Hacker News Reader

Comment by bsaul | Hacker News Reader

otoh : it seems to me covid models used to dictate public policies proved to be off by at least one ( if not two ) orders of magnitude. Countries that didn't confined ( brazil and sweden) didn't show that much of a difference compared to some others that did (france). The spread and impact of the virus proved to be extremely hard to understand and model correctly.

Which of course, should have people wonder if climate change model couldn't also be off by orders of magnitude ?

Now of course because one thing proves to be hard to model doesn't mean all are.

But looking a the history of public climate catastrophic announcement, how do we know that this time is the time when models of immensely complex phenomenon will be correct ?

View on HN

>Countries that didn't confined ( brazil and sweden) didn't show that much of a difference compared to some others that did (france).

Huh? France had an early outbreak, instituted lockdowns and have since crushed their numbers. Brazil is still on fire.

Here's a graph for Brazil: https://91-divoc.com/pages/covid-visualization/?chart=countr...

Here's a graph for France: https://91-divoc.com/pages/covid-visualization/?chart=countr...

These look nothing alike.

I agree that they are different graphs, but to the parent's point are they at all concerning?

Here's the logarithmic graph for Brazil:

https://91-divoc.com/pages/covid-visualization/?chart=countr...

It's a breatiful plateu, which indicates that we do not have the super-scary exponential growth we were all warned about by these models. As the parent said, the models were off by an order of magnitude at least.

It is not the case that Brazil (at a state level) is doing nothing. It is the case that Brazil and France did things to very different degrees, and as a result have seen stunningly different results, as evidenced by those graphs. (And contradicting the assertion of the GP poster.)

Yes I totally agree with both of your comments. I wanted to change the topic of conversation to the point the parent made. Maybe I should have replied to parent with that point. Thanks for showing me that site!

I still want to know if anyone thinks the parent's skepticism of climate change models is/isn't warranted given the fact that the scientific research community (epidemiologists, virologists, and scientists generally [AFAIK]) basically aligned completely with the view that anything other than extreme responses to Covid would end in disastrous ruin by generating and pointing to these models. I want to be clear that I am not questioning the science of climate change here, but how do we know that the models we rely on to predict the future effects of climate change will be accurate? Could Covid really be an indicator that the scientific community is the boy who cried wolf? I'm not saying I think that, but I don't know enough to completely discredit the position.

They look nothing alike because they're not plotting the same data?

In the links you posted, the graph for France shows new deaths per million, the one for Brasil shows new cases.

This is a graph for deaths comparing France (total lockdown) and Sweden and Brasil (no lockdown):

https://91-divoc.com/pages/covid-visualization/?chart=countr...

Brasil never had the peak that we had in Europe. And the difference between the total lockdown of France and Sweden's business as usual is visible but not dramatic.

I find it incredible, and humbling, that we understand so little about the spread of a simple respiratory virus.

If I pasted the wrong link, you can use the small drop-down menu to switch from one metric to another. You'll see essentially the same trends.

I think the difference here is that climate change has been studied for decades, meanwhile COVID-19 is more or less a new phenomenon since Christmas. The former has had a lot more time to be observed and have hypothesis’ tested.

What’s your guess that our climate change predictions are off? And by how much?

i think the number of times we've seen articles saying how there's going to be no ice anymore in 5/10 years on the poles, during the last 25 years, should be a hint that estimates aren't even reliable on a 10 years time range. I wouldn't be too afraid when the same people predict horrible things in 100 years.

i sometimes have the feeling that those models don't do much more that extrapolate exponentials and lines, without any kind of comprehension of the underlying phenomenons ( at least that's my feeling with covid models.. i know climate models are actually better than that)

I've never seen a reasonable (or any, in fact) person claim there's going to be "no more ice" on really any timeline. Could you link one of these studies? Sometimes people will make claims that are wrong, but I would think twice before doubting the consensus of (practically) the entire scientific community (the consensus being climate change is real, and is real bad. To the point that we need to do something immediately)

https://youtu.be/MsioIw4bvzI

> i sometimes have the feeling that those models don't do much more that extrapolate exponentials and lines, without any kind of comprehension of the underlying phenomenons

At the end of the day, all we have is our feelings. In terms of our decision-making and understanding of the world, though, I think it's important to recognize the general distributed consensus from observations around us. If you're feeling those pangs of doubt, I'd encourage you to try to find the dry citations behind the articles that you've mentioned.

Covid is new, but infectious disease isn't and pandemics aren't. Covid models are just epidemiology models with a given set of covid specific input parameters.

If a climate model doesn't work with a certain set of unpredictable input parameters, that means the model is either broken or irrelevant.

Same for these covid models. Don't confuse the data for the algorithm.

Maybe because at the time they were modeling Covid for a few months.

We've been modeling climate change for something like 50 years and have seen our best models confirmed by reality.

Sweden did much worse than their next door immediate Nordic neighbors, Norway, Finland. They had more than 20x per capita deaths directly because of that.

>Which of course, should have people wonder if climate change model couldn't also be off by orders of magnitude ?

Does the world even see the accuracy of modelling that dictated public policies?

Actually, I am hopeful that the general sentiment of science-led public policy is more likely to be improving this year. Exactly because countries that didn't confine showed a VAST difference compared to some others that did.

Privileged countries with functioning governments were able to trust their experts, act early and decisively, resulting in not digging mass graves.

https://91-divoc.com/pages/covid-visualization/?chart=countr...

Does the world see this? I hope so.

Sweden has it way way worse than all its neighbors. Six thousand people left dead. Compare to 150 in Finland.

Comparing the French and the Swedish numbers seem to indicate that the virus ran rampant in France for a shorter period, while it was less severe but was active over a twice as long period in Sweden.

To me it is a bit of a mystery why France, who arguably did a lot to contain the virus, ended up with a more or less identical result as Sweden did.

Sweden: https://91-divoc.com/pages/covid-visualization/?chart=countr...

France: https://91-divoc.com/pages/covid-visualization/?chart=countr...

Sweden actually had 11.057 deaths in December 1993 which is still a bit more than the 10.458 deaths for April 2020. But the difference in public and government reaction to those two events is like night and day.

Source: https://emanuelkarlsten.se/more-swedes-died-in-one-month-199...

>To me it is a bit of a mystery why France, who arguably did a lot to contain the virus, ended up with a more or less identical result as Sweden did.

Are they, though? The long tail in Sweden seems to suggest a much larger per capita death count overall doesn't it?

Wouldn't the initial peak in France and Sweden being the same suggest these are all deaths from the very start of the pandemic? People take a while to die. The work France did to contain the virus seems to explain the fairly rapid tail-off to zero, vs the slow tail-off in Sweden.

I'd imagine -- with no evidence to back it up, mind you -- that there are likely also a lot of Swedes who took their own protective measures despite their government?

Well, since Sweden did little to "flatten the curve" they had more death, earlier. That won't change the total. Look at the number again in two or three years.

Not that there's a reason to lose sleep over six thousand dead in a country of ten million (some 80000 die every year anyhow).

It was less virulent and more deadly than some people thought originally. There were estimates that the factor of real cases compared to positive test result was really high - that a lot of people were infected but asymptomatic. Turned out to be wrong.

So the "let it spread, it can't be stopped" strategy turned out to be wrong and caused a lot of deaths.

The countries who had more serious isolation measures had a lot lower amount of deaths.

> how do we know that this time is the time when models of immensely complex phenomenon [climate change] will be correct ?

Models of all degrees of complexity agree, from zero-dimensional energy balance equations, through 1D, 2D, and 3D with progressively more and more detail and biophysical processes added.

They all agree, within the amount of detail that each complexity level is capable of.

We also have the evidence of history, as we do for pandemics, but our current political leaders paid no attention to the history because "this time is different".

Saying that some modeling is inaccurate is not a sufficient argument to disregard other modeling. Beyond that, we have already seen catastrophic climate change results in nature, we just haven’t felt catastrophe level impacts for most humans.

We don’t know if the models are correct, but we know the direction is accurate and the impacts are already meaningful. Models are just educated guesses, no one having a good faith argument would say otherwise.

Now those are completely different topics and I must wonder why those are referred together. One had the potential of killing up to 2% (earlier estimates were as high as 5%) of humans (mostly sick and elderly) of which there are plenty, while the other has the potential of wiping out all life on earth (which makes it all life as far as we know).

Your sibling comments are seem to be mostly trying to poke holes in the statement about COVID. Independent of our performance modeling COVID, we need to admit that we get things wrong and this is applicable to climate modeling. AND there's uncertainty. The chance of models getting it correct is exactly 0% (all models are wrong, some models are useful). How far off is a big question (uncertainty quantification). I think the scary part with climate modeling is that things are looking scary even if we're over-estimating warming. I think models are probably under-estimating. Bifurcations are an important, but tricky piece of chaotic dynamics. I'm really worried that we don't know all the bifurcations, "tipping points", and feedback loops. But, I'm also not a domain expert.

Edit: Sorry, also meant to ask for a reference on _all_ life disappearing due to climate change. Perhaps all human life, but life in general seems pretty resilient. Eg. recent headline [1]: Scientists revive 100 million-year-old microbes from deep under seafloor.

[1] https://www.reuters.com/article/us-science-microbes-idUSKCN2...

I don't care whether global mean temperature rises one or two Kelvin in the next 30 or 50 years. The discussion around that is a fatal distraction. I care whether humans will be able to live here in hundred and life in general exist in five hundred years. Hence I don't care about the minutiae of the models. It is quite difficult to predict how much influence decreasing albedo has on the rate by which the temperature increases, same for thawing permafrost in Siberia releasing an unknown amount of Methane. It is not difficult however to realize that those are positive feedback effects. There is no life possible on Venus. Now, earth is a bit further away from the sun, so won't reach the same temperature, but it doesn't take lead to melt to kill all life as we know it.

Reply on news.ycombinator.com