Hacker News Reader

Kevin Anderson (@KevinClimate) speaks prolifically [1] about how the (perverse) incentives in climate science are actually in the opposite direction from your expectation, towards less alarmism. Policymakers want to hear that the problem is fixable with little disruption to the status quo. These policymakers and their representatives are more likely to dismiss climate predictions that require radical action. This leads many climate modelers to include tricks such as negative emission technologies (NETs) -- which are pure science fiction -- in order to keep emission reduction prescriptions friendly to their audience. Thus, the models become more moderate while the situation grows more dire.

One recent example of such a dynamic was Germany's recent plan to phase out coal (by 2038). The committee that drew up the recommendations was overruled on several key elements, such as the decision to not to open a new coal power plant this year. Although the committee was and is correct that opening new coal power plants is not compatible with our CO2 emissions goals under the Paris agreement, which are in turn our best estimate of the maximum we can emit and still have civilization past the next couple decades, lawmakers didn't want to hear it.

1. https://www.youtube.com/watch?v=mk49vAmNoiQ

With the violent hammering that scientist raising the alarm have received over the past 20 years or so, I would suspect any results-selection bias to have pushed in the other direction. What me may be seeing now may be reduction in bias, caused by the ample observations coming up.

> I'm afraid that the incentives are to credit models that predict more warming just a bit more than those which predict less;

You've got it backwards. The models are known to be conservative because "alarmist" scientists have received a bad reputation when they are proven wrong. It is safer to pretend to be ignorant and underestimate the numbers.

I'm reminded of Feynman's famous anecdote about the Millikan oil drop experiment: Millikan underestimated the charge of the electron, and over the ensuing years, subsequent measurements slowly but surely converged on the correct value.

Why? Because experimenters who got the correct value figured it must be too high, and ran the experiment again until they got a more-erroneous value, one that was nonetheless higher. Eventually it worked itself out.

But climate modeling isn't measuring something objective and static like the charge of the electron. It's prediction. I'm afraid that the incentives are to credit models that predict less warming just a bit more than those which predict more; the expected effect would be that the predicted warming effect would only slowly approach the more dreadful reality.

Please note that I'm not calling climate science a "Cargo Cult", as Feynman put it; I accept the basic premise of anthropogenic warming on first principles, and see that historic measurements (not to speak of recent ones) show a warming trend. There is cause to be extremely concerned.

Nonetheless, it would be surprising if climate modeling, which is inherently inexact and involves irreducible guesswork, didn't drift toward the less alarming, due to similar social forces to those identified in Feynman's essay.

You're not actually saying anything of merit here. Flipping the sign on your paragraphs requires almost no work to get a meaningful result (which doesnt contradict known facts). You've made an assertion (climate models are likely to overpredict warming rather than underpredict) and handwave at some random shit that makes it sound nice. The actual assertion has zero argument behind it.

So close, but not quite.

The oil drop results (Millikan was less than 1% off, by the way) show the pressure on scientists to conform with the current status quo, the pressure NOT to release shocking or unexpected results. In climate science, that means underplaying the results, biasing them towards less catastrophic findings, and there's a great deal of evidence that that is what has been happening. For instance, the gold standard climate documents, the IPCC reports, represent a consensus and therefore if 99% of scientists believe 100 units of climate change is coming, and 1% of scientists believe no more than 30 units of climate change is coming, 30 units is what is reported (the amount everyone can agree on). This is a huge underreporting bias, only slowly corrected with each new report.

Good on you for your scientific approach, you just basically reversed the signs in your equation.

I've read the AR5 synthesis report, and your description is simply inaccurate.

To use your numbers, the IPCC reports that 99% of scientists believe 100 units of climate change is coming, and 1% believe no more than 30.

It's less than 200 pages, I recommend actually reading it, rather than just guessing how the sausage is made, as you've done here.

The ultimate proof is actual data that's steadily being collected as time moves forward. That surely culls off bad models as they're proven wrong. But I think it would only be helpful at reducing bias if we just look at the surviving old models rather than making new or updated ones that incorporate the new data or else the new models might still have a systemic bias in one direction or other.

> I'm afraid that the incentives are to credit models that predict more warming just a bit more than those which predict less; the expected effect would be that the predicted warming effect would get larger over time.

This actually doesn't make any sense! In your valued Milikan example, the problem was subsequent experiments were biased towards lower values, meaning that for climate models the bias would be expected towards under heating models. In other words, the lesson from Milikan is that there is a risk our models aren't heating fast enough.

So by the quoted example we should be worry that our models under predicts, and yet Milikan is stood up to argue the models are too wild in their warming predictions?