Sunday, May 7, 2017

Bret Stephens, Alan Jacobs, and climate change

Alan Jacobs (via
My aggregator list has about 30 different blogs on it.  All of them represent voices that I found interesting and important when I added them; many have now fallen silent, and I regret that, and wonder whether I should remove them or whether they might, perhaps, come back to sparkling life.  One blog that is very alive and  consistently fascinates me is Alan Jacobs' Text Patterns.

Jacobs is currently a distinguished professor at Baylor and before that was at Wheaton College.  He wrote a fine biographical study of C.S.Lewis - one of the best, I think - and more recently has published a history of the Book of Common Prayer - I'd love to read that as the BCP has been a steady guide to me in my Christian journey.

He also writes a wide-ranging blog which right now is revolving around two aspects of our present age which are both loudly announced (by some people) and which seem to be mutually contradictory: on the one hand, that this age is the dawn of the Anthropocene, the age when the human race is getting "big" enough to become the central influence on our planet's ecology; and on the other hand that it is also the dawn of the posthuman, the era when human beings are transcended and (according to some) superseded by machines that are faster, stronger, more agile, precise and intelligent that we are.   "Ours; not ours", writes Jacobs. "It is in the light of this twofold reality that theology in our time should be done."

That is all by way of an introduction to Alan Jacobs for those of my readers who don't know him already.  But last week he posted a piece called "Help Wanted".  It is short enough that I can reproduce it in full.

Suppose you’re a person who believes that anthropogenic climate change is very real and very, very bad news. Suppose further that you believe that portrayals of a future of chaotic weather and massively destructive rises in sea level — e.g., the portrayals we see in Kim Stanley Robinson’s recent novels — are not manifestations of apocalyptic alarmism but are sober, well-thought-out, plausible projections from the best current data. And suppose further that you read that Bret Stephens op-ed and think that it’s not only reasonable but self-evidently correct.

Where would such a person go to be taught, in calm, clear, and rational prose, why that last supposition is in conflict with the previous ones?
Well, I had read the New York Times op-ed he is referring to, titled "Climate of Complete Certainty". It uses the example of Hillary Clinton's vote-modelers in the 2016 election to begin a conversation about the evils of misguided certainty which concludes, "Perhaps, if there were less certitude about our climate future, more Americans would be interested in having a reasoned conversation about it."  This was Bret Stephens' first column for the NYT -  he had, I gather, been brought on board as part of an explicit decision by that newspaper to broaden the range of editorial-page voices that its readers hear. I didn't know that, though, and was completely unaware of the upwelling of outrage that his first article had apparently generated. But I had just been writing some material for my "Math for Sustainability" textbook explaining the probabilistic nature of climate forecasts, and why that does not absolve us from responsibility for (potential) action now, so Mr Stephens' insinuation that "uncertainty" might justify less action, even if more conversation, aggravated me a bit.  Anyhow, I finally determined to try to write something in response to Alan Jacobs' help-wanted request.   Here it is.

Mr Stephens begins his piece with a reminder that we don't know as much as we think we know, however well-informed we are with data and statistics; and for his example he chooses the data-heavy Clinton 2016 campaign, which crashed and burned despite their advisers' apparent near-certainty of a win going into Election Day.   To illustrate the point here, let's imagine that this graphic represents what the Clinton team believed to be the probability distribution of likely outcomes (number of electoral votes) on Election Day:

Maybe Clinton hoped for this?
The gray shaded area shows their estimated chance of a Trump win - a couple of percent or so.  If, however, the team had been more alert to the inherent uncertainty of their data - had less of a "limitless faith in the power of models and algorithms to minimize uncertainty and all but predict the future" (to quote Mr Stephens), they might perhaps have allowed a substantially wider range of error (variance) and so produced something like this

But ended up with this?
which shows about a 30 percent chance of a Trump win (roughly what Five Thirty Eight was suggesting the day before the election). Then they might have done something differently (Stephens doesn't say exactly what different actions greater epistemic humility might have produced, but that's fair enough, since he is using the Clinton campaign only as an example).

It's worth observing that the outcome of an election is a step function of the underlying state variable.  If Clinton obtains less than 269 electoral votes, she loses; more than 269, she wins. Setting aside the tiny chance of an Electoral College tie, these are the only two possible outcomes and they are at opposite extremes.  Most natural processes do not have outcomes of this step-function, yes/no type. The range of outcomes in the climate system is not x percent chance of the apocalypse versus 100-x percent chance that nothing bad happens at all, but a probability distribution of outcomes running from "dodged a bullet there" through "quite bad" and "terrible" and "disastrous" to, yes, "apocalyptic" at the other end.  I am sure that Mr Stephens understands this, but by beginning with an example with a step-function outcome he may dispose his reader to the belief that the same kind of outcome function is relevant to climate change.

To contrast with the (completely imaginary) curves that I drew above, here is a real example of a probability distribution for the most important quantity related to climate change, the so-called climate sensitivity, which is the expected (medium-term) average warming of the Earth in response to a doubling of atmospheric carbon dioxide concentrations:

This is from Gerard H. Roe (no relation!) and Marcia B. Baker. Why Is Climate Sensitivity So Unpredictable? Science, 318 (5850):629–632, October 2007, but many similar graphs and tables can be found in the more recent literature (including the IPCC reports). The scale along the bottom represents the amount of warming in degrees Celsius in response to a doubling of CO2 (a point that we have not yet reached, but one that we are well on our way towards; the doubling of CO2 is used as a reference value, as the corresponding distributions for other percentage increases in CO2 can be calculated from this one).

What is there to notice?  Well, first of all, there is a wide range of uncertainty - the "high confidence" range for climate sensitivity in the latest IPCC report is between 1.5 and 4.5 degrees Celsius.   Second, the distribution is heavily rightwards skewed: in other words, a hugely worse outcome (7 degrees, anyone?), though unlikely, is more likely than a neutral outcome (0 degrees, it was all a scientist's bad dream).  Because of this skewing, if I pulled the same trick with this distribution as I did with the Hillary-election one above (modeling "epistemic humility" by increasing the variance while leaving the mean more or less the same), we would end up with a picture where the outcomes were strongly weighted towards the disastrous.

"Anyone who has read the 2014 report of the Intergovernmental Panel on Climate Change knows that...much that passes as accepted fact is really a matter of probabilities" writes Mr Stephens, presumably referring to information like that contained in the graph above.  I find myself puzzled by this assertion.  What exactly is it that "passes for accepted fact" that is really a matter of probabilities?  You will not (I think) find anyone claiming that a 2 degree climate sensitivity, for example, is an "accepted fact".  When Mr Stephens went on to say, "By now I can almost hear the heads exploding", I checked my own head and found it disappointingly unexploded.  Maybe I am just missing the point, but the idea that our grasp on the future is uncertain and involves probabilities seems simply banal. (That we can narrow down those probabilities, in the case of the huge and complex global climate system,  as much as we have - that seems pretty amazing. But also a different discussion.)

So, our grasp on the climate future (as on other aspects of our future) is uncertain.  A probability distribution like the one above represents the best currently available state of our knowledge, but allows plenty of room for uncertainty.  Does that mean that a rational person would take no action, awaiting greater certainty?  The conclusion does not follow in general.  In the book, I use the example of auto insurance to illustrate this.  Suppose you buy a new car, worth, I don't know, $20,000.  You wonder whether to take out (comprehensive) insurance - coverage that will pay you back the value of the car if it is wrecked in an accident.  If you knew for sure you were going to wreck your car next year, then buying insurance would be an easy decision. But that level of certainty is not required for the purchase of insurance to be rational. What you need to know is that a wreck is a substantial risk (you can set your own criteria for what counts here) and that, if it occurred, it would do you substantial damage (maybe you can no longer get to work, you lose your job, you can't support your family, your relationship comes unglued - all of these are real possibilities).  Then buying insurance, if you can afford it, is rational.   If you are rich enough to write off the loss on Wednesday and go buy a new car on Thursday, you don't need insurance.  If you are sure enough that you will never suffer a major accident, you don't need it.  Otherwise, if you can afford it, you do.

All of this thinking is transferable to the climate situation.  The probability distributions above, together with our current "business as usual" trajectory, suggest a substantial risk of significantly damaging our planet.  If that damage were to occur, we are not rich enough to afford a replacement Earth - nor, so far as we know, is there any showroom where shiny new planets are on offer.  Action to mitigate climate risk - "insurance" - seems to be rationally indicated.

At this point, I think Mr Stephens might want to draw attention to the phrase "if you can afford it" that I inserted above.  He writes,  "Demanding abrupt and expensive changes in public policy raises fair questions about ideological intentions."  Again, he uses the language of step functions ("abrupt and expensive changes") to characterize policy responses to climate science: once again, I feel, he may dispose his reader to believe that there is a binary choice between an "abrupt and expensive" response or no response at all.  But many climate responses - perhaps particularly those that have a market element rather than setting hard regulatory boundaries - can be "switched on" slowly or rapidly, and ramped up to whatever level seems to bring about the needed effects.  For a specific example, this would apply to the "carbon fee and dividend" proposal advocated by the Citizen's Climate Lobby (a revenue-neutral carbon tax, whose proceeds are returned equitably to all citizens rather as Alaskans today receive dividends from the Permanent Fund).

I don't feel I've done a good job in meeting Alan Jacobs' challenge: that is, explaining why there is a conflict between believing that "portrayals of the future like those we see in Kim Stanley Robinson's novels (shout-out, by the way, to Mr Jacobs for getting me into reading them, they are fascinating) are plausible projections from the best current data" and believing that Mr Stephens' article is "not only reasonable but self-evidently correct".  Maybe I don't see too much of a contradiction either: after all, I already described one of his key points as "banal", which surely includes "self-evidently correct".   I'll go further and say that this also seems to me "self-evidently correct":
Censoriously asserting one’s moral superiority and treating skeptics as imbeciles and deplorables wins few converts.
The problem with climate change - for that matter, the problem with many related environmental issues as well, like deforestation, water scarcity, resource depletion, you name it - is that we are the problem, all of us, even those of us who think we are doing our best to be "green" or "low impact" or whatever we like to call it. We are all in the grip of a system that is bigger than us, what the New Testament might identify as one of the "principalities and powers", and because of that there is simply no place for moral superiority here - not to mention that it is pretty useless as a communication strategy, as Mr Stephens reminds us and as much recent research has confirmed.   Allow me a personal aside here. A few years ago I decided that one of the best things I could do as an educator was to teach math-related sustainability ideas, including the basics of climate change, to nonspecialists (hence the forthcoming book, and many posts on this blog).  I want to equip students with the tools to evaluate the data for themselves - risk, modeling, change, decision-making under uncertainty, all these are tricky topics on which a mathematician can help (I say no more than that) a conscientious citizen to achieve clear(er) thinking.  There is no assertion of moral superiority in my classroom and (I hope) no-one is treated as an imbecile.  In fact, I do not find the students by and large to be skeptical of the scientific consensus on climate change - I would be glad for some more skepticism now and then, and I think it would be educationally beneficial too.  Moral issues more often manifest themselves in a sort of environmental Pelagianism - if only we would all recycle our water bottles and take shorter showers, the world would be fine. I am still thinking about the most effective way to respond to that.

So can I find any conflict for Alan Jacobs, as he requests? Well, I am disappointed - and I think he should be too, if he seriously understands KSR's novels as plausible projections from best current data - by Mr Stephens' concluding sentence
Perhaps if there were less certitude about our climate future, more Americans would be interested in having a reasoned conversation about it.
Undoubtedly, reasoned conversation is important. Undoubtedly, moral smugness is an obstacle to such conversation.  But, by leaving things here, Mr Stephens lets it appear as though, given that "much that passes as accepted fact is a matter of probabilities", more "reasoned conversation" is all that can be justified at this stage.  Is that what he actually believes, or does he also accept the argument, sketched above, that "uncertainty, informed and bounded by science, is actionable knowledge" (quoting former NYT columnist Andrew Revkin)? I think his article can be read either way.

Once again, the "sophisticated but fallible" models that produce the probability data above are the best guide that we currently have to our future planetary trajectory.  To use their acknowledged fallibility as an excuse for inaction is like not hitting the brakes as you approach the hairpin bend because you are not sure whether the speedometer reads 71 or 72. If that is what Mr Stephens is recommending, I think there is an ethical conflict for someone who believes KSR's novels are plausible projections of the "business as usual" future.  But if you read him simply as suggesting ways in which climate messages can be communicated more effectively (I think Mr Jacobs uses the word "winsomely" in another post) - well, I think his advice is good, and I don't see a conflict.  And I don't honestly know which reading is the correct one. I look forward to reading future writing from Mr Stephens for the NYT in which he clarifies this point.

No comments: