I don’t like when people read a book and then become experts on a subject because they read one book on a topic. For instance, I didn’t fancy myself as knowledgeable on economics because I read one book on the subject. I fancied myself smart enough at it after I read about three books. So there’s my threshold. Why am I telling you this?
At Virginia Tech, George Mason, and now at Colorado Tech I’ve taken statistical modeling courses. A model is something you generate based on old data. Based on this old data, it can make predictions on what will happen with similar data in the future. Generally you can use these for system failure rates – like if you’ve got a system that’s been running for some period of time, and you know how long it has been between each failure, you can use that date to predict when another likely failure can occur.
Another useful model could be any grocery store logistics chain. They get their food from somewhere, in a timely manner. Using sales data they can predict when certain things will be in a higher demand. For instance, on the 4th of July, I bet you couldn’t avoid hot dogs, sodas, potato chips, and ketchup in any grocery store you went to. The statistical data they keep using from those club membership cards allow them to model their stocks to be stuff they know that the majority of their customers will buy. Pay attention to when turkies start showing up – they usually show up about three weeks before Thanksgiving, because statistical data has shown that a significant portion of shoppers will buy them that early. There isn’t as much clam juice on the shelves ’cause people don’t buy it. It’s not because the store doesn’t like clam juice!
Same for Halloween candy, and other seasonal items. They don’t come out with Valentine’s day stuff right after Christmas stuff just because they’re trying to make money – statistical data has shown that, within a degree of certainty, people are buying all that stuff that early!
Models depend on the data and depend on limiting variables to a reasonable amount. The more variables in a model, the more likely it is to be incorrect. If my system that I’m trying to predict has 100 points of failure, it gets a lot harder to predict which point could fail. 1000 points of failure? You’ll need a good computer. 10000? You’ll need a super computer.
Weather models are a whole different animal. If you put a sensor at every corner of every square mile of the earth, you’ve limited your variable set to… what’s your best guess as to how many particles of air are in a square mile? How many of those square miles are over oceans or ocean currents? How much wind does it take to move one air particle, and when wind moves an air particle, how does it collide with other air particles while it’s moving, and how do the particles that are being collided into get moved by the wind, in a 360 degree space? Imagine the variable space of one square mile of the earth’s surface! I’d say there are at least thirty.
Worldwide, I think there are twelve models that are scientifically accepted as being able to predict weather patterns. On average, they are tolerably accurate. This was information passed on to me by my professor at CTU (my professor was not Jack Bauer) for one of our exercises, and I can’t find the reference to the article we used, and I probably won’t try again. One of the modeling systems came up with a result that shows the global temperature going up alarmingly. This is the one being bandied about in An Inconvienient Truth and used by the UN in their climate panel. They picked the worst set of data from this one particular model – and compared to the other modeling systems, this one is like the Jerry Falwell (bless his soul) of climate models – it gets way spun up but nobody knew why.
Turns out they used bad data to feed their model. Or they misinterpreted the data – one of the two. This link explains how the data got warped, and why the graph has gone from “oh shit” to “oops”. The author of that blog post has more time to go and find/check references than I do, and he’s a much better writer – it’s an interesting read if you’re bored at work. If you’re not, go look at some furry pictures or something more entertaining.
Most other models are much more normalized. And even then – these models don’t have the benefit of sensors at every square mile of the earth’s surface. So the fact that they’re even 75% accurate could be because of blind luck – real science is just that picky. The way to learn about this phenomenon of “climate change” is to look at the numbers yourself, and not just regurgitate whatever the people you already agree with tell you.
Then, read this, ’cause it’s funny. And then take solace in the fact that the planet will be fine. We may all be dead! But the planet, it’ll be fine. This is also funny, and so is this. Warning – the link has the word conservative in it, so if that doesn’t pass your confirmation bias filter, you might not want to click on it. That means you’ll automatically get angry or disagree with whatever is said in there!
