Hacker News Reader

Comment by bsbechtel | Hacker News Reader

Why do we believe climate scientists when they can't even predict the weather 10 days out? Yes, weather doesn't equal climate, but climate is weather in aggregate. That means we should be able to predict short-term, local weather events more accurately before we even begin to start predicting long-term, global weather phenomenons. If our short-term weather models are so pitifully inaccurate, what on earth makes us believe our climate models predicting what will happen 5, 10, 20, 50 years from now are anywhere close to being correct?

View on HN

In many systems you can predict larger scale events (when the pot of water will boil) despite being unable to predict smaller scale ones (turbulent flow inside the pot).

Your pot of boiling water prediction is based on past experience, not modeling of the smaller scale system. Climate science is akin to modeling turbulent flow inside the pot to predict when the pot will boil.

All modelling is based on past experience. The usefulness of a model is the degree to which you can abstract it from that experience.

In the case of the pot of water, you can observe the energy input into the system and combine that with some very generic, simple, and reliable physical models to determine when it will boil.

I'm not claiming that climate science is that simple, of course. My point is that some systems are more easily predicted at a larger scale than a smaller one.

To take a similar example, you can derive the ideal gas law from the kinetic theory of gasses. Which is to say, you can start from the idea that gasses are composed of small particles that move constantly and collide with each other and with their container, and turn that into an accurate description of macroscopic behavior. This works well despite the fact that it remains impossible to predict the motion of an individual gas particle for more than an extremely short distance or time.

You're comparing apples to oranges here.

Predicting the exact type of precipitation in a specific locale is one thing. Extrapolating a global trend based on past data is another.

They don't need to know whether it will be raining next month in Istanbul to tell you that the world is warming at a specific rate and what are the likely culprits.

I predict the average value of fair dice thrown into the future will be 3.5.

I predict the next value thrown by a given fair die is _____.

Sometimes aggregate predictions are easier than fine-grained ones.

Reply on news.ycombinator.com