav Klaus, the Czech President, became the 27th and final leader to ratify the Lisbon Treaty. It will now come into effect from December 1st, creating official President and Foreign Secretary roles and changing virtually all EU-wide lawmaking from unanimous to majority decisions. This, in the language of the EU, is "simplification" rather than a ratcheting up of power. I'm anti-Europe, but I don't think this change is as catastrophic as others have made out - in particular, we now only need a majority vote to begin repatriating our sovereignty, and centre-right politics are in ascendance across the continent.

Since this is now unavoidable - unless some kind soul were to cause a general election in the next four weeks - I support David Cameron's decision to drop the referendum calls. As William Hague says, it's pointless - it can't unratify the treaty and it's too late to stop it. The idea of putting it in as a manifesto pledge is much better. Unlike a referendum which might go either way (see Ireland), a Tory victory will guarantee a mandate for action - and, if the polls are to be believed, such a victory is on the cards. I guess we'll see what Cameron has to say today.

There's been another huge bail-out of the banks by the UK government - £40bn more thrown at RBS and Lloyds. Alex was in quickly with the news, and reports on today's FT coverage. "Worlds biggest bank bailout" as Metro's front page shouted. City AM leads with the story as well, but focussing on Neelie Kroes' influence.

GM has pulled out of the Magna/Opel deal, which is probably bad news for the Vauxhall workers - GM is now going cap-in-hand to the UK and German governments. I can't see Lord Mandelson changing his stance, especially as finances are getting ever tighter, and the German promises were for a different deal and were made prior to an election that's now happened - very easy to change.

M&S are now selling other brands - I thought this was nationwide when they started the trial in April, but apparently it is now. Good. I don't buy shoddy own-brand colas. This now makes their lunch deal much more attractive as I can get a bottle of Diet Coke with the much-better-than-Boots' sandwiches.

Toyota has pulled out of F1, with immediate effect. Poor Kamui Kobayashi... I hope he can get another drive after his fantastic performance in the last two races. If not, we should support him by going and eating sushi at his family restaurant! Bridgestone are going, too, which is a shame. I was hoping there'd be tyre choice back on the strategy menu now that fuelling is off.

- KoW

Labels: , , , , , , ,

Tuesday, November 03, 2009

Government IT Failure

Yet another government database project ends in disaster. £161m spent on the project cannot be accounted for. A spokesman claims that "Steps have been taken to ensure that the mistakes made are not repeated.", but isn't that the same line trotted out every single time one of these projects goes tits-up?

What "steps" have been taken? What "lessons" have been learned? And why, if those aren't blatant lies, do things keep going wrong year in, year out? Why is a government so obsessed with databases so inept at their implementation?

The Technology in Business programme and the Government IT Profession were set up several years ago to address these issues, yet what have they achieved? The policy statements for the GITs include this gem:
Enabling organisations and individuals to develop the capability required to deliver excellence through advice and guidance on embedding professionalism and using the skills frameworks, and by creating and signposting learning and development opportunities.
That is gibberish. Complete and utter tosh. Semantically empty. They've produced at least four versions of the skills framework, all talking in generalities and buzzwords. Sound and fury. No doubt the authors of that claptrap would claim I'm "not thinking abstractly" and that classification, categorisation and meta-analysis are vitally important parts of the work. Bullshit.

Databases, despite the impression all this might give, are not all that difficult to engineer. There are around 85k prisoners in the UK. I have a database with more than 170k rows across a dozen tables, used for the back-end to some web services which I wrote in a weekend. That's running, acceptably fast, on a shared server costing me a few quid a month. I'm under no illusions and know that a proper setup - redundant infrastructure, dedicated and distributed hardware, security accreditation, crypto, development and testing of the applications - would cost considerably more, but a million times more? Several thousand pounds for each prisoner in the database?

I know I'm not cut out for the civil service: I actually value results - rather than process and paying lip-service to results. IT projects aren't made by consultants and reports and policies and skills frameworks; they're made by analysts and programmers and admins and engineers. Yes, these people need some direction, but direction in and of itself cannot produce the desired system - you cannot make a database by executive fiat!

- KoW

Labels: , , ,