Hacker News Reader

It shouldn't even be possible to compare wildfire severity/frequency to climate change. There was no forest "service" or firefighting 100 years ago and prior. The forest and wilderness were wild. But now we play superhero and put out every fire we can (then the government parasite takes over and needs more funding to justify their bloated existence). Then the fuel level reaches unprecedented levels. We need to start logging to thin the forest to its natural levels, and stop putting fires out. We have known for decades they are part of the natural cycle and essential to some species to reproduce. But how will we justify our jet air fleet?

The problem with climate change and wildfires is a moral risk.

If climate change can be blamed, it means bad management of forests is not the problem, so there is no need to spend money and effort on it.

In fact, anything that raises the risk should mean that you put more effort into better management.

The other problem is that the models are extremely complex and there are limits on the data available for back testing. Its similar to the problems with economic modelling, rather than those typical in say physics. We also do not have enough planets or time to test hypotheses!

>the regression is going to end up with non-zero coefficients on it

This is just not true. Plenty of regressions have coefficients that arent statistically distinct from zero.

>But this approach cannot tells us if those inputs are drivers, much less the main ones.

Again, untrue. There are plenty of statistically appropriate ways to estimate causality. You might consider looking into the latter-day work of Judea Pearl, a well known computer scientist. "The Book of Why" seems like a decent place for you to start, because ut is for the layperson, and you have a ton of fundamental errors in your statements of "fact."

>unfortunately with media, politicians, and the general public so science illiterate that they're unable to engage with this research on its own terms

You should also add "the confidently incorrect" to your list!

> This is just not true. Plenty of regressions have coefficients that arent statistically distinct from zero.

Sure. But you are responding to something I didn't claim, or at least didn't intend to claim. If you throw in a spurious piece of data that happens to exhibit the same trend, it's going to end up a non-zero coefficient. But that doesn't mean there is a causal relationship.

[ https://www.tylervigen.com/spurious-correlations ]

> There are plenty of statistically appropriate ways to estimate causality.

I didn't claim otherwise, but that doesn't help for an extremely underpowered analysis which not only didn't even consider causality but didn't consider alternative hypothesis. (In particular, it didn't even consider the null hypothesis except perhaps in some p-hack sense that they may not have published at all if the none of their coefficients had significance according to R's GLM or whatever package they used). That wasn't it's goal, I'm not even accusing the authors of bad science (at a minimum it passed the bar to get published)-- but the conclusions the media were drawing from it couldn't be supported by the work. It's an easy error to make because there is a gap between what we want to know and what we have the data to tell us.

> The question of human climate change's impact on things like wildfires is an interesting, important, and very difficult one

Perhaps it is difficult to prove scientifically, but as humans we can operate on common sense if we choose to, and it's common sense to anyone that has interacted with wood and fires a few times that making a forest drier due to warmer temperatures will increase the flammability of it. I personally don't need scientific proof that climate change makes forest fires more likely. It's f***ing obvious.

If you're going to policy reason on gut, that's fine-- most policy is set that way. But we shouldn't disguise gut as "science" and make it unquestionable or just offensive to debate. (Or instantly resorting to personal attacks and accusations of ignorance.)

Your gut analysis isn't as obvious to me: Fire can't happen without fuel, drier forests grow less fast. In climates where wildfires are a regular and natural feature, all that grows will eventually burn-- and it's literally impossible for it to burn unless it grew first. On that reasoning one might expect that drier climates might lead to fewer wildfires. At least at the reasoning level of "gut". Not too many wildfires in death valley after all, even though it is very dry and hot there while there are wildfires in Seattle which is usually wetter than a sponge. :)

Moreover magnitudes matter a lot-- I'll accept your 'obvious conclusion' but it doesn't set a magnitude. If wildfire activity is modulated only a small amount by dryness but a large amount by where we choose to build, our irrigation practices (e.g. creating pockets of growth in naturally dry places and diverting water from wildlands we don't care about), our forestry practices (aggressively suppressing small inconsequential fires, until they run out of control), and so on-- then that calls for different interventions (and would also be really good news because if existent climate change really were the only cause of recent activity we're screwed because there is no plan to turn back the clock on that, only plans to decrease the rate).

There is also a definitional question: What is a wildfire? The things most people care about under that banner aren't just the metaphorical tree burning in a forest with no one there to hear it... they are incidents of out of control vegetation burning leading to mass property damage and loss of life. And by that definition of a wild fire where we build and how is obviously a determinative factor over which we have a lot of influence. And so a premature fixation over factors which we have less ability to influence and may have a much less significant effect in what we actually care about could result in poor decisions that ultimately lead to a lot more damage and loss of life, which we can all agree would be terrible.