The entire universe is not deterministic. It's stochastic, i.e. driven by probabilities, even down to an atomic level. There's a probability you will be killed by a car collision today, either as a pedestrian or in your vehicle, or even in your home when a Hummer runs amok and comes barreling through. There's also a probability a piece of a commercial airliner will fall off and hit you tomorrow. However minuscule it might be, it's not zero, unless it falls off and hits you today, resulting in your demise. There's even a probability you'll be the victim of a home invasion robbery some time tonight, and another one, ever so slightly higher of being the victim of one some time during the next month as it's a larger time window.
All of your insurance premiums are driven by probabilities, and it's the same as placing a wager, whether it's life, accident, homeowner, auto, etc. You're betting bad stuff will happen sooner and more often. The insurance company is betting it won't happen as soon, and will occur less often. Think about it. It's also why a National Weather Service forecast is worthless much farther out than the next 72 hours. They make them for 10 days because the American public won't accept the reality that day 4-10 aren't worth the paper they're not printed on. They demand to know if it will rain on their picnic a week and a half from now and will not accept "we don't know" for an answer. After all, if we could send a man to the moon, why can't we predict the weather for a given city two weeks from today? Consulting the Farmer's Almanac that was published some time last year for this calendar year would be just as reliable.
I've dealt professionally with probability events and predictions by examining the relevant processes, and the probabilities of various outcomes from those processes. One can make predictions, but they only have a confidence of something less than 100%, sometimes much less. The point I got from this is that you really cannot measure "global warming" effects, if there are any, except over many decades. Everything else is statistical variation about a mean. For those that didn't recognize it, the probability distribution he showed is a Poisson, it's not a Normal Bell Curve. It's an excellent example of a Poisson, and exactly what I'd expect to see when plotting numbers of hurricanes in North America per year over 50-100 years or so. What we're seeing this year is completely within the expected variation for number of hurricanes per year for the last 50 to 100 years.
There are three kinds of models:
- Descriptive - how does a system currently behave
- Predictive - how will a system behave in the future
- Prescriptive - what can be done to change current system behavior to a more desirable one in the future
He's showing a very basic descriptive model of the probability of "x" number of hurricanes in North America for a given year with a simple and mostly non-quantitative explanation.