You'll never get rich.

June 6, 2003 9:33 AM

I guess we all suspected it, but those one-armed bandits were fixed all along.The fruit machines we gamble on in our local pubs follow pre-determined sequences to get you hooked; there is nothing random in their behaviour. How do we know? This U.K. campaign group has built PC emulators to run the actual programme code from the gaming machines's ROM chips. They claim they are not only fixed but illegal (via NTK).
posted by rolo (53 comments total)

At my neighborhood bar, there is an Albino guy who's addicted to these things. On any given night (and I do mean any), he makes the rounds of all the other nearby taverns to see which machines are "hot". I estimate he has put about 7 billion dollars into these fruit machines. There also seems to be some kind of under the table pay-outs being made, as no one will give me a straight answer as to why one would play a game such as this "for fun" (payouts on these are illegal here in Wisconsin). The funny thing is, the albino guy thinks he's got the system pegged-- he plays with a confidence and dedication that is something to behold.

Like the Lottery, these games seem to be a tax for stupidity...
posted by sharksandwich at 9:50 AM on June 6, 2003

Interesting, though not totally unsurprising. Still, you know what they say about a fool and his money. If you "gamble" on a fruit machine you get what you deserve.
posted by salmacis at 9:51 AM on June 6, 2003

I love having my buttons pushed so easily. I have no idea how I'll live to the age of 50 without a heart attack killing me.

Not to derail this thread with the first comment, but this is exactly why electronic voting is so troubling.
posted by PigAlien at 9:53 AM on June 6, 2003

Although, on second thought, I like the idea of people 'auditing' electronic voting booths with emulators...
posted by PigAlien at 9:54 AM on June 6, 2003

When I worked in probation in Oregon, I saw as many lives ruined by video gambling -- which is run by the state lottery -- as by alcohol, drugs, or violence. Of course the games are fixed, but they don't even need to. If someone shovels their life into a machine dollar by dollar, they need to be fixed, not some code.

Like the Lottery, these games seem to be a tax for stupidity...

Or a tax for the poor.
posted by Ignatius J. Reilly at 9:57 AM on June 6, 2003

This is really fascinating and a great testimony to the value of emulators. I don't know anything about gambling laws in the U.K., but I'm very interested nonetheless to see how this plays out.

I agree that lotteries and some other types of gambling tax the stupid, but that doesn't stop me from buying a ticket from time to time. I understand how that it's almost incomprehensibly unlikely that I'll win, but it's also quite comprehensibly impossible that I'll win if I never buy a ticket. I do pity folks who believe in systems, laws of averages, or merit-based luck, and I see them all the time choosing their games at the lottery counter.

Hey, aren't cigarette and alcohol taxes also taxes on stupidity?

Also, PigAlien, electronic voting is rife with possibilities for this sort of thing, and that's why its inevitable realization will have to have a very smart and open design in order to have any chance of fairness. It will need well chosen cryptographic protocol design as well as an audible paper trail. Our voting system has a lot of room for improvement, and I hope that intelligence dominates over lobbying and incumbency when the big choices are made.
posted by Songdog at 10:07 AM on June 6, 2003

I agree that lotteries and some other types of gambling tax the stupid, but that doesn't stop me from buying a ticket from time to time.

Uh-oh. Please try to resist the temptation.
posted by Songdog at 10:08 AM on June 6, 2003

I have a few friends who work for GTech, one of the worlds largest lottery supply and slot machine management companies. I don't know about British law, but in the US, any lottery machine or slot machine is required to have a random number generator chip, and goes though extensive testing by your local state lottery commission, just to be sure.
posted by SweetJesus at 10:18 AM on June 6, 2003

Oh, and if you're smart, you check the machine's pay schedule, and get a feel for the pay cycle. Slot machines are required by law to pay out a percentage of what they take in, in x amount of time.

Usually the software will payout in a few big bursts, so if a machine just paid off big, stay the hell away from it.

And if you're like me, you just play blackjack, and stay away from the slots all together.
posted by SweetJesus at 10:26 AM on June 6, 2003

I don't see how what is described here is "fraud". It sounds like you put your coin in, and at that point it is pre-determined by the machine whether you are a big winner or not - no matter what buttons you push later on. In that respect, this is no different than buying a scratch ticket - once you do, it is pre-determined what, if anything, you win, regardless of the order you scratch all the dots off.

The ariticle seems to imply that the "gamble" should be in playing the odds between betting "high" or "low" when you are offered a choice, and that the machine's predetermination of the outcome of your choice constitutes fraud. I disagree; the gamble is when you put your coin in the machine in the first place. At that point you're either going to win or lose, all the eye candy in between is just for "entertainment".
posted by yhbc at 10:27 AM on June 6, 2003

Cigarette, alcohol and lottery taxes are taxes on poor people, not stupid people. Otherwise you would have rich idiots playing the lottery. I’m working the graveyard shift at a convenience store right now, and the predominant buyers of cigarettes and lottery tickets are poor people, especially those on disability and welfare (I’ve seen a guy spend a hundred and fifty-dollars of a disability cheque in thirty-minutes on Nevada tickets).
posted by Pseudoephedrine at 10:32 AM on June 6, 2003

More information on Nevada's "Regulation 14" which describes, among other things, "Minimum standards for gaming devices" here. For the most part, the minimum payback of all wagers is 75% (there's a reason one bank of machines will have hundreds of grannies circling them like sharks and another bank is growing cobwebs -- the pay-off percentage in latter case is much lower). And I don't believe there are many (if any) machines in the US that allow you to make a choice to go "up or down" after barfing up two reels except I seem to have some hazy memory of slot machines emulating a certain card game with a similar choice available.
posted by user92371 at 10:38 AM on June 6, 2003

Seems to me that if you play slots, you're going to lose in the long run anyway - else why would slot owners bother putting a machine in? Maybe it's 'fraudulent' in the sense that it's using psychology to get you to play more - 'Oh I could have won last time if only ...' but I think that's stretching it. Before all this programming, people were *still* losing at slots, and slot owners were still making money - weren't they?
posted by carter at 10:45 AM on June 6, 2003

I agree that lotteries and some other types of gambling tax the stupid, but that doesn't stop me from buying a ticket from time to time.

Uh-oh. Please try to resist the temptation.

I have a close personal friend who won the big one (UK National Lottery - £3.3M). Uh, wait. That in no way invalidates Songdog's advice. Don't buy em.
posted by Dan Brilliant at 10:51 AM on June 6, 2003

The article says that a machine that claimed to pay out 90% could decide not to pay out any money at all until a million years in the future, and that would be perfectly legal. That's clearly fraudulent. When you put your money in, you should have the same chance of randomly winning as anyone else, but currently it would be theoretically legal to have nobody in the lifetime of the machine win.
posted by Pretty_Generic at 10:55 AM on June 6, 2003

Um, I just had a thought... if the interval between people winning is somehow predetermined... and the code can be examined using the emulator, to work out how that interval is determined... couldn't you use tactics to always win?
posted by Pretty_Generic at 10:58 AM on June 6, 2003

Usually the software will payout in a few big bursts, so if a machine just paid off big, stay the hell away from it.

Except of course, that's complete nonsense. Since it's random (at least, it's supposed to be, and I'm fairly confident the ones in most regulated areas of the US are), then it's got exactly the same chance of coming up big on the next spin as it did on the winner. It's just that the chances are so small that it's not likely to happen. (That being said, I've seen it happen.)

Slots tend to be pretty far down on the return rate, and generally aren't all that great. (Although the secret is the higher denomination slots tend to have better pay scales). In some areas you *can* find banks of machines that pay back as much as 98%, but they're not too common.

Table games tend to be much better, especially blackjack, craps, and baccarat. Double 0 roulette is quite possibly worse than many slot machines. Video poker can be one of the better bets in the house a) if you know what you're doing and b) if you can find a decent pay scale. If you can find the right video poker machine, with the right slot club, and play perfectly, it's theoretically possible to actually play with a greater than 100% return. Course, not too many people can play perfectly.
posted by piper28 at 10:58 AM on June 6, 2003

Bill Bennett's going to be furious.
posted by uosuaq at 11:02 AM on June 6, 2003

Oh, they were losing even *more* back then. Unscrupulous owners could "bug" the machine - bugs were sold anywhere from 1900 to 1950, and would prevent the machine from displaying a pre-chosen (most likely) high-paying symbol. Sometimes a bug required police payoffs to work properly. There were other ways to gaff the machine, and owners also had to skirt different sorts of gambling regulations -- for a time, slot machines would pay off in rolls of mints, candy, or gum, and the owner would then buy this booty off you and restock the machine (a practice which continues to this day in Japan, except with little trinkets. And it's pachinko).
posted by user92371 at 11:03 AM on June 6, 2003

SweetJesus, have you ever heard of the gambler's fallacy? You're soaking in it!
posted by argybarg at 11:06 AM on June 6, 2003

Otherwise you would have rich idiots playing the lottery (or slots)

Bill Bennett's going to be furious.
posted by TedW at 11:12 AM on June 6, 2003

I believe that the only gambling options in which skill can provide any edge at all are blackjack, poker, and horse racing. Is this true?
posted by rainbaby at 11:36 AM on June 6, 2003

Random != pseudo-random.

There is no way to generate random numbers by deterministic means.

Philistines.
posted by spazzm at 11:41 AM on June 6, 2003

SweetJesus, have you ever heard of the gambler's fallacy? You're soaking in it!

I don't really gamble all too much, but I have a few friends who fix slot machines here in Newport, RI, where you can gamble.

The Gambler's Fallacy is committed when a person assumes that a departure from what occurs on average or in the long term will be corrected in the short term.

That's not what I'm saying. Gambling is all about playing the odds. If a slot machine hits a large jackpot, odds are it's not going to hit it again any time soon, but it's possible. But you don't bet on the possible, you bet on the probable.

Except of course, that's complete nonsense. Since it's random (at least, it's supposed to be, and I'm fairly confident the ones in most regulated areas of the US are), then it's got exactly the same chance of coming up big on the next spin as it did on the winner. It's just that the chances are so small that it's not likely to happen. (That being said, I've seen it happen.)

It's not complete nonsense. If were talking about modern progressive slots, there is a little chip that is going to determine exactly what you make at time x, so that the machine is in step with payout schedule y.
posted by SweetJesus at 11:43 AM on June 6, 2003

I'm shocked. Shocked.
posted by beagle at 11:56 AM on June 6, 2003

Rainbaby: No, if you define "skill" as understanding some of the underlying mechanics of the game that people usually don't think about -- like the fact that roulette dealers can, with some regularity, make a ball land (or not land) wherever the hell they want it to.
posted by user92371 at 11:59 AM on June 6, 2003

THis fixed-ness seems to make it more likely that you can win - if you can work out the pattern, you're all set. If it's truly random, there's nothing to work out, no strategy that will give you an edge.
posted by mdn at 12:00 PM on June 6, 2003

I believe that the only gambling options in which skill can provide any edge at all are blackjack, poker, and horse racing. Is this true?

Depends on your feelings about sports betting.
posted by COBRA! at 12:01 PM on June 6, 2003

Except of course, that's complete nonsense. Since it's random (at least, it's supposed to be, and I'm fairly confident the ones in most regulated areas of the US are)

The point here being that for these machines, it isn't random, as the emulation shows (and as every fruit machine expert in the UK 'knows'). The machine has pre-selected a maximum payout for this spin, and rigs any further user choices (note that UK slots have a million bells and whistles, they make Vegas slots look incredibly primitive (also note I say 'look', as the clever stuff in both, is all hidden from the mark)).

Quite how legal this all is hangs on the precise nature of the local law. For example, from the article:

It may also fall foul of "heartstopper" regulations, which forbid scratchcard manufacturers, for example, from creating a disproportionate number of cards with "nearly" wins on them (eg two jackpot symbols in a row and then a losing one), which tempt people to play again.

I grew up in a seaside town, where I acquired a minor fruit machine problem. I kicked it a long time ago, but I still don't like hanging around the slots on the odd occasion I'm in Vegas.
posted by inpHilltr8r at 12:09 PM on June 6, 2003

Cigarette, alcohol and lottery taxes are taxes on poor people, not stupid people.

I hate to tell you, pseudo, but this is bullshit. If I offer you a proposition where you give me as many dollars as you want, and I will return 90% of those dollars, and you take it, you are stupid. I don't mean to seem insensitive, but don't you think that the intersection of "poor" and "stupid" might be larger than you're willing to admin?

Of course, I know that there are many fine, intelligent people who are in bad financial shape for various unfortunate reasons. Tell me with a straight face, though, that throwing away a large chunk of a limited income away on addictions and bad math is a smart thing to do.

I’ve seen a guy spend a hundred and fifty-dollars of a disability cheque in thirty-minutes on Nevada tickets

Tell me that this isn't stupid. Maybe there's a reason this guy is poor? This is not a case of "blaming the victim". This is a case of "blaming someone who has no idea how money works".
posted by majcher at 12:14 PM on June 6, 2003

Try this in America and it's likely a DMCA violation.
posted by monkeyman at 12:21 PM on June 6, 2003

SweetJesus:

Well, if you're arguing that the machines spit out wins at regular intervals (with some limited fluctuation), then you're right.

But if the chips are truly random (and whether they are is a separate issue) -- that is, they have a 1/n chance of spitting out a payoff per regular interval, then the machine is just as likely to pay off right after a payoff as it is at any other time.

But you don't bet on the possible, you bet on the probable.
I'm talking about the probable here, and I'm saying the probability doesn't change. I'm sorry to rely on the old coin-flipping basis here, but it's the easiest to work with. In one coin flip, the odds of tails coming up is 50%. The odds of tails coming up in two coin flips is 75%. The odds in three is 87.5%, and so on. As we consider a larger and larger size, the odds of a particular outcome in the aggregate goes up.

But the odds of tails coming up in any given flip is always the same -- 50%, no matter where in the sequence you pick the flip. This is true even if you pay out for tails. So it doesn't matter where in the sequence you choose to bet -- whether tails has just occured, or even a freakishly long sequence of tails, or of heads, or anything else.

Failing to make the clear distinction between the odds of an event over time and the odds in one event is part of the gambler's fallacy.

Again, if the algorithm is rigged, the subject changes. But I don't think that's what you were claiming.
posted by argybarg at 12:28 PM on June 6, 2003

If I offer you a proposition where you give me as many dollars as you want, and I will return 90% of those dollars, and you take it, you are stupid.

I paid $15 to see a band a few nights ago, and not only did they not give me 90% of my money back: they didn't give me anything back. Does that make me extra stupid?

Point being, many people consider gambling a form of entertainment, and most people are willing to pay for entertainment.
posted by mr_roboto at 12:29 PM on June 6, 2003

Of course, I know that there are many fine, intelligent people who are in bad financial shape for various unfortunate reasons. Tell me with a straight face, though, that throwing away a large chunk of a limited income away on addictions and bad math is a smart thing to do.

No one tried to say. Nor did anyone try to say that being desperate makes you hyperrational or some shit. Have you ever been really hungry? Did it reduce your standards as to what you would eat (I'm so hungry I could eat a horse)? Imagine being that hard-up for money, you'll do the financial equivalent of gulping milk without sniffing it first.

Playing the lottery for many destitute people is like saying "the five dollars I have to my name ain't shit, so I can throw it away trying to get rich" -- at first. Regardless of the initial reasoning, or lack thereof, once one is addicted to gambling it has nothing to do with winning, much less making wise fiscal decisions, it is about the thrill of having the money on the line. A serious gambling addict gets the same blank look on their face when the hand is over regardless of whether they won or lost.

That's might be sustainable if you're not already poor. If you are, it can ruin you.
posted by Ignatius J. Reilly at 12:34 PM on June 6, 2003

I know at least two people who suppliment their incomes by hitting slots at the right time. You get to know their patterns and spot when they are getting hot. That's another reason they tend to move the things around a lot.

Then of course I also know one guy who regularally puts his paycheque into slots and walks away with an empty wallet...
posted by twine42 at 12:44 PM on June 6, 2003

There is no way to generate random numbers by deterministic means.

This may be a stupid question, but why don't they use non-deterministic means?
posted by skryche at 12:58 PM on June 6, 2003

Again, if the algorithm is rigged, the subject changes. But I don't think that's what you were claiming.

I'm not talking about pure games of chance, here. I'm talking about computer controlled video gaming machines. These machines, because of the current limits of computer science, are not truly random.

Most randomization algorithms involve picking a seed number, and performing some calculations on it to generate a random number. Now, I don't profess to know exactly how complex their randomization functions are, but the seed number is always pre-determined (the date, the time, something like that), which means even if you toss in a million other variables into the calculation, the number will never truly be random.

The programmers who write the code for these things are pretty smart. I have no doubt in my mind that the algorithms are designed to make the machines work on the house's terms, as much as legally possible.
posted by SweetJesus at 1:02 PM on June 6, 2003

An easy way to make deterministic (pseudo-random) algorithms less deterministic in a game like a slot machine is simply to continually generate numbers even when none are needed, i.e. when the machine is waiting. Several mililon can be generated in the time it takes the player to react to the last spin. Even if they tried, players couldn't use this to influence play, as their resolution simply isn't high enough to pull the lever just when the right number comes up.

The old Apple II used a similar variant of this, incrementing two bytes in memory whenever the machine was waiting for a keypress. This number could then be used as a seed for a more standard PNG since obviously, you can't expect the human to hit a key every time a new random number is needed.

On a game I once wrote for the Apple, I just scanned through the machine's ROM to generate random numbers. There were some patterns in the resulting data, so I kept a counter of how many animation cycles had gone by since the player pressed a control key, and each time the player did press a key, I skipped that many numbers from the ROM. That made it seem significantly more difficult to predict.
posted by kindall at 1:36 PM on June 6, 2003

and each time the player did press a key, I skipped that many numbers from the ROM

The problem with that strategy is, there are certain numbers that are probably not in the ROM at all, as they are not valid 6502 instructions. Mind you, you might not have had time to perform any complex mathematical juju and still keep the rest of the game running.
posted by arto at 2:00 PM on June 6, 2003

One way to define random numbers is to say that the sequence of values cannot be described in a more compact way. This is equivalent to saying that the list of values doesn't follow any kind rule that lets you predict what the next one is.

That means that you can't generate these things using a computer program, because a computer program would be a counter example - a set of rules that does let you predict the next value.

However, you can pretty close. It is possible to generate sequences of numbers using computer programs that are "very like" "truly random" sequences.

This "very like" means that, amongst other properties, two adjacent values are not related in a measureable way. So if a sequence like this directly controls the pay-out of a machine then you can't say anything about the likelihood of a big pay-out after one has just occured.

However, there are two qualifications to make to that, both related to "state" - values that are remembered over time.

First, the game (or the underlying implementation) may include additional state. An obvious example is a "pot" that slowly increases as time passes without a win. Winning the "pot" will return more money after a longer perios of no winners, obviously.

Second, while computer generated "pseudo-"random numbers can be very good, they will always give teh same result if you "go back in time". This is what the people in teh article have done. They have taken the slot machine program and run it to a certain point, then made a copy of teh program, including all the "state" (all the variables in memory). Now each time they start that program they see the same result.

This raises several interesting points. Most obviously (to a programmer) - they didn't need to go to all that trouble. Any computer scientist could have told them that if the program has no extra source of random numbers from "outside" (you can buy physical devices that use the "noise" in the environment, or radioactive decay, to generate "real" random numbers) then it has to repeat every time.

Also, just because it repeats every time it's started from "the same point" doesn't mean that the program is "fixed from the start". It might be that the program really does calculate the next random number to work out what the "high" button should do. The problem is that this random number will be the same each time the program runs because it has started from the same frozen point.

This is a subtle point. You can argue that there's no difference between two different ways of writing the program:

(a) You put the coin in the slot, the amount of money you will win (or not) is calculacted immediately, and you then see lots of lights and press buttons, but the final result is what was calculated at the start.

(b) You put a coin in the slot and, each time you have to do something, the program calculates the next "random" number to decide the outcome.

From a very theoretical point of view (a) and (b) are identical. If you freeze the prgram (including all the values in memory) and then restart from that point, you'll always win the same amount (it might be different for (a) and (b), but (b) will always give the same amount if it's run again and again from a certain point, the same for (a)).

However, as a programmer, I feel (b) is fairer, because there's less chance of an over "plan" to manipulate the user. But this is a rather emotional argument that I cannot justify on purely rational grounds (yet).
posted by andrew cooke at 2:12 PM on June 6, 2003

ye gods. i did preview that. isn't typing "teh" supposed to be l33t?
posted by andrew cooke at 2:17 PM on June 6, 2003

oh, and kindall has a good point. that's using the "user" as an external source of "noise" in the context of my argument. they could have done that (should have done that). i'll go have dinner now.
posted by andrew cooke at 2:19 PM on June 6, 2003

If I offer you a proposition where you give me as many dollars as you want, and I will return 90% of those dollars, and you take it, you are stupid.

That's a really, really simplistic way to look at the issue, and I suggest you consider that some people find gambling to be an entertaining and social activity.

Also, I'd like to note that not all the games have high casino rakes. Properly played blackjack (without counting cards) has about a 0.5% casino advantage, baccarat is about 1.2% and craps is around 1.4 or 1.5%. so for every $100,000 of well-placed bets at those games, you can expect to lose between $500 and $1500, though obviously some nights you may lose $2,000 while others may have you winning $2,000.

We work and work to make money... why should our relaxation have to involve not spending it?
posted by mosch at 2:41 PM on June 6, 2003

I'm pretty certain you can also buy truly random numbers. Enough numbers to last for the life of a slot machine would probably fit on a CD.
posted by Wood at 3:32 PM on June 6, 2003

Of course some people gamble for entertainment. I have taken out a set amount, and gone down to play a sucker game like roulette myself, fully aware that, more than likely, that money would be gone at the end of the night - in exchange for a bunch of "free" drinks, and the momentary illusion of being a high roller. I won't deny this for a moment.

We're not talking about entertainment here, though. We're talking about people who go in with a system, or feeling lucky, or expecting things to turn their way after X amount of time, and losing everything. Calling it a "tax on the poor" seems wrong to me; sure, a poor person losing a couple hundred bucks from their welfare check is proportionately worse than your average middle class tourist dropping a chunk of their vacation buget. However, spending money that you don't have - money that you need to pay the bills or feed your kids - on "entertainment" is just stupid.

I'm all for entertainment. Just don't claim that people are being anything but ignorant when they complain that they've lost everything in a game that they knew was stacked against them.
posted by majcher at 3:33 PM on June 6, 2003

a sucker game like roulette
posted by inpHilltr8r at 3:51 PM on June 6, 2003

The problem with that strategy is, there are certain numbers that are probably not in the ROM at all, as they are not valid 6502 instructions.

Well, I was throwing away some bits anyway, since I only needed the random numbers to place obstacles on the 40-column text screen. So it was good enough.
posted by kindall at 4:28 PM on June 6, 2003

As many have pointed out, if the slot machines had a "system" of any kind, some player would find this out, therefore it is better for the house to make them random. A properly designed slot machine used regularly will have a predictable return.
Gambling is all about probability, the house doesn't have to rig or cheat or even be better at the game, since the game is designed to make the house money. The only exception to this are games like poker, where there is no house ie. all the players are on equal footing.
Sports betting is a whole different ballgame, although many play it like they play the lottery, just picking horses semi-randomly. Proper sports betting is all about skill, skill in determining whether the odds set by the house are accurate.
I spend a few euros on a lottery ticket every few weeks, it's chump change, and I'd only buy something unhealthy with it anyways. This "investment" lets me day-dream of winning, after all, somebody does win. I also sometimes play blackjack in the bar, while drunk. It is great fun.
posted by lazy-ville at 6:22 PM on June 6, 2003

Mmmm, , Thomas Bass. I have both versions of this book (The Eudaemonic Pie and The Newtonian Casino - both well-worn), and I've always fantasized about following in their, uh, footsteps. Of course the "sucker" part of that gig nowadays is when they catch you, and escort you out, Casino-style.

Playing straight odds, though, the house has about a 5.25% advantage, unless you're a true sucker, and take the five-number bet with an edge close to 8%. Still, a lot classier than those humps slaving away in the blackjack pits...
posted by majcher at 7:02 PM on June 6, 2003

"This may be a stupid question, but why don't they use non-deterministic means?"

Because that's very hard. You can, as several of the above posters have pointed out, use the player as a source of noise.
Wether the player is deterministic or non-deterministic is another matter, she will certainly try to influence the chances of winning by modifying her behavior.

Another solution would be measuring the amount of time between two alpha particles in a small sample of radioactive matter, since quantum events are non-deterministic - but players could influence this as well by carrying a source of high-energy electrons, for example.

Some people have come up with some pretty strange ideas in order to generate truly random numbers. Packing it all inside an arcade machine, at a reasonable cost, is another matter.
posted by spazzm at 7:58 PM on June 6, 2003

An easy way to make deterministic (pseudo-random) algorithms less deterministic in a game like a slot machine is simply to continually generate numbers even when none are needed, i.e. when the machine is waiting.

My understanding of how the typical slot machine (at least in the US) is this is exactly what it does. So while the underlying number stream might not be perfectly random, the element of exactly when the person presses the button is pretty much so. I'd be astounded if you could ever produce a non-random string from a machine in anything other than a lab setting where you're reading the numbers directly from the chip. (And if you could, I'd be willing to bet the casino's would be lining up trying to find a way to fix that).

(Incidentally, the odds sited for craps above are actually a bit off. They look like the odds if you play a straight pass line bet only. You can actually improve them to under 1% for the house by playing odds bets. Unfortunately I've always found craps kinda intimidating, even knowing how to play it).
posted by piper28 at 9:19 PM on June 6, 2003

Table games tend to be much better, especially blackjack, craps, and baccarat.

Sometimes true, untrue, and untrue. Blackjack's odds depend on the state of the deck (and how many decks there are) and the fact that there is an uneven payoff that favors the player, not the house (you get 1.5x your wager if you have blackjack, the house only gets your wager if they have blackjack.) If the deck is loaded with aces and ten cards, and you trade blackjacks with the house, you'll come out ahead. You also need to play close to perfect -- making poor choices will quickly drop you back into the "house wins over time."

In craps, the house will win over time, and the basic bets are very close. The house edge on the pass line is 1.4% -- if you wager $100, over time, you'd get, on average, $98.60 back. Don't pass is 1.36%, add in the various Odds bets, and you can get a payback rate close to 99.7%.) But no bet on a craps table will reverse the house edge -- play long enough, and the house will win all your money.

Baccarat isn't played against the house, it's played against other players. It would normally make no money at all for the casino -- except the casino extracts a small sum from each wager for providing the table. Play long enough, trade wins with someone, and you'll find that the house has all your money.

Poker works the same way. The idea of the table games where you're playing against other players, not the house, is that you can play better than the average joe, and win more from him that you give to the house for playing at tier table. Remember the oldest rule of poker: When you sit down at the table, look around. If you cannot spot the sucker, get up -- you are the sucker.
posted by eriko at 6:53 AM on June 9, 2003

Oh yeah. Note that the house edge depends on you playing over time. If you keep better on a roulette wheel, the house will win your money over time.

If you want to roll the dice, so to speak, decide on your risk-to-reward ratio, place all the money you want to gamble on a comparable bet, and make *one* play. Randomness makes the house advantage weaker if you don't play often -- you could hit a roulette number, and walk away with a 35x payout. Give them enough tries, and statistics dooms you to lose, give them few, and simple randomness make or may not make you a winner.
posted by eriko at 6:57 AM on June 9, 2003