Hacker News new | past | comments | ask | show | jobs | submit login

"All models are wrong, but some are useful"

https://en.wikipedia.org/wiki/All_models_are_wrong




Apart from climate models of course. Science is settled.


That's why researchers should always report uncertainty bounds. On that note, I don't think I've ever heard a researcher give a definitive answer as to what climate change _will_ do. We can very precisely measure the change in ocean surface temperature and acidity and come up with reasonable guesses for the amount of CO2 added to the atmosphere every year, but the future always involves additional uncertainty.


This simply isn’t true. Climate science has an incredibly long history of failed predictions, with the less specific predictions being obviously more resilient to falsification. If you care about the existence of climate change deniers, then you should know who Competitive Enterprise Institute are, most writing on climate change denial can be traced back to their publications. The reason being that they have an essentially limitless supply of alarmist predictions that have been falsified.

https://cei.org/blog/wrong-again-50-years-failed-eco-pocalyp...

https://cei.org/blog/manufacturing-alarm-dana-nuccitellis-cr...

https://wattsupwiththat.com/2017/10/30/some-failed-climate-p...

Just look at all the other child comments, most of them immediately jumped to the defence of supposedly irrefutable climate models. The reason climate deniers have so much support is that climate scientists, their communicators and their advocates have little credibility outside the true believers. If you don’t want to take my word on it, here’s a Stanford study on this exact topic:

https://www.sciencedaily.com/releases/2019/10/191018112145.h...


Climate models are a perfect example of the above statement. They're not good enough to predict the detailed future with certainty (weather forecasts only go out for a few days), but they're useful and valid for reasoning about large scale trends.


I wholeheartedly believe in anthropogenic climate change and I disagree with this comment with every fiber of my being despite not knowing which "side" you're actually on.

I'm actually kind of impressed.


I’m on neither side. I also wholeheartedly believe in anthropogenic climate change, but I think the commentary on climate science is about as ignorant and anti-scientific as the views espoused by young earth creationists. Climate models are flimsy as hell and have an atrocious track record of making predictions. Yet “the science is settled” (or a close approximation to it) is the only acceptable view in most circles. This gives climate change deniers a remarkably strong platform, because they can just say “climate scientists (or more often climate science communicators) are full of bs” and then point to countless examples where that is absolutely true. Science is never settled, in any case that you think it is, you’re not dealing with science, you’re dealing with an opinionated dogma. A statement which almost everybody would agree with, on any topic except climate change.


Computer models are used to try to predict the local effects of climate change, not prove whether it’s real. The basic science is settled, and no one needed computers to do it. Heck the hypothesis was first proposed over 100 years ago.

Computer climate models have trouble returning accurate predictions for exactly the same reasons it’s hard to make accurate weather forecasts. They’re attempting to do things on the edge what we know is possible. Personally, I would not call that “flimsy as hell.”


Relativity is a model. Yet it is settled.

Still 'wrong' though.


Seems like a pretty creative use of the word ‘settled’.


F = M * A is never correct. Yet, it’s simple and useful enough in most situations to be ‘settled’ and people will probably still use it in 10,000 years. Thus, demonstrating these are different qualities when it comes to models.


The key feature of science is that all theories are open to further scrutiny and falsification. To say “the science is settled” is probably the least scientific statement you could possibly make. A deeply ironic position to take for people who’s key complaint about their detractors is that they don’t appreciate science.


When comparing two alternatives, we can most definitely conclude that the science is settled, at least until actual evidence to the contrary shows up. Being open to accepting new evidence in the future in no way means we can't act confidently on what we know now, taking into account our state of knowledge/uncertainty.

The science on a heliocentric solar system is settled. The science establishing a link between anthropogenic carbon emissions and average global temperatures is also settled - built on foundational chemistry and physics, supplemented with observational evidence here on Earth, and validated by the planetary science done to understand Venus's temperature.

Sure, there's a lot of remaining uncertainty about short term effects, but mostly in the category of "just how bad is it likely to get?"


Science as a search for truth cares deeply about accuracy, engineering doesn’t. Discovering say F=MA was wrong was a huge deal, but existing results still apply. New models fill in gaps, but the improved models needed to account for everything the old one correctly predicted.

So while science moves on, engineering does not care about the 15th decimal place. Which is why we keep getting value from approximations.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: