[jules' pics] Political Correctness Gone Mad

Many San Franciscans have a Portuguese water dog of just the right age. Bet they're feeling silly now it's not cool to like Obama anymore.
[beach - on the west side of San Francisco]

Sadly, for the rest of the week we must once again sit in windowless rooms and listen to geeks like us present interesting science badly.

--
Posted By Blogger to jules' pics at 12/05/2011 10:39:00 AM

The Oxbridge Myth

Methinks the lady doth protest a little too much...

In our collective cultural imagination, however, Oxbridge means the dreaming spires - a kind of massive Bullingdon Club where rich undergraduates still flaunt their boaters and blazers, and dotty dons get sozzled every night on the college port (while hoping to avoid the Inspector Morse-style murders in the quad). Of course, for most of us, this isn't remotely how it is (I've not tasted port for 30 years...

Not tasted port for 30 years? Well, it can only be because she doesn't actually like the stuff, there is hardly a shortage last time we checked.

As for the tradition of allowing despots to buy places for their offspring, I'm pleased to see that this is still going strong.

Another IPCC whitewash

Typical, isn't it?

They set up a so-called "investigation" and having failed to even interview the relevant people, find "no evidence of a criminal offence". Well, there's a surprise for you. In tomorrow's news, maybe we will hear that turkeys did not vote in support of the celebration of Christmas (shh, but don't tell the ducks, cos I already ordered one of those).

Oh, and the report is being kept secret for now, under the IPCC's longstanding "openness" policy (to be fair, they do say they will publish it in 2012).

Yet more evidence (if any were needed after this) that a thorough reform is needed.

Further details here.

[jules' pics] Health and Safety

Today we had health and safety origami with advanced knotting. Actually it was the most interesting and involving CPR/first aid/safety thingie I have attended. When we arrived in Japan a decade ago, I don't think the phrase health and safety had been invented yet, but there seems to be a bit of pressure towards it now, especially from the boat people of JAMSTEC (boats being such excellent places for losing body parts if you don't know what you are doing).

--
Posted By Blogger to jules' pics at 12/01/2011 08:28:00 PM

[jules' pics] Let's pink

In case anyone was wondering who on earth buys those brightly coloured SLRs... young women are powerful consumers in Japan... 

"The real world is the one on the back of my camera"  

[Hachimangu, Kamakura, Japan]

--
Posted By Blogger to jules' pics at 11/29/2011 03:21:00 PM

[jules' pics] Vitamin C overdose

I suppose it might be the same in all those exotic countries in which citrus grow easily, but I find it odd that Japanese people don't eat the fruits of their own trees. Instead, like in the photo, the fruits remain on the garden trees as decoration all winter long. A British person did once tell me that he walked the lanes and successfully gained access to his neighbours' trees in order to make marmalade. Japanese people do eat a lot of citrus at this time of year. There are many different varieties of delicious satsumary things available, and a family may have a big boxful to work on during the New Year holiday. I'm not sure if what my Japanese friend told me can really be true - that children sometimes turn orange from eating too many!? I also don't know why the Japanese do not grow grapefruit - they are available but imported from the USA. The grapefruity thing in the picture is probably a yuzu, which is generally smaller and sweeter and less juicy than grapefruit. The tree was growing in someone's garden in Kamakura, with the boughs overhanging the road. I could have picked it if I was James' height.

--
Posted By Blogger to jules' pics at 11/28/2011 12:49:00 PM

More on Schmittner

OK, so the Schmittner paper is out, along with a commentary in Science, and I've had a few days to digest it more thoroughly. What I said before about past v future asymmetry still holds true, but there is another point which may be more interesting.

The model results actually don't fit the land data very well, being generally too warm. A key plot is the sensitivity analysis where they compare results when land and ocean data were used separately, versus together. Clearly, the combined analysis looks almost identical to the ocean-only results, and the land-only results are radically different. In fact, they barely overlap with the ocean-only results.

Of course, there is no reason why these results should match exactly, or even closely - remember, they are not estimates of "the pdf of sensitivity" but rather, probabilistic estimates of the sensitivity - but they do need to overlap in order to be taken seriously (if they don't, at least one has to be wrong). The true value has to lie in their intersection, which is rather narrow in probabilistic terms - the 90% range of the land-only pdf is 2.2-4.6C, that of the ocean-only is 1.3-2.7C.

The explanation for this near-disjoint pair of distributions is that the model does not represent the land-ocean temperature contrast well (this is a characteristic behaviour of this sort of model, as the authors acknowledge), so can only fit one set of data at a time. When faced with both, it prefers the ocean, partly because these data are more plentiful, and partly because it is given the prior belief that the land data are less accurate (which they probably are, to be fair). The poor fit to land data then results in the statistical method assigning even less weight to these data through the spatial error term mentioned in the supplementary on-line material, and in the end result they are almost ignored. In the final analysis, the cooling over land (and perhaps also the polar amplification) seems to be significantly underestimated, leading to their rather warm LGM state which is only 3C cooler than the modern (pre-industrial) climate. One might reasonably expect that their future simulations also underestimate the temperature change over land, meaning the sensitivity estimate is on the low side, too.

Jules has also been looking at some of these data recently, particularly in comparison to the PMIP2 experiments - that is, simulations of the last glacial maximum by several state of the art climate models, most of which also mostly contributed to the CMIP3/IPCC AR4 database of modern/future projections. One telling point is that several of the PMIP2 models actually appear to fit the data better than Schmittner's best model, even though these were not specifically tuned to fit the data. Moreoever, these models are all clearly colder, in terms of global mean temperature anomaly, than the -3C value obtained in this latest paper. We haven't done a thorough analysis of this yet but I think it is safe to say that there is a significant bias in the Schmittner fit and that the LGM was really more than 3 degrees colder than the present. The implication of this for climate sensitivity is not immediate (since there are also well-known forcing biases in the PMIP2 simulations), but this line of argument also seems to suggest that it may be reasonable to nudge the Schmittner et al values up a bit.

It is still hard to reconcile a high sensitivity with the LGM results, though.

UPDATE: similar comments from RC.

New leak?

Some people might be surprised to hear me say it, but I think this new leak provides damning evidence of shoddy behaviour. There is clearly inept leadership at the heart of the organisation, plenty of back-biting, and the way in which junior and more conscientious colleagues who refused to toe the party line were bullied and ridiculed is shameful. Many of these people who I had trusted to do their honest best are clearly motivated far more by money than the desire to do their jobs properly. There certainly isn't much evidence of the sort of ethos that we are entitled to expect from people in their position.

I find the whole thing truly shameful, and call upon all those involved to resign. It's time for a new broom.

More details can be found here.

Cancer survival: Macmillan hails major improvement

The first thing I thought of when I saw this story was the talk given by Gerd Gigerenzer last year at this odd but interesting workshop. The gist of it was that "survival time" as a measure of performance in medical science could be very misleading, as it does not necessarily indicate any increase in lifespan or reduction in death rate. Increasingly aggressive and sophisticated screening and diagnosis procedures will automatically result in increased "survival time" even without any improvement in treatment, simply through spotting the cancer earlier in its progress. This isn't a purely theoretical point, he had plenty of statistics to back it up too. That's not to say there haven't been genuine advances too, but 5y survival rate doesn't necessarily measure them correctly.

[jules' pics] Spot the difference

For those who found the last puzzle too difficult...

--
Posted By Blogger to jules' pics at 11/20/2011 09:21:00 PM