Hacker News Reader

Comment by justadudeama | Hacker News Reader

I am not a climate change denier by a long shot, but stories like these give a ding to the credibility of climate scientists.

> "So we can't throw them out yet."

This might be an elementary view of science, but I think there is a danger here that while everyone is making their models, if anyone is an outlier they go and tweak their model to match the patterns of others: 'Klaus Wyser’s group "switched off" some of the new cloud and aerosol settings in their model, he said, and that sent climate sensitivity back down to previous levels'. That seems to me a questionable reason to "switch off" part of the model - you should create the most accurate simulation possible and trust the output, not tweak the inputs to match literature data.

View on HN

Tweaking the inputs would help one to identify what factor is causing the difference, which allows you to focus on that. It doesn’t mean you permanently leave it out.

And it does not make a credibility issue. This is how science works — it is a process that strives to be as accurate as possible, not come up with a static answer and stick to it (that’s what religion does).

The ability for science to change and adapt is what makes it so strong. That is the message that needs to always be driven when discussing science; not throwing hands in the air and saying “why do these scientists keep changing their answers? They obviously don’t know what they’re doing”

These are valid points, and it is important to figure out what changes in the model is causing such a dramatic change and scrutinize it more.

> The ability for science to change and adapt is what makes it so strong

To me this article is them _resisting_ change, looking for a reason NOT to accept it, because it goes against what they previously said. Maybe this is all to early, and in a year or so we will be seeing 4 or 4.5 degree predictions, but this article describes them trying to change their inputs to match previous answers, not getting new answers.

If your model suddenly changes its predictions, you need to make 100% sure you know why and that is valid.

Just imagine they would go public with this "Scientists found climate change is worse than previously thought with updated model" and then weeks later find an error in the changes made to the model "Scientists made mistakes, climate change less severe than previously thought".

What do you think would hurt their credibility more?

Well this is the problem. People are reticient to make public a private finding that is different, because it is probably wrong and they'll probably lose credibility. But if everyone is doing that...

They are not reticent to make it public, they are publishing papers and are asking for peer review. There is a public newspaper article about it with quotes of the involved scientists. They couldn't be any more transparent about this.

They are not throwing away their findings and they are not burying it. They are working to valdiate/invaldiate them, because they are scientists.

Again, that's the way that science tends to operate - a la Kuhn and his Structure of Scientific Revolutions. You have some novel research that seems to contradict well-established consenus. It comes under close scrutiny, because science tends to be conservative. If after giving it a good kicking, its clear that the old consensus is wrong, science takes the wrecking ball to it. But it prefers not to get the wrecking ball out of it can help it.

> You have some novel research that seems to contradict well-established consenus. It comes under close scrutiny, because science tends to be conservative.

Indeed, even people we nowadays consider brilliant minds of their time, encountered plenty of resistance [0] during these very same times because their proposed ideas just seemed too outrageous.

[0] https://en.wikipedia.org/wiki/Criticism_of_the_theory_of_rel...

It certainly does hurt the credibility of people saying we need drastic changes now based upon the models. If the answer might change next year, then making a drastic change based upon the current projections seems foolhardy.

The answer is going to change every year, with new inputs from the prior year. That is how prediction models work. Using your logic would lead to never making any changes, which I don't think will really help.

They say we need drastic changes based on the best evidence we have along with given the likely outcomes if change isn't introduced.

What you're saying is that you have established a prior that undermines all further evidence presented to you, because it has changed substantially in the past. However, that seems irrational to me.

Yes it would be terribly foolhardy for us to stop polluting the atmosphere prematurely.

This isn’t really a helpful reply, because of course it’s not a question of magically reducing pollution.

There are two basic modes of reducing pollution. Technological advancement, and conservation.

In the case of technological advancement, you get less pollution per unit of output as the same or lower cost. As long as capital is fairly cheap, these advances tend to propagate quickly through a competitive market.

In the case of conservation, it’s more a question of which groups of people should conserve (i.e. suffer) for this outcome, and by how much.

Why do governments fail to take advantage of this opportunity to inject money into their economy? Didn't Trump promise to bring the manufacturing jobs back? How is he going to achieve that if he's not going to rebuild the current power infrastructure or discourages companies from investing into the lucrative EV market?

I'm seriously wondering why there is any reason to oppose technological progress at all. What can they possibly gain by sticking to the old technology? Who's going to benefit from something so short sighted?

We might be steering into an ice age and Co2 induced global warming is the only way for humanity to survive.

That would be cool, please post links to your model's source code and data set.

FUD. The statement that we need drastic changes is not dependent on the details of the projection and is not based on it. Further, climate models have correctly predicted the heating we have already caused.

We have known the rough level of the climate sensitivity (delta T per doubling CO2) in the current macro state for decades. We have observed it. This science is incredibly settled. We don't know where the tipping elements are and we don't know how the impacts will be distributed.

Questioning the credibility of people calling for action might look like you are being the circumspect and rational one, asking for action proportionate to the evidence. In reality you are endorsing an ignorant position that ignores the evidence we have. If this is not wilful ignorance then please read up on this. The IPCC report summary is a good spot to start. Sceptical Science also often provides good writeups: https://skepticalscience.com/climate-sensitivity.htm

The fundamental physics that drive the sensitivity are straightforward. Energy balance and water vapour. That gets you a ballpark figure that is confirmed by the last 100 years of emissions.

Individuals, corporations, and the politicians who set policy all operate on imperfect knowledge in many many domains as a routine matter of course.

It is a denier talking point to claim that we just don't know enough, or that the models are always being tweaked and we should just wait until scientists have "the final answer" before doing anything.

Hey, let's stop giving NOAA weather forecasts because they aren't 100% accurate. Let's close down the stock market until everyone can decide what the values should be. Let's disband the military because not all military leaders have the same opinion.

If we are mature enough to recognize that imperfect but credible models are worth acting on in other domains, why do we hold climate change to impossible standards?

Sadly no, the hypotheses that are now being shown to have been too optimistic were the ones that implied drastic changes. The updated models are telling us something else, you'll figure it out, and/or it will figure you out.

Try thinking from a basic Risk Analysis point of view. If climate scientists are wrong and we do something we will invest heavily in infrastructure and efficiency, reduce our environmental impact, and reduce air pollution for our children. If they are right and we do nothing then the 9 to 11 billion of us still living on Earth will be crammed up against the North and South poles while sea level rise claims many small nations. We will run low on arable land, many of our animal and plant species will die, and the planet our children inherit will be a dirty hellscape that’s too hot to live on.

Given that, how could you possibly advocate for doing nothing?

But it should make the scientists more circumspect about public policy recommendations. They're not making evidence-based policy, if yesterday's models are evidence and today's models are not.

> This is how science works — it is a process that strives to be as accurate as possible, not come up with a static answer and stick to it (that’s what religion does).

That's fine. Many public policy advocates throw around the term 'settled science'. But based on what you said, is there such a thing?

But it should make the scientists more circumspect about public policy recommendations.

From the article:

Climate models have been doing a fine job projecting warming for a long time. A recent study compared models as old as 1970 with observations made in the decades since. Some models warmed up too much, and some too little, but 14 of 17 past projections turned out to be consistent with the measured path of global average temperatures.

Scientists can not give definite answers in such complex systems, but it is the closest we can get in extrapolating the trajectory that we are on and computing what the effect of various measures is.

I am not sure what else one could base policy on.

> That's fine. Many public policy advocates throw around the term 'settled science'. But based on what you said, is there such a thing?

Yes, there is. Just because there is no absolute certainty, doesn't mean it's all just baseless speculation. Just because, as a matter of scientific methodology, atomic theory is not absolutely certain (i.e., we could hypothetically find evidence that contradicts the model), doesn't mean someone just made it up and it has no predictive power.

I think you're missing the point. From how I read the parent comment, the problem they see is how scientists are using the data available to them. When you make a hypothesis, you are supposed to validate or invalidate it based on its predictions - not change the predictions to fit the evidence.

If you change your model (hypothesis) to fit the facts, your model is merely descriptive of the evidence that is already available, and useless for prediction - as proven by the act of changing the model. That's how science works.

Uh, you realise this article is about work in progress, right?

If you want to cast it in the rigid hypothesis testing language (not a very good description of the actual practice), the description is this:

A physically motivated modification to existing tools lead to unexpected results that are not understood yet. As part of figuring out the cause of the new result they are running a series of smaller experiments to test various hypothesis of what aspects of the modifications matters most for the new results.

In effect, objecting to the practice of turning things off and on again is to object to the very notion of doing experiments, which is not very scientific.

I don't know how you could have reached that malicious interpretation of why that part was switched off. To go and impugn all of climate science in the basis of a misinterpretation of motivations would indicate that you have internalized the denialists claims and look to seek to justify them.

To point out just how wrong your misinterpretation is, the very next sentence has the "why":

> A new research paper co-authored by Zelinka from the Lawrence Livermore National Lab likewise pointed to the role of virtual clouds in determining the results.

The key phrase being "role of virtual clouds." The motivation wasn't to slavishly reproduce results from the literature, and to even suspect that it completely outside the bounds of reasoned discourse. It comes across as if you are trying your hardest to find something to nitpick, some small phrase that you can misinterpret in order to raise doubts in those who aren't paying close attention.

This is exactly how denialists and propagandists behave.

That's the main problem I have with climate science. They don't really have the luxury of proving by experimentation, and I have seen enough perfectly good backtestings in my industry (finance) that go wildely off as soon as they go live, that I have a fundamental mistrust of backtesting as a way to calibrate a predictive model.

Especially because they seem to test it by making it predict history, but we don't have anywhere near the same amount of historical data. So what are you testing, really? Just that it accurately finds average temps?

Climate models as far back as the 1970s have predicted the temperature changes in the time from then to now -- those are not predictions that are coming from hindsight, but rather predictions where their effects were observed over the last 50 years.

If you are unfamiliar with climate sensitivity, this is a good place to start: https://skepticalscience.com/climate-sensitivity.htm

From the article: "A recent study compared models as old as 1970 with observations made in the decades since. Some models warmed up too much, and some too little, but 14 of 17 past projections turned out to be consistent with the measured path of global average temperatures."

So we actually did an experimental test of the predictions made in 1970, and those predictions turned out to be largely correct.

We should probably halt this experiment on ethical grounds.

Typically it's not obvious what to switch on and off for "accuracy".

If your model produces wildly different results from previous models, either you hit on some really big discovery or you made a mistake. An ethical scientist will try to rule out the latter possibility before accepting the former.

It goes deeper than that. If you can tweak your new model to correspond to an older one for which you have results available, you must run tests in which you apply these tweaks and verify that you get the same results. If you get different results, one of the models simply isn't doing what you think it's doing and you need to understand that. And in the vast majority of cases, it's likely a mistake in a model, not any kind of new science.

One pretty general debugging/analysis strategy for a complex model is to deactivate it piece by piece until you get it to a point where it should reproduce a result that you can verify by other, independent means. E.g. tweak the parameters of the model so that reproduces a result that is known analytically. For instance, whenever it is possible in a model, a good sanity check is to set parameters that must lead to e.g. a perfect conservation of certain quantities.

I'd propose that has more to do with the headlines than any of the actual science. Model tweaking is exactly how a lot of this works - models are supposed to reflect reality, so part of how we explore them is to find out what models fit past (and predict current/future) reality. This is both intuitive, and based on theoretically sound inference (see Approximate Bayesian Computing for an exploration of the connection between simulation and formal inference).

> Why didn’t they discover that the new number was higher right away?

http://calteches.library.caltech.edu/51/2/CargoCult.htm

The reason clouds are starting to be simulated now where they weren't before is probably because we have only recently had the computing power available to do this at the level of detail required. Switching the 'clouds' feature on and off is part of the model verification process. Listen to this podcast[0], it's a in depth technical discussion about weather forecasting models with the actual scientists that create them. They discuss model verification for weather models and they touch on how this also applies to climate modelling.

[0]http://omegataupodcast.net/326-weather-forecasting-at-the-ec...

The climate models are not settled.

Sadly the political activists have hijacked the normal scientific process. And this allowed some pseudoscience (with alarming predictions or alarming observations) to mix in too.

In some science news I read that the Arctic Ice-sheet is growing back, and that Envisat showed that the oceans were not rising. So very likely the climate is more complex than some alarmed scientists thought.

This idea, along with the null hypothesis, are simply not honored by this part of the scientific community. Climategate proved this ten years ago. It's a doomsday cult and they'll stick you with the "D" word if you don't narrative-fit. Check out Ned Nikolov's work for a better alternate theory.

This part of climate "science" always makes me put on my Popper hat. The social aspect of being the outlier (grants dry up, career goes away) makes me skeptical of the IPCC reports, just not in the usual direction. I fear reality could be much worse.

I can't ding the climate scientists too much, though... I'm not sure there's a better way to study this stuff.

In this case have grants actually dried up or careers gone away? Nope.

I don't think I implied they did. I'm only saying that humans are social creatures and that scientists are susceptible to peer pressure (even in "hard" sciences with lots of possibility for experiment). I imply no maliciousness, but am trying to voice my fear of the impossibility of knowing whether or not the consensus warming scenario is a "reversion to the social mean", that is, we just have to wait and see if warming is worse that expected.

But I'm only a bit surprised this would be such a controversial opinion here.

Reply on news.ycombinator.com