Seeing Further (49 page)

John Houghton, former director of the UK Meteorological Office and the IPCC Working Group I leader for the first three assessment reports, was initially very reluctant to get into the surprises tangle. I recall a very clear exchange at a climate meeting in Oxford University in 1993.
⁸
Houghton thought the public discussion about ‘surprises’ was too speculative
and would be abused by the media. ‘Aren’t you just a little bit worried that some will take this surprises/abrupt change issue and take it too far?’ he asked. ‘I am, John; we have to frame it very carefully,’ I replied. ‘But I am at least equally worried that if we don’t tell the political world the full range of what might happen that could materially affect them, we have not done our jobs fully and are substituting our values on how to take risks for those of society – the right level to decide such questions.’
⁹

In the end, despite the worry that discussions of surprises and nonlinearities could be taken out of context by extreme elements in the press and NGOs, we were able to include a small section on the need for both more formal and subjective treatments of uncertainties and outright surprises in the IPCC Second Assessment Report (SAR) in 1995.
¹⁰
Chapter 11
, ‘Advancing Our Understanding’, was about what to do later, and so was not directly assessed in the more politically sensitive conclusions of the report. Thus, John did not object to the few sentences on those topics in that chapter. As a result, the very last sentence of the IPCC Working Group I 1995 Summary for Policy Makers (SPM)
¹¹
addresses the abrupt non-linearity issue. This made much more in-depth assessment in subsequent IPCC reports possible, simply by noting that ‘When rapidly forced, non-linear systems are especially subject to unexpected behaviour.’

Now we had licence to pursue risk assessment of uncertain probability but high consequence possibilities in more depth; but how should we go about it? The basics are that scientists can help policy-makers by laying out the elements of risk, classically defined as
consequence
x
probability.
In other
words, what can happen and what are the odds of it happening?

The plethora of uncertainties inherent in climate change projections clearly makes risk assessment difficult. The inertia in the climate and socio-economic systems and the fact that greenhouse gases emissions will continue to rise, given the absence of strong mitigation policies (or unexpected events like a prolonged recession), indicate that globally most policy-makers have been reluctant to make long-term investments beyond their expected terms in office. But that is changing both in some regions like the EU and even in the US. These kinds of decision-makers are increasingly wary of making what is known as a Type II error – fiddling while the Earth burns. A Type I error is a false positive, which in this case would mean taking action against climate change which subsequently proved relatively needless. Scientists are often leery of making a Type I error when data are scarce for fear of misleading society into unnecessary actions and being blamed for undue alarm. The other kind, a Type II error, is a false negative, and in this case would mean assuming it is preferable to do little or nothing until there is less uncertainty, and subsequently finding that serious climate change ensues unabated with much more damage than if precautionary policies had been undertaken to adapt to and mitigate the effects. So it appears that many scientists are often Type I and our future-oriented decision-makers Type II error avoiders. A less charitable interpretation of those reluctant to invest in precautionary adaptation and mitigation measures is that they know that the really adverse outcomes will likely occur in the future when current decision-makers are not in office and not likely to be held accountable. The short-term incentives are to delay action and pass the risks and the recriminations on to the next generation. None of this is scientific risk assessment, but value judgments on where and how to take risks and make investments in policy hedges – in short, risk management. But risk management is put on a much firmer scientific basis when the managers are schooled in the best risk assessments that state-of-the-art science can produce.

To help decision-makers, the IPCC produced a Guidance Paper on
Uncertainties in 2000
¹²
which was a foundation for the 2007 Fourth Assessment Report.
¹³
I prepared the original draft with Richard Moss, now a Senior Scientist, Joint Global Change Research Institute, after convening a meeting in 1996 in which about two dozen IPCC lead authors met with decision analysts to fashion a better way to treat uncertainties in scientific assessments. The final guidance eventually agreed to within the IPCC was a quantitative scale. We would define ‘low confidence’ as a less than one-in-three chance; ‘medium confidence’, one-in-three to two-in-three; ‘high confidence’, above two-thirds; ‘very high confidence’, above 95 per cent; and ‘very low confidence’, below 5 per cent.

It took a long time to negotiate those numbers and those words in the Third Assessment Report cycle. There were some people who still felt that they could not apply a quantitative scale to issues that were too speculative or ‘too subjective’ for real scientists to indulge in ‘speculating on probabilities not directly measured’. One critic said, ‘Assigning confidence by group discussions, even if informed by the available evidence, was like doing seat-of-the-pants statistics over a good beer.’ He never answered my response: ‘Would you and your colleagues think you’d do that subjective estimation less credibly than your Minister of the Treasury or the President of the US Chamber of Commerce?’

So we had two things we wanted everyone to use – a set of numbers defining the probability ranges for words such as ‘likely’, and a set of qualitative phrases for our confidence in the results, going from ‘well established’ if there were a lot of data and a lot of agreement between theory and data, to ‘speculative’ without much data and when there wasn’t much agreement. We had ‘established but incomplete’ and ‘competing explanations’ for the intermediate cases.

And then for the next two years Richard and I became what a journalist later called ‘the uncertainty cops’. I read three thousand pages of draft material for the IPCC’s Third Assessment Report. People did not always

Assessment and Reporting in R. Pachauri, T. Taniguchi and K. Tanaka (eds), Guidance Papers on the Cross Cutting Issues of use uncertainty terms according to our simple rules. For instance, they would say that because of uncertainties, we can’t be ‘definitive’. I wrote back, ‘What is the probability of a “definitive”?’ Early drafts would put the range of outcomes anywhere from a one to five degrees Celsius change in temperature. And then they would say in parentheses ‘medium confidence’. That was completely incorrect. It was ‘very high confidence’, because they were talking about the fact that
between
one and five degrees was a very, very likely place to arrive. But people didn’t want to say ‘very high confidence’ because nobody felt very confident about the state of the science at the level of pinning it down to, say, one degree. So Richard or I would help them to rewrite, and say that we have ‘low confidence’ in specific forecasts to a precision of a half degree, but we have ‘high confidence’ that the range is one to five degrees. Simple things like that were needed to achieve consistency of message.

Meanwhile the political chicanery of ideologists and special interests was shamelessly exploiting systems uncertainty by misframing the climate debate as bipolar – ‘the end of the world’ versus ‘it’s good for you’. The media compliantly carried it in that frame much of the time, too. But those were and still are, in my view, the two lowest probability outcomes. The confusion that bipolar framing has engendered creates in the public at large a sense that ‘if the experts don’t know the answers, how can I, a mere lay citizen, fathom this complex situation?’ To this, industry-funded pressure groups added the old trick of recruiting non climate scientists who are sceptical of anthropogenic climate change to serve as counterweights to mainstream climate scientists. This spreads doubt and confusion among those who don’t look up the credentials of the apparently contending scientists – and that, unfortunately, includes most of the public and too much of the media. The framing of the climate problem as ‘unproved’, ‘lacking a consensus’, and ‘too uncertain for preventive policy’ has been advanced strategically by the defenders of the status quo. This is very similar to the tactics of the Tobacco Institute and its three-decade record of distortion that helped stall policy actions against the tobacco industry, despite the horrendous health consequences and eventually billions of dollars in successful lawsuits against big tobacco.

In the face of such tactics, the IPCC assessment reports are intended to be the best achievable statement of current scientific consensus. But ‘consensus’ is not necessarily built over conclusions but the
confidence
we have in a host of possible conclusions. With that kind of information policymakers can make risk-management decisions by weighing both the possible outcomes and the assessed levels of confidence – we know it well, sort of know it, or hardly know it at all. Scientists should just say what we do
know and don’t, and not leave something out because it isn’t a well-established consensus yet. It is the job of society, through its officials, to make the risk-management decisions informed by our conclusions and accompanying confidence estimates.

Again, the groups preparing IPCC reports had many hot, contentious discussions on that issue. Working Group I, for example, initially balked at the notion of including subjective estimations, and then embraced it, but then said that they needed to have finer gradations, because they had real data, not just subjective judgments, and they wanted to have a 99 per cent and a 1 per cent. There were also interesting disciplinary differences. Linda Mearns at the National Center for Atmospheric Research, one of the few lead authors in two working groups, helped reconcile the physical scientists in Working Group I who were leery of subjectivity and risk management and the ecologists and social scientists in Working Group II who felt that society, not scientists, should choose how to take risks after
all
the possible conclusions were reported. It took us quite a long time to get both sides to first understand and eventually respect the other point of view. My role was not to endorse one or the other, but rather to be sure all our reporting was explicit about assumptions, so we could have a ‘traceable account’ of all underlying processes behind important conclusions. That process is building, but is not yet complete across the IPCC or the scientific community in general.
¹⁴

As I’ve said, normally science strives to reduce uncertainty through data collection, research, modelling, simulation, and so forth. The objective is to overcome the uncertainty completely – to make known the unknown. Short of that, new information may narrow the range of uncertainty. No doubt further scientific research into the interacting processes that make up the climate system can reduce uncertainty about the response to increasing concentrations of greenhouse gases. This is very unlikely to happen quickly, however, given the complexity of the global climate and the many years of high quality data which will be needed. Meanwhile, even the most optimistic ‘business-as-usual’ emissions pathway is projected to result in dramatic, dangerous climate impacts. That means making policy decisions before this uncertainty is resolved, rather than using it to justify delaying action.

Risk management also means understanding what is truly uncertain, and what is not. Sometimes critics claim that there should be no strong climate policy until the science is ‘settled’ and major uncertainties resolved, whereas supporters of strong policies suggest the science is already ‘settled enough’ and it is time to proceed with action to reduce risks. The science which demonstrates a significant warming trend over the past century is settled; moreover, it is virtually settled that the past several decades of warming have been largely caused by human activity and that much more is being built into the emissions pathways of the twenty-first century. Sounds like the ‘settled already’ side has won the debate: warming is occurring and human activities are the primary driver of recent changes.

That leaves the uncertainty about how severe warming and its impacts will be in the future, especially when projections for ‘likely’ warming by 2100 vary by a factor of six. The task then is to manage the uncertainty
rather than master it, to integrate uncertainty into climate research and policy-making. This kind of risk-management framework is often practised in defence, health, business and environmental decision-making. But the thresholds for action often seem lower. The US has a military arm, of course, and although I may not like everything we do with it, I don’t know anybody who says you should get rid of it because a nation has to have security precautions, even against only very low probability – but potentially dangerous – threats. Well, the climate change threat is not 1 per cent. It’s more than 50 per cent for many really significant troubles, and maybe 10 per cent for absolutely catastrophic troubles.

In my personal value frame, it is already a few decades too late for having implemented some policy measures against such risks. Had we begun mitigation and adaptation investments decades ago, when a number of us advocated them,
¹⁵
the job of remaining safely below dangerous thresholds would be easier and cheaper. Similarly, beyond a few degrees Celsius of warming – at least an even bet if we remain anywhere near our current course – it is likely that many ‘dangerous’ thresholds will be exceeded. Strong action is long overdue, even if there is a small chance that by luck climate sensitivity will be at the lower end of the uncertainty range and, at the same time, some fortunate, soon-to-be-discovered low-cost, low carbon-emitting energy systems will materialise. For me, that is a high-stakes gamble not remotely worth taking with our planetary life-support system. Despite the large uncertainties in many parts of the climate science and policy assessments to date, uncertainty is no longer a responsible justification for delay.