This crafter knows that real random systems are extremely streaky, much more than most realize. Sure, the computer's system is not a true random number generator, yet when I take the time to collect sufficient detailed data, I've never yet seen this one fail the standard statistical tests of goodness of fit. Compound that with the very low intrinsic event rate, and observer confirmation bias and you get these threads.
Originally Posted by MithrielWielder
Success rate = 0.7, the expected probability of 5 consecutive successes = 16.8% at the limit. Short sampling may see a very large difference in the recorded result and the central limit theorem expected results. For 6 consecutive successes = 11.7% at the limit. For 5 consecutive failures, you get 0.2%, which if it happens to you makes it memorable (and this contributes to observer confirmation bias). For 6 consecutive failures, you get 0.07% and that is very memorable if it happens to you. What you forget is all of the 'normal' results and you are completely oblivious to the results of all of the other players on your server at the time.
Last edited by cdq1958; Dec 07 2013 at 09:10 PM.
"No sadder words of tongue or pen are the words: 'Might have been'." -- John Greenleaf Whittier
"Do or do not. There is no try." -- Yoda
Indeed, in a world and life full of change, the only constant is human nature (A is A, after all :P).
We old vets need to keep in mind those who come after us.