Skip to the content ForoHackers
Log inJoin

I promised to pick a fair music track with the coin that lives on my desk. The coin leans.

Probability and statisticsTesting
#15367 · · Quote · Report

My desk coin has a blob of solder stuck to its face. We tried it a bit, it isn't obviously fifty-fifty, and my bandmates want the next listening choice settled before we all die of politeness. Can we use this same shitty coin without having to learn its exact bias? Assume successive tosses are independent and the bias stays put. Yes, yes, that's a big assumption for a desk coin.

#15620 · · Quote · Report

Two tosses. Call HT record A, TH record B. HH and TT get another pair. With a constant head probability p, the two accepted patterns both have probability p times (1-p). Unless the coin always gives the same side, you eventually get a choice with probability one.

#15621 · · Quote · Report

How eventually? Suppose my putative coin has p = 0.9. I'm going to be seventy years old before he gets to the guitar solo.

#15622 · · Quote · Report

I make it 0.18 acceptance per pair. A little over five and a half pairs on average. And yes, the first sixteen pairs could still all suck. I have already got the opening track stuck in my head though.

#15623 · · Quote · Report

Don't quietly change how you throw it on the second toss to shake loose a result. Then the two products need not be the same.

#15624 · · Quote · Report

lol the drummer threw one into the carpet and balanced the next on his elbow before launch. we have agreed to use the method for the maths and let him put his record on for the music.

You are not logged in, so this goes up anonymously: no account, no name and no address is stored with it, and it may carry no link and no image. Log in or join to post under your name. Our web server's logs keep the address and time of every request for up to 14 days; to keep yours out of them, use Tor. To reach the staff in confidence, use the drop. Sign as a returning source (a key kept only in this browser)

Sending takes a few seconds here: your browser first works out a proof of work against spam.

Reply