PS has anyone cracked the code on what the computer offers in "Deal or No Deal?" - I'm guessing the average of the remainder less say 10 or 15% "casino" margin (to encourage you to keep gambling)
I wondered that too, and spent a while trying to work it out. It turned out that there is no exact formula they use, and a lot of human input is used. They often put jokes in (like when the person says at the start that they want to win $20,000, a $19,999 offer is likely to come up, or if they say they want to buy a Mazda 323 they'll get an offer of $323 etc). Generally the offer is approximately the average of the remaining figures minus 10-30% but sometimes the offer will be drastically lower than the average, and occasionally it will even be higher. The lower offers tend to come either after a very high case is knocked out (probably for dramatic effect), or when the median value of the cases is much lower than the mean (because this is a much riskier position for the player and they will be more likely to want to take the safe option).
I suspect they have a complex formula which takes the mean and median values into account and puts a low weight on upper outlying values. This figure would be immediately displayed to the 'banker' (probably also with the median, mean and other values) and he can quickly enter a figure based on it (using any humourous figure within the general range of the calculated suggestion). I suspect they may have multiple formulae which are all displayed to the 'banker' at once, some being used at different times (eg, there is a recommended one for when the player seems confident, happy, worried, etc) and also the formulae most frequently used probably change between the beginning and end of the game.
Yes, I am a nerd.
Incidentally, I have been on the show twice (once our block was not called up all day, the other I was one of the idiots on the podium), and I have won a grand total of nothing! Two days wasted! I have watched a total of seven games played (I didn't stick around for the last two games on the second day), including the one I was a minor contestent on. They invited me back, but I decided enough was enough, and third time lucky wasn't enticing enough.
As for the random number guessing game, I think there might be a small flaw, and the optimal strategy might include one more variable (clearly making it slightly better, although calculating how much that extremely slight edge would be might be nasty), but I will need more time to go through the thread to see if it has already been brought up. It would mean sometimes making your guess earlier and sometimes not taking a maximum value to date after the 370th value, and waiting for another. I am quite certain that the best strategy is not to rigidly put you name on the highest value to date after the 370th no matter what.
After looking at the solution, I see I was on the right track, but I would have had a lot of trouble working out exactly where the point to start waiting for a higher figure was. I probably would have guessed closer to 500, and I think my gut feeling was a little above 500.