The anecdote isn't as farfetched as you might think. (IAAS -- I am a statistician -- so I can speak from experience here.) Physical systems, like casino wheels, are terribly hard to predict. But there are ways...
I know of one story where an wily engineer exploted an imbalance in a casino roulette wheel. Basically, an error in the alignment meant that one side of the casino wheel came up more often than the other. It still wasn't possible to predict any one number coming up, but you could say, *statistically*, that certain numbers came up more often than the odds paid for betting on them. If you make enough bets, then after a while the difference will result in profit. Casinos actually *encourage* the punters to write down the past numbers, but without some heavy-duty analytic software to back it up, this won't help anyone -- and you can't bring your PC into the casino, that's for sure. This engineer has some kind of transmission system built into his shoe, as I recall. But roulette wheels are regularly checked for balance for exactly this reason, so don't try this at home, kids.
When RNGs are used instead of physical systems, there are other ways to exploit the system. I remember one time when the government of my home state in Australia introduced an early form of on-line gambling: a Keno system (like lotto -- draw 10 numbered balls from 100, and try and guess all 10) with drawings every five minutes. It was all computer-controlled, with displays in pubs and bars around the state. You could place bets at an online terminal at the bar. The government got a statistical consulting business I was involved with to help set the odds and work out what the expected payoff would be (so they'd have enough mony to cover winnings in the early days -- the jackpot was a significant amount -- early payoffs were to be backed by insurance).
Unfortunately, the jackpot was won on the first game of day three, which caused some embarassment all round. (They'd predicted a miniscule chance of it being won within a month -- there was a chance of this, of course, but the insurance premium paid didn't cover this eventuality.) A bit of investigation revealed what had happened though. On the first drawing of day 1, the numbers were (say) 16 19 22 27 30 49 58 65 74 97. On the first drawing of day 2, the numbers were again, 16 19 22 27 30 49 58 65 74 97. Someone noticed this and placed a bet for the same numbers for the first game of day 3, and won the jackpot. As it turns out, the programmer had reset the "FOR day IN..." loop, meaning the first results of each day were the same. (The RNG was used between games in the ticket issuing process, so it didn't affect later days, which was why it wasn't noticed in development.) Just goes to show the importance of good QA when dealing with RNGs.
When I buy anything -- not just software -- the fewer options I have, the better. I don't have the time to sit down and try and decide what works best for me, or to experiment with the various options in the interface over a period of months until I find something that works. What I expect is that the manufacturer has INVESTED THE EFFORT to learn about what is best for me, implement it, and market it to me.
It's like, when you go to a nasty restaurant and ask for a salad, you get a million options. What kind of salad? What kind of dressing (from this choice of sixteen) -- and will that be low-fat and on the side? Croutons? Bacon bits?
But when I go to a GOOD restaurant and ask for a salad, they just bring me a GOOD SALAD, no questions asked. They've put the work into developing a salad that I'll like, and that's why I've chosen that restaurant. Would that software were the same way...
At that speed it would have taken ~150,000 years to get here. Since that is assuming a straight line...
150000 years is but a twinkle in the eye in the scale of the age of the Universe, or even the age of the Earth. And a straight line is the only option (at least in the sense of General Relativity). The likelihood of an extrasolar meteorite dropping on earth is hard to judge (mainly due to the observer bias inherent in the anthropic principle) but it doesn't seem that farfetched to me.
Slashdot Mirror
I know of one story where an wily engineer exploted an imbalance in a casino roulette wheel. Basically, an error in the alignment meant that one side of the casino wheel came up more often than the other. It still wasn't possible to predict any one number coming up, but you could say, *statistically*, that certain numbers came up more often than the odds paid for betting on them. If you make enough bets, then after a while the difference will result in profit. Casinos actually *encourage* the punters to write down the past numbers, but without some heavy-duty analytic software to back it up, this won't help anyone -- and you can't bring your PC into the casino, that's for sure. This engineer has some kind of transmission system built into his shoe, as I recall. But roulette wheels are regularly checked for balance for exactly this reason, so don't try this at home, kids.
When RNGs are used instead of physical systems, there are other ways to exploit the system. I remember one time when the government of my home state in Australia introduced an early form of on-line gambling: a Keno system (like lotto -- draw 10 numbered balls from 100, and try and guess all 10) with drawings every five minutes. It was all computer-controlled, with displays in pubs and bars around the state. You could place bets at an online terminal at the bar. The government got a statistical consulting business I was involved with to help set the odds and work out what the expected payoff would be (so they'd have enough mony to cover winnings in the early days -- the jackpot was a significant amount -- early payoffs were to be backed by insurance).
Unfortunately, the jackpot was won on the first game of day three, which caused some embarassment all round. (They'd predicted a miniscule chance of it being won within a month -- there was a chance of this, of course, but the insurance premium paid didn't cover this eventuality.) A bit of investigation revealed what had happened though. On the first drawing of day 1, the numbers were (say) 16 19 22 27 30 49 58 65 74 97. On the first drawing of day 2, the numbers were again, 16 19 22 27 30 49 58 65 74 97. Someone noticed this and placed a bet for the same numbers for the first game of day 3, and won the jackpot. As it turns out, the programmer had reset the "FOR day IN ..." loop, meaning the first results of each day were the same. (The RNG was used between games in the ticket issuing process, so it didn't affect later days, which was why it wasn't noticed in development.) Just goes to show the importance of good QA when dealing with RNGs.
It's like, when you go to a nasty restaurant and ask for a salad, you get a million options. What kind of salad? What kind of dressing (from this choice of sixteen) -- and will that be low-fat and on the side? Croutons? Bacon bits?
But when I go to a GOOD restaurant and ask for a salad, they just bring me a GOOD SALAD, no questions asked. They've put the work into developing a salad that I'll like, and that's why I've chosen that restaurant. Would that software were the same way...
150000 years is but a twinkle in the eye in the scale of the age of the Universe, or even the age of the Earth. And a straight line is the only option (at least in the sense of General Relativity). The likelihood of an extrasolar meteorite dropping on earth is hard to judge (mainly due to the observer bias inherent in the anthropic principle) but it doesn't seem that farfetched to me.
Then don't buy Warcraft III. No-one's forcing you to play on battle.net. Or just enjoy the single-player campaign.
Data mining is statistical analysis of structured or unstructured data to discover unknown relationships.
At best, this is voyeurism. At worst, it's espionage.