Zoue wrote:
You predicted they would collide in Mexico and were correct because they did. But you have a 50% chance of being right there, so I don't think you can use that as evidence of any science behind your calculations.
I don't believe you can use past contact as evidence of future conduct, since every race (and corner) is different. There are too many variables, from car positions relative to each other, other players in the mix, distance to 1st corner, angle of corner etc. I don't agree that two drivers making contact in e.g. Canada increases the probability that they will do so again in e.g. Brazil.
This is not how statistics work.
The available outcomes are;
1) collision
2) no collision
There is not a 50% chance of each outcome (being right)
The available outcomes of winning the lottery are;
1) win jackpot
2) don't win jackpot
There is not a 50% chance of each outcome
Furthermore, I did not predict the collision in Mexico. I attributed a probability to an outcome. I stated there was a 30-40% chance of a colliision in Mexico. So I was saying it was more likely they wouldn't collide.
If a weatherman says there is a 10% chance it will rain and it does rain. He did not predict rain, he was actually saying there was a 90% chance it wouldn't rain and it did. He got it completely wrong.
The second part of your comment I agree with, you can not predict. That is why you use probability ranges. A prediction would be "they will crash", a probability range is completely different. We know for certain that percentage lies somewhere between 1 and 99%. Then you can start to examine the factors to attribute a probability.
This is how all probability works and whilst you can always say "you never know what would happen in each individual case" that doesn't render probability meaningless. Probabilities are used to decide nearly all major government and individual decisions.