Horizon - Science under Attack

I finally got around to watching the Horizon programme last night; it's been a good series this year. As can be deduced from the title the topic was how science was being treated; how those who work in certain areas seem to be attacked by the media and others.

In this instance I'm going to start with the conclusion our host, Sir Paul Nurse, came to. Scientists need to do more to display their data, communicate the uncertainties, and better explain how they came to develop their theories. I agree with this in its entirety.

There we go end of entry... well no I started with the conclusion because I found it amusing that the first half exhibited some of the exact problems that cause science to be under attack.


Science works with data and observation to produce a chain of logic that results in theories. With that in mind I'll relate the following events from the programme.

A hypothesis is presented in the programme that climate change is caused by solar activity not man-made emissions. At NASA we see a projected sphere of the Earth* and the scientist states that the evidence shows that such has minimal impact on temperatures; on the other hand we're pumping out 7 Gigatonnes of Carbon a year compared to a massive volcanic eruption that would only produce 1. Therefore it's man-caused change.

But we have a break in the chain of logic there. As presented by this show we go from "it's not solar" to "therefore it's the carbon we produce" with no evidence presented that carbon has any affect whatsoever on the climate. Please note I'm not saying it doesn't (or that it does) merely that we have a logic break here that the scientist presenting does not fill-in at any point.

In another instance Sir Paul tries to explain the complexity of cause and effect and I'm afraid does so in a way that leaves things slightly muddled. I'll unravel it -

We have Cause A that results in symptom/effect B a little later symptom/effect C shows up. If you are unaware of Cause A then the logically conclusion is that B is the cause of C. To put it another way you contract influenza (A) this results in a blocked-up nose (B) and you start to increase in temperature (C). If you were unaware of influenza you could conclude that the blocked-up nose causes a fever.

What's important about my example is that you can look at other evidence. It's possible to have a blocked-up nose and not have a fever; it's possible to have a fever and not a blocked-up nose. Using such observation and data you could come to the conclusion that although they can are related they're not causal; in which case you start to look for a third common factor.

Again as presented this is the flaw we see in the carbon emissions. We see an increase in temperature and we see an increase in carbon emissions therefore they are connected. Again I'm not saying that's a wrong conclusion to draw; merely that it's not explained here.

Next we have Sir Paul wondering how people don't want to accept the consensus or the data. He makes a good point that if a consensus of doctors say you need to do this you're more likely to accept that then one loan doctor - how is climate data any different? Sadly undermined by his discussion with an HIV positive individual who was told he'd be dead in a year if he didn't take these (consensus) medicines. He didn't and that was 13 years ago. Now of course doctors should be all over him to found out why and what makes him different; and most importantly if his thoughts hold true. Yet we see nothing of that.

Consensus is fine, however when dealing with peer-review a flaw can appear that exists within any profession that deals with specialised knowledge. Once you get to a certain level less people are 'qualified' to judge your work. When you have a consensus within that group then you can have a paper in which the results contained already agree with the results that the peer-reviewer already holds. Now this bias is known and steps are taken to guard against it; yet still when a consensus forms it may be said that any paper that agrees with it is likely to be treated less rigorously than one that doesn't, which in turn emphasises the consensus viewpoint when it gets published.

In another flaw we're shown a screen in which the top half is real data and the bottom is the modelling software and shown how the two match up. Except we're not. We're shown very briefly the two screens with the two men standing in front blocking the view. We're shown close-ups of one of the screens while at the same time being told "You can see how closely they match" well no we can't because you're not showing it to us.

Next the ClimateGate emails were examined with the "hide the decline" and the "Nature trick" what was going on? It was explained that temperature readings were based on tree rings and that from the period that independent temperatures were taken up to about 1960 there was a high correlation between the two sets of data.

To clarify what that means in simplistic terms is that if you measure a tree ring at 2mm and a temperature reading at the time shows 2 degrees; then another tree ring of 2mm would correlate to a temperature reading of 2 degrees. Using this you can build up data so that you know that a tree ring of X means a temperature of Y.

So why the dodgy language. Well as I've said this works up until around 1960 and then the correlation diverges. When you plug in the tree data past 1960 and derive a temperature using the pre-1960 equation it doesn't match the recorded temperature. As a result the data up until this point is tree ring and after actual recorded temperatures.

On the face of it that seems fine - if you know one set of data is a bit screwy and one set isn't you simply splice the one onto the end of the other. The fault lies in that this wasn't explained and Sir Paul makes that point in that it should have been. Good; except another fault exists that wasn't explained. The reason for the divergence isn't known. Now if the data during the period in which it can be tested against measurement works only up to 1960 and you don't know why it changed after that; with what confidence can you use readings derived before it can be tested? How do we know that another divergence didn't occur a few years prior to testable measurements maybe this period it which everything aligns is an aberration rather than a norm?

And finally although aiming for why science as a whole is under attack the majority of this programme dealt with Climate Change; and Sir Paul's qualifications... he's a biologist.Nothing wrong with that, but it's difficult to state that we should listen to the people who know the science they're talking about when it's presented to us by someone who doesn't deal with that field.

I therefore found myself agreeing with the conclusion while being amused at a programme that exhibited most of the flaws complained about.

*I wonder how long NASA have had that sphere? I had a similar idea as part of a thought experiment - "How I'd have presented the Millennium Dome" back in 2000 in which a giant projected real-time Earth sphere would make the central attraction with galleries around it so you could view it top and bottom and all around. I thought it would be pretty awesome.

[Update - prototype in 1995, patent in 2005; go figure]