Cheating Fruit (Slot) Machines

ebbdr writes "Ever think that fruit machines cheat you? You would be right, at least in the UK. This article provides proof that fruit machine outcomes are predetermined and that the players inputs have little, if anything to do with it. And it lets you download the emulators and machine code required to test the hypothesis for yourself...."

Comments

Vegas Machines??

[Anonymous Coward]

for the love of god, does anyone have the ROMs for the machines in vegas?? i'd love to see what the hell is going on there

Re:Vegas Machines??

[JJahn]

Vegas is completely regulated as well, so I would expect that they are all predetermined also.

where'd they get the rom from?

[Wakkow]

I can't seem to find -where- they got the ROM from? Seems like a crucial part of it to say what particular model/version.. I mean, even the screenshots have different quality graphics.

Not to say they're lying, but I'm not convinced of their "proof". Anyone else see something I missed?

Re:where'd they get the rom from?

[terpia]

That's exactly what I was wondering from minute one of reading the site. They make not only make no mention of how they obtained the roms, but there is nothing saying "we'd like to, but cannot disclose how these roms were obtained/delivered". These are the only ways that I can think of:

• 
  Someone unlocked a machine and stole the board, and put the chip on a rom reader and dumped the code. Obviously, highly illegal.
  
• 
  Someone has a connection on the inside with someone who has access to the raw code. Not likely, (job security issues) unless someone with access to the code recently got sacked or is otherwise disgruntled. Again, obviously highly illegal.
  
• 
  This could simply be a hoax.
  

Of course, if the Roms were obtained illegally - there's a perfectly good reason as to why they wouldn't disclose how they were obtained. But nonetheless, when someone is accusing a big money industry of something so severe and underhanded it seems to me that they shouldn't take the readers who are keenly interested in this for granted and expect them to believe all of thier accusations without some sort of vague effort to let the readers know that the information they are relying on is in fact somewhat trustworthy.

Plus, if they disclosed their methods, people may be able to test the Vegas and Reno machines in the same manner, which suck up more money than all of the UK machines easily, at least I'd imagine they suck up more money given the extraordinary volume of machines and players in Vegas.

Re:This is a surprise?

[cpaluc]

That's the same here in Australia where, I believe, we have the highest number of of pokies (slot machines) per capita.

Where I live, the machines actually have little stickers on them that state the percentage return of the machine (usu. about 85% i think). The stickers must be compulsory - otherwise I don't see why the manufacturers would put them on.

But how good is the emulator?

[Anonymous Coward]

I saw this posted on The Register yesterday, but I wasn't convinced by the argument presented. Exactly how accurate is the emulator that they have running the ROM? Being able to execute the ROM doesn't necessarily cut it. For example, is there a good quality random number generator in the machine that isn't being emulated properly? A disassembly of the offending machine code showing where such cheating has been implemented would be more convincing. I imagine that the code isn't too terribly obfuscated, so it should be doable.

Without a doubt the gambling companies are bastards, but the evidence presented in this case seems weak to me.

Re:This is a surprise?

[pm]

They didn't say that you never win on the "double of nothing" games, they said that it's predetermined whether or not you will win. So you have an option to play double or nothing as to whether you can guess if the next number will be higher or lower than the number that they show on the screen. It is predetermined that you will win or lose this regardless of the number that they show and what you happen to choose.

This isn't a claim that you can't win - but that winning is predetermined.

Re:How else would they work?

[jon_eaves]

In Australia (or Victoria at least) it's 87%, and it's done by manipulation of the payoff tables. So, a Royal Flush pays 500-1 when in fact the odds are much higher. (For video poker type machines)

The industry is heavily regulated and government monitored. I had a friend who built the hardware and software for some of the systems, and they have hard-core maths people working for them.

Very funny story though, there was a machine that was in one of the suburbs that had a very high frequency of migrants (Vietnamese) that was consistently paying out above the 87%. The company was suspicious they were doing something illegal causing the machine to pay out when it shouldn't.

Turns out the guys playing the machine were statistics professors (from Vietnam) that had analysed the payoff tables and found a weakness in the payoff and under certain "unusual" circumstances (like breaking 3 of a kind and throwing away 2 Kings to go for Royal Flush) the payoff could be increased.
The maths guys at the company were somewhat embarassed as they had to change the tables to account for this.

Ah the lottery

[Patrick Cable II]

Why not just stone the winners instead?

(it's not offtopic, if you've read Shirley Jackson's The Lottery)

Don't Trust Machines!!!!

[yintercept]

The problem with digital gaming machines is that it is too easy for the programmer to add twists to the algorithms that tweak the odds. It seems odd that they would bother, since the laws of probability come out in the casino's favor, they don't need to tweak the algorithm, just do a little basic math first.

As I recall, the Nevada casinos are required to post the expected payout and odds on the machines. For example, the expected payout might be 98%. That means the casino collects on average 2 pennies every time a patron shakes the hand of a one armed bandit with a dollar bet. The casinos don't need to pull any tricks beyond calculating the expected payouts for the different states of the machine and make sure the expected payout is less than one.

It is disconcerting knowing that there are machines which go even further.

As I understand, a well run gaming commissions tries to assure that casinos don't bend the rules any further than that.

Re:Don't Trust Machines!!!!

[doktor-hladnjak]

I was watching some TV program a couple of years ago on how people scam Vegas casinos. Part of the program talked about what measures (technological and otherwise) the casinos take to combat the fraud. This business with the "machines playing themselves" I think is that the machines are constantly updating their random state.

For example, the simplest way of designing a computer-based slot machine would be to calculate the random values on each play. The problem with this approach is that random numbers generated algorithmically are not really random. Supposedly, if you knew enough about the machine's design and a certain number of the last outcomes, you could determine sequence of psuedo-random numbers being generated and therefore the sequence of future results. What really happens is that the machines are constantly "playing", that is generating these pseudo-random numbers. This means that knowing past results is not really so useful for predicting future results.

All in all it was an interesting program. Apparently the biggest fraud of all time involves sticking some stick with a light up the are above the pay out tray. Slots use optical methods of counting coins being payed. If you stick the light up there then it thinks the last coin it tried to dispense somehow did not come. So, it just keeps paying out until it runs out of coins.

Re:ho-hum

[rice_burners_suck]

Seems to me that all you have to do is work out a winning sequence for a given machine at home on your emulator, or, if the RNG is different for each machine, on the machine itself, then make sure you're the first one in the casino every morning when they turn the things on. You'll clean up every time.

Now who's cheating?

And guess what... When YOU are the one using this "feature" to gain money, the casino owners will kick you out of the casino. It's ok when they cheat you by controlling the outcome of slots, but when you cheat them, they kick you out. That's why:

1. I don't go to casinos.
2. Except when I go to Indian casinos (for the food) and when I do, I do not "gamble."

Gambling is just plain stupid anyway.

Re:Psychology of gambling, the study involving rat

[Dr Reducto]

In the study involving rats there were three groups:

Group 1: Every time they pressed a lever, food came out.
Group 2: When lever was pulled, they sometimes got more food, but most of the time, none at all.
Group 3: Did not get any food when they pushed the lever.

Groups 1 and 2 constantly pushed the levr in order to get food. Goup 3 stopped pressing the lever after the lack of food. At one point, the researchers stopped providing food. Group 1 stopped pressing levers, given tht there was no food. However, Group 2 thought that the big payoff and kept pushing levers anyway.

Re:what?

[deanj]

Actually, there probably is a random number generator....the thing is, it's always being initialized with the same seed. That's how they can be sure of what will happen in each spin of the game in the examples.

The place where it isn't being used is the "high-low" pick (and other places). That's the kicker.

casinos are bad

[bigbigbison]

As a casino employee, I can tell you that casinos are some of the most depressing places in teh world. It is amazing that these people come back, some of them every day (one customer told me he has been there every day for 18 months) and lose and lose, but when they win the sligtest bit they feel like winners. Hey moron, I just watched you put a ton of money in that machine and you're excited about $250? Wow, you're only down $300 now!
The thing about casinos is that people think, oh maybe I'll win the nix spin/hand whatever, so they keep playing. Then when they do win, suddenly they thing they're "on a roll" and poof their goes the money they just won plus some more.
You want to know what is even more amazing? At least half of the people who work there are just as adicted to gambling as teh customers. You would not beleive how many of my fellow emploees spend their days off at the casino down the road. See, a lot of them were former customers of the casino i work at, lost a bunch of money and were forced to get a job, so they got one at the casino. One would think that this would cure them of their addiction, but I suppose it is like an alcoholic working at a bar.
Long story short, don't go to casinos. if you do leave your credit cards and checkbooks at home.

wrong

[martin-boundary]

The only way to guarantee this is by determining what the payout is as soon as the money's in the slot. The "pick high or low" and all of these other things are just meant to help keep the player interested, so that the player keeps playing.

There's absolutely no need to make payout decisions before play starts. Slot machines and related devices are very simple to analyse probabilistically. All you need is to write down the state transition probabilities, and run matrix mutiplications. It's all elementary Markov chain theory. I analysed a couple of systems of this type for a job interview at a gaming company once (I didn't take the job).

It's totally possible to tweak probabilities to get the desired overall payout average. The fruit machines in the story are a scam.

Re:Reminds me of the lottery...

[FunkSoulBrother]

I know this is a joke, but this is very similar to what the mafia does.

If there is a pick 3 game (1 in 1000 shot) in your state which pays says $500 on a $1 ticket, the local mob is probably offering $700 dollars on the very same ticket.

Now they are both terrible bets, but at least the mob is offering you better odds than the government.

Of course the gov't will cry bloody murder, because it wants to keep its lottery monopoly.

Mersenne Twister

[ajs318]

A Mersenne Twister is just a fancy name {probably made up by some marketing type} for a pseudo-random number generator. It's a mathematical process which I found quite instructive. To quote from the ZX81 BASIC Programming Manual, chapter 5, Exercise 6 {I have taken the liberty of altering the original, pure-mathematician's style to more closely resemble a programming style}:

Let p be a large prime, and let a be a primitive root modulo p.
Then if b[i] is the residue of a ** i modulo p (1 .le. b[i] .lt. p - 1), the sequence
(b[i] - 1) / (p - 1) is a cyclical sequence of p - 1 distinct mumbers in the range 0 to 1 (excluding 1). By choosing a suitably, these can be made to look fairly random.

65537 is a Mersenne prime, 2 ** 16 - 1. Use this, and Gauss's Law of Quadratic Reciprocity, to show that 75 is a primitive root modulo 65537.

The ZX81 uses p = 63337 and a = 75, and stores some b[i] - 1 in memory. The function RND involves replacing b[i] - 1 in memory by b[i + 1] - 1, and yielding the result (b[i + 1] - 1) / (p - 1). RAND n makes b[i] equal to n + 1.

Skipping the heavy maths, what this is basically saying is: you start with some number, and you do the following in order: add one, multiply by 75, subtract one (*), and take note of only the last 16 bits; then put the answer through the same process, and so on over and over again; then you will get a sequence of 65536 distinct values, each possible combination of 16 bits occurring once and only once before the sequence repeats itself.

This sequence is not random at all, but completely predetermined. Re-seeding merely jumps within the sequence to some point along its length. Re-seeding must be done at intervals to ensure some degree of randomness. Traditionally this has been done using some measure of time elapsed between switching the computer on and starting the program running; the units used must be small compared to the variance of this interval.

True randomness would require some totally random event, such as the time interval between particle decays in a radioactive substance, or thermal noise in a semiconductor junction. {The static picked up by an unconnected input of a sound card would be an example of this phenomenon, but would also include local noise sources such as mains hum and local RF interference.} In the case of a fruit machine, the time between player operations, and the sequence of hold and cancel operations, if applicable, can be used as entropy sources.

(*) I know, I know. You can multiply by 75 and add 74 and no-one will know or care.

As someone who used to play these way too much...

[cliveholloway]

This is common knowledge and not a surprise. These things work on cycles, but the cycles are spiked with extremely baron periods and extremely genorous ones.

Ten years ago the cycles were a lot more reliable. I used to hang out in an arcade on Leeds railway station during rush hour.

The way to play was to pay attention to how much other players were dumping in machines. If a machine had received 75% of the jackpot without paying out, as soon as the current player left, go over and put in up to 50% of the jackpot value. Play until you win a jackpot (ignore/gamble smaller wins). 80% of the time, you'd get it. The hard part was to walk away if it hadn't paid by then :) But if you stuck to that, you were pretty much guaranteed to come out up overall. I used to make around 20 quid ($30) a night when I played - over a couple of hours, so the payback wasn't that great :)

But then the cycles gradually got longer, with longer baron patches followed by an occasional triple jackpot (paid over three pays to avoid breaking the law!). At that point it was no longer statistically worth while playing.

The manufacturers though are experts at intermittant reinforcement. It took me a while to quit while losing.

Now I live in California so I don't have to worry about being able to do anything dangerous, addictive or interesting because the State very kindly makes all my bad habits illegal :) To paraphrase Eddie Izzard, "We all go down the library for a wild time" :)

.02

cLive ;-)

Re:Reminds me of the lottery...

[RabidOverYou]

Read The Autobiography of Malcolm X. He was a numbers man at one point. You'd pick a number, and they'd use the close of the Dow Jones on the next day, down to the appropriate decimal. Public info, impossible to spoof, and pretty darn random.

It's a terrific book!

Re:How (IGT, perhaps other) slot machines work

[Anonymous Coward]

Forgive me for being light on details, but I got interested in this a very short while ago and it's given my team something interesting to do over the weekend.

We are a team of reverse engineers, both software and hardware, ordinarily working in... shall we say, other fields.

We have gathered a sample machine (quite old but still in use) from a local club, with the permission of the manager (though not the machine owner, so we will have to be careful to return it intact before tomorrow afternoon - hence us being light on details, short on time and anxious to get through with another project). This has been a very entertaining project.

There is no hardware entropy generator in the chipset of JPM Monopoly.

There is no software entropy generator (from, for example, low bits of timings between button presses) in the ROM.

It doesn't need one. The generator doesn't get seeded from entropy.

The number sequences for hi/lo are programmed, Elite-style (table index generated from a ROM-seeded LFSR). There's a condition in there too, based on some running values and flags, which skip the generator and force the table index to a 0. The code interprets a 0 for a forced lose - no matter which way you gamble, you will ALWAYS lose that gamble. The source code must be a treat.

This isn't about simple predetermination. It's about a forced lose. Misrepresenting a lose/lose/collect situation as a win/lose/collect situation like that is criminal fraud, card-shark style.

We're going to spend the rest of the afternoon breaking the LFSR, eating pizza, and trying to figure out if there's an easy way to recognise generated patterns immediately prior to big payouts. One of the control bytes controls the frequency at which a particular light blinks during attract mode, so we think there may well be.

By the way, the technique you mention above (wildly exaggerating the likelihood of a near miss relative to the odds) is known as a "heartstopper" and is illegal in the UK. The National Lottery scratchcard people were, shall we say, politely asked to stop doing it.