For reasons that need not concern you, I found myself the other day reflecting upon the claimed 97% climate change consensus. Ninety-seven has always struck me as a strange yet ubiquitous figure that seems to crop up in surveys with a frequency that hints at non-randomness. Why does it never seem to be ninety-six or ninety-eight percent? The answer I have given myself is that these two even numbers are not spiky and idiosyncratic enough to carry the required message of scientific credibility. Furthermore, ninety-six is just a tad too low to impress, and ninety-eight is just a tad too high to be believed. Only ninety-seven hits that sweet spot that Goldilocks cannot resist.

To confirm my suspicions that ninety-seven occupies a favoured position in the survey results popularity charts, I decided to google it, just to see the variety of claims that fall into the same fake credibility niche occupied by the climate consensus. I grant you that a mental space in which such an exercise might seem a good idea is a strange one to occupy, but we deniers are queer folk, as Lewandowsky et al have been keen to point out. Unfortunately, my research proved futile since, rather than finding a myriad of claims based upon the mythic ninety-seven percent, I found just the one, repeated endlessly:

97% of women say they have been victims of sexual assault.”

Yes, it seems that the ninety-seven percent meme has now been co-opted to further the idea that women are no longer safe in society. If the number ninety-seven finds employment in the service of any other agenda, then I’m afraid the evidence of it must be buried on page ninety-seven of my google search results (not ninety-six or ninety-eight, mind you). Fortunately, however, this single example currently choking the meme pool does at least perfectly illustrate the underlying problem with all such claims:

It simply makes no sense to make a precise claim regarding a vague notion.

In the case of the sexual assault statistic, the vagueness stems from the lack of a reliable definition of what constitutes sexual assault – at least in the minds of those answering the surveys. At one extreme we are dealing with crimes of the most heinous nature, but at the other extreme there are events that may simply be the unwanted attentions that are inevitable given the nature of the mating game. When does an approach become an assault? It can’t surely be left at the whim of the victim to declare. But if it is, we have a problem because the concept will have become hopelessly subjective.

In fact, the concept of ‘assault’ is a classic example of the sort of vague noun beloved of linguists and philosophers of vagueness. There has to be borderline cases that suffer from arbitrariness and the sorites paradox (e.g. if an unsolicited smile is not an assault, then a smile that is barely distinguishable from a leer must be equally innocent, and so on). As a consequence of this vagueness, any claim to a precise quantification has to be bogus. Similar examples can be found in medical science where statistics relating to vague notions, such as obesity, are quoted with far too much confidence and precision.

The same problem, of course, exists with the precise claims relating to climate change consensus. When quantifying the consensus, the vague definition of the belief under investigation gives enough latitude for the researchers to gravitate towards any precise quantification that floats their boat – but it still makes no sense to claim a precise value when the position of belief under study is so loosely defined. The ninety-seven figure may have been chosen for its credibility, but the truth is that its precision is the very thing that signposts its untrustworthiness. Even more damning is its apparent constancy in the face of scientific progress. The consensus should be increasing, but the declared value is not; the enchantment of the number ninety-seven is just too strong. It is not a constant of nature, but when there is an idea to be nurtured, it becomes a constant of nurture.

Another classic example of a constant of nurture within the field of climate science is the stated range of plausible values for the Equilibrium Climate Sensitivity (ECS). This was first guessed at back in the sloppy seventies and that guess has retained a constant presence ever since. As knowledge is gathered one can either use it to make an estimate more precise or one can use it to increase confidence in the currently stated level of precision. Strangely, over forty years, it is only the latter that seems to have happened with regard to the ECS value – we started out thinking we don’t know the value precisely, but now we know for certain that we don’t know it precisely. We seem to have done little more with our newfound insights than to bolster our confidence in that regard. This strange phenomenon only starts to make sense when one stops thinking of ECS as a constant of nature and starts thinking of ECS as a constant of nurture. Back in 1997, Jeroen P. Van der Sluijs, Professor, Theory of Science & Ethics of the Natural Sciences, Bergen University, explained this constancy by stating:

“[T]he stability can also be seen as a function of an implicit social contract amongst the various scientists and policy specialists involved, which allows ’the same’ concept to accommodate tacitly different local meaningsThe maintained consensus about the quantitative estimate concerning a central scientific concept in the anthropogenic climate change field, namely climate sensitivity, operates as an anchoring device in science for policy.”

So the main advantage of holding the ECS uncertainty range constant down the years is that it provides a firmly constant framework within which to perceive the level of risk posed by climate change, particularly if that view is informed by the concept of ‘actionable uncertainty’, as proposed by Lewandowsky (That’s the second time I have mentioned his name; as with Beetlejuice, I will have to be very careful not to mention it a third time).

As such, the accepted range of values for ECS is a ‘constant’ that guides the science just as readily as Newton’s constant of gravity (G). The only difference is that G is a constant of nature and the ECS range is a constant of nurture. A narrower range for the latter would be more valuable to the science but less valuable to the politics, for which too many social contracts have already been signed.

For different reasons, there is an arbitrariness to both the ninety-seven percent consensus figure and the ECS range. In both cases, we are invited to think that the values are purely the outcome of an unbiased analysis of the available data. In practice, however, there is no such thing as unbiased analysis, and the degree to which bias may have been applied can be seen as a reflection of the degree to which a result serves to nurture favoured ideas. It is under those circumstances that a constant of nature can morph into a constant of nurture.

I’ll leave the last word to Stephan Lewandowsky.

Oh my God! What have I just done?


  1. John, I hate to do this to you, but you obviously don’t read the Guardian with sufficient intensity: 97% is SO last year, or even last decade.

    “The fight against climate change is down to us – the 99%
    Naomi Klein”

    “Is the climate consensus 97%, 99.9%, or is plate tectonics a hoax?
    A new study argues the 97% climate consensus estimate is too low, while deniers claim it’s too high”

    “‘No doubt left’ about scientific consensus on global warming, say experts
    Extensive historical data shows recent extreme warming is unprecedented in past 2,000 years”

    “The scientific consensus that humans are causing global warming is likely to have passed 99%, according to the lead author of the most authoritative study on the subject, and could rise further after separate research that clears up some of the remaining doubts.”

    And elsewhere, there’s this:

    “More than 99.9% of studies agree: Humans caused climate change”

    You may once have been correct to say:

    “Why does it never seem to be ninety-six or ninety-eight percent? The answer I have given myself is that these two even numbers are not spiky and idiosyncratic enough to carry the required message of scientific credibility. Furthermore, ninety-six is just a tad too low to impress, and ninety-eight is just a tad too high to be believed. Only ninety-seven hits that sweet spot that Goldilocks cannot resist.”

    But now, I’m afraid, it seems that anything less than 99.9% is derisory. Just like the registered voter turnout in the North Korean elections in 2019:


  2. Actually, Mark, I was well aware of Klein’s’ attempt to overturn the ninety-seven percent trope. It did not affect my thinking, however, because it isn’t a genuine removal of a constant of nurture but an attempt to replace it with another one, i.e. the Domestos statistic. Time will tell if she ultimately succeeds, but for the moment I still think the 97% meme holds sway. In fact, it is telling that, by challenging the 97 figure, the Guardian only really succeeds in drawing further attention to it.


  3. “Domestos statistic” – very good!

    It’s curious, isn’t it, that just like the North Korean regime, they can’t bring themselves to claim 100%. Apparently nobody would believe that, but 99.9% IS believable? I suppose the 0.1% of permitted doubt gives them a get-out if all fails to pan out as the hysterics insist it will.


  4. “The twin threats of flooding and drought”

    “Lorraine Watson’s chip shop has been badly flooded five times in 10 years.

    The Carron fish bar in Stonehaven takes its name from the river which flows beside – and all too often through – its basement.

    The last time that happened was in the early hours of 12 August 2020 when the spate lifted freezers from the floor and left behind a sticky, stinking soup of mud and grease.

    The bill ran to £8,000, according to Ms Watson, who says the floods have become more intense and more frequent in her decade running the shop.

    On the same summer’s day, a train hit a landslide at Carmont just outside the town, killing the driver, a conductor and a passenger.

    Is climate change a factor in Aberdeenshire’s repeated, devastating floods?

    “One hundred per cent” declares Ms Watson, adding: “We must take action. Everybody must take action.””

    Is that the first sighting of a 100% claim?

    Still, the new approach to climate change is handy – too much rain: climate change. Not enough rain: climate change:

    “Summer rain storms are forecast to become rarer but heavier with a risk of flash floods but at the same time the nation may be grappling with the opposite problem – water scarcity as a result of a 7% drop in overall summer rainfall.”

    Sorry if I’ve wandered O/T.


  5. I should also make it clear, FWIW, that I approve rather than challenge your analysis. 🙂


  6. From Toby Young:

    The Italian Higher Institute of Health had adjusted downwards its estimate of the number of people who’ve died from Covid, as opposed to with Covid, from over 130,000 to under 4,000. Yes, you read that right. Turns out 97.1% of deaths hitherto attributed to Covid were not due directly to Covid.

    It”s the point one that makes all the difference.


  7. Mark,

    I had wondered. But more to the point, I find myself reflecting upon the deafening silence with which the article has been met. Maybe it is because people feel it is stating the obvious. If so, I think it is a self-evidence that still bears pointing out. All of the controversy regarding the climate consensus surveys could be avoided by making the simple observation that the linguistic vagueness undermining the survey renders any precise quantification as meaningless. This violates the ‘maxim of quantity’, as formulated by the linguist Paul Grice. This maxim states that in communication:

    “One should be as informative as one possibly can, and give as much information as is needed, and no more.”

    Applying this maxim, one immediately recognises that the precision of the 97% claim is gratuitous and so cannot be trusted. People can save themselves an awful lot of time and effort if their first instinct when evaluating a statement is to ask themselves whether the maxim of quantity is being violated. If it violates the maxim, don’t bother wasting any more time on it.

    Vagueness is an important element of uncertainty that I think is all too often overlooked.


  8. John,

    Re-reading my comments, I was bothered that I may have left the wrong impression, and so was anxious to put the record straight. My arrows were aimed instead at those who propagate (nurture) claims of 98, 99 and even 99.9% certainty. I made the comparison with claimed turn out levels in North Korean “elections” very deliberately.

    As for the lack of comments here, your guess is as good as mine. If I’m honest, sometimes I’m disappointed at low numbers reading one of my articles. On other occasions I’m very happy with the number of people reading an article, but left bemused by the lack of comments. Maybe we should pat ourselves on the back and say we’ve covered it all in the article, so there’s nothing to add. 😉


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.