scott wrote:
>> For example, just the other day I was on the London Underground, and I 
>> wondered: How the **** do they get the trains onto the track? 
>> Something like the Circle line is completely subterranian, and I'm 
>> pretty sure the factory where they make the carrages isn't. And it's 
>> not like you can just take a train carrage down the escalator with you...
> 
> The Circle line will have connections to other tracks that come up above 
> ground.  There are lots of connections that are not marked on the public 
> underground map.  They are needed for things like train repairs and 
> replacements, and if some lines have fewer trains during busy periods, 
> they need somewhere to park the other trains!

Yeah, I guess so...

>> This analysis works only if events are statistically *independent*, of 
>> course. In the case of a flood, they might not be.
> 
> Yep, floods are influenced by things like sun-spots, which follow patterns.

I was thinking more along the lines that a flood might well change the 
characteristics of the flooded land / water courses through it, and 
these changes may well affect the probability of another flood happening 
(although I couldn't say in which direction).

>> In the second analysis, if a dice is rolled and almost never turns up 
>> a 6, this could be taken as evidence that the dice is biased, in which 
>> case the probability is not actually 1/6.
> 
> Also the probability of a flood likely changes dramatically over time, 
> before you get a chance to calculate a good estimate of the probability.

Probability is of course relative too.

The probability that the next letter in a sentence is a U is about 2.7%. 
However, the probability that the next letter is a U *given that* the 
previous letter was Q is more like 90%.

You might argue that that's because these events aren't independent. But 
consider that the chance of rolling two 6's is 1/6^2, but the chance of 
rolling two 6's *given that* you've already rolled one is 1/6 (i.e., 6 
times more probable).

Probability: It depends on what you calculate.

Post a reply to this message