Hacker News Reader

As I write this, I think you have a typo there; 94-97% payout on $1000 is $940-970. (But I am confident it is just a typo.)

And you sort of state this, but I'd empasize that's per spin; for a $1 spin, the casino expects to make 5 cents and give back 95 cents... but then they expect you to put those 95 cents right back in with high probability, meaning that you can now expect .9025 of the original dollar, which again they expect you to put back in... it doesn't take much math to understand where this value is headed if you keep feeding the machine its own output.

I have thought before that if you must play the slots, but you want to retain the possibility of winning, the right way to do it is to take $X, and play it through the machine once, keeping all the output separated. I mean, the right right way, statistically, is to set X to 0, but if you want to play for the enjoyment or whatever and still have a chance of winning something, this seems like the best strategy.

The "right" way to gamble is to decide how much fun gambling is and how much it's worth to you and then spend that amount. Like for example I could go skydiving for $350 or play a few hours of blackjack and poker for that much and decide to choose the latter. So you spend that much to get the expected result.

Sometimes you bust out quickly, sometimes you make money, just like sometimes you see a bad movie or concert for full price. But as long as the calculation is actually utility all is well. Of course in practice that's hard to do, our brains aren't quite wired for it.

Exactly! I love playing craps. The ups and downs are a big rush, plus it is super social. I set my personal limits pretty low so I cannot play much anymore in the bigger casinos where minimums are $25+. Find a smaller casino with $2 minimums, a rowdy crowd, and it's a great time :)

Yep, same for me, my mum taught me as a kid that you should never aim to come out of a casino with money. Your wager is your investment in the fun of that experience, take £20 in to play an hour on the 1p/line slots.

In the same way I play the lottery, don't expect to win, but for me the cost is worth it for the moments of daydreaming I get from it.

Yes, totally agree with this. This is the most appropriate way to gamble. The casinos are viewed as an entertainment industry, and that's exactly what people should treat them as.

For me, I realized years ago that even when I won, I didn't enjoy playing. That was an odd observation/realization to experience when it finally occurred to me. I was down in New Orleans for a bachelor party and we all won a bunch of money because the table was hot - everyone roughly made enough to pay for their trip. Which was great. But then I realized that even though I was thrilled that I just got a free trip, it occurred to me I didn't actually enjoy the experience (time spent) of winning the money. I derived little to no entertainment/pleasure from it as it was happening, even though the process was addicting. Luckily, we managed to all agree to leave while we were up big.

I think if most people actually sat down and thought about it, I think they would find the same true for them. Ever actually looked at gamblers when they are playing? They all look miserable. High stakes, low stakes, slots, table games, doesn't matter. Most of them look miserable yet remain there.

Yes, that is a typo (I had 100$ in my mind ...).

Since these machines are fixed odds games each result of a spin is completely exclusive of the result from the other spin. So a 97% avg payout for a spin or over a billion spins (a game cycle) is the same.

The problem is that a player cannot take a hit a continue playing while a casino can keep doing it theoretically forever.

The problem with slots as with most games is to know when to quit. When a player who has lost 5 pennies coming back to bet 95 pennies is not the problem. The problem is for a player who has put in a thousand pounds over a year, wins a jackpot of 4000 pounds, does not cash out put again goes back to betting those 4000 pounds - that keeps losing.

Correct, as you said, know how much I want to spend on entertainment, spend it and return back. I've actually seen that in land-based casinos, people go out with a fixed moolah in hand, play and lose that and come back. In online casinos, though, there's no stopping them.

>the low income, unstable jobs, low literacy, blue collar jobs or are already on social support.

The few times I've been to a casino, I was really surprised by the crowd. It wasn't exactly Casino Royale. Just a lot of people who came off as lower class to me (clothing, speech, mannerisms, etc) and/or retirees, who also weren't exactly living it up. Heck, I even saw a woman slyly shove a $5 prime rib into her purse for later. The entire steak too, and without a ziplock or anything. She just shoved it in there to eat later. She was also wearing a bedazzled sweater with a pair of dice. I suspect a lot of these people end up identifying with a 'casino lifestyle.'

There was also something depressing about the slot machines that had cute cartoon characters on them. I wonder what the emotional and intellectual age of the people playing are. They seemed more like children who have yet to have the maturity and education to understand risk and financial security than adults.

>This industry is highly regulated

I imagine its regulated like most regulations in the USA: by the whims of the industry itself. They'll tolerate gambling hotlines and will fund them because ultimately that keeps real regulation away. The same way cigarette makers tolerated labels but, of course, were never pulled from store shelves.

That high percentage payout that you descibe is very misleading. There's an excellent description of this at http://www.abc.net.au/news/2011-05-27/cummings---poker-machi...

wouldn't a payoff of 97% means that with a fixed amount of money going in, over time you end up with $0? the payoff is exponential in the wrong direction, right?

The graph for the payouts would look like exponential decay.[1]

Lets say you walk in with 1000 dollars in your pocket.

If the payoff is 97 percent, you would expect the money in your pocket to be .97^n where n is the number of times you took the bet. Each bet reduces the money in your pocket by 3%

While the curve approaches zero, the real limit is the minimum bet of that particular game.

1. http://teacher.nsrl.rochester.edu/phy_labs/AppendixD/Appendi...

Not really. The graph for payout would be jumping up and down. With an exponential decay, you would mean no player would make any money.

You should remember, the industry would not have been so lucrative or addictive if there were no wins at all!

That is the wrong interpretation - atleat the math says so.

The 97% payout is a payout over millions (billions) of 1$ bets made on the machine - called the game cycle. You have to remember that players win and lose too.

Theoretically, if I were to wager 1$ over a million spins, I should be left with 970K$. That is what happens too if we consider the payout of the machine over the full lifecycle of the slot game.

The problem is that a player's funds are always limited. So when they hit a losing streak, they would lose all money by the time a winning streak can come their way. The house however has theoretically infinite funds at their disposal.

With a mechanical slot machine I can understand the reel thing. But with electronic screens, software making up reels, how do I know that the software does not fool me even more?

Take the given example with first slot 1/16, second 1/32 and third 1/44 chance of winning big. Now assume the generated numbers are really random, and that day or week the machine has given two big prices. Then a third big number combination is created, and here a new rule takes over, changing one of the numbers so the price is smaller.

Is this regulated or how does this work?

Gambling software in casinos is regulated extremely closely.

I've personally seen casino multi game slots that have had either options built in to them via admin panels or grievous enough software issues that mis-wired multi use buttons could allow a player to either lose horribly or continually win/compound upon wins. The coin or credit in/out stat counters would also be affected and be pretty much worthless. These issues just throw ANY concept of magic "odds" chips right out the window.

They're not regulated that closely at all.

One of the regulations is that each reel is independent. You cannot change any reel outcome based on what the other reels are doing, and you cannot change any reel outcome based on past results.

I'm not 100% sure this is the case but I'm pretty confident.

Not necessarily. The math for video slots can be changed to handle that functionality, but it has to be fair and RNG driven. And of course has to be audited.

From a USA point of view: it's highly regulated at the state level. State gaming commissions hire third-party labs to audit the mathematics and software to make sure everything matches what the manufacturer says it does. Things like changing the odds dynamically is never allowed. The casino may load up a new program with different odds or mechanics, but for any particular game the odds must match what is published to the gaming board.

Now, this doesn't necessarily apply for non-governmental casinos, like ones on indian reservations. They are their own enforcers. Use that information as you will.

> From a USA point of view: it's highly regulated at the state level. State gaming commissions hire third-party labs to audit the mathematics and software to make sure everything matches what the manufacturer says it does.

Can you point me to more info on this "auditing" process? For all the recent coverage of "algorithmic fairness" (an area where I do some research) I'm not aware of any other usecases where mandatory software audits are common, so this is interesting. Who are these labs?

One of the most popular labs in this line of work is GLI (Gaming Laboratories International)

http://www.gaminglabs.com/

Here's a doc from their site about RNG testing:

http://www.gaminglabs.com/pdfs/GLI%20Comp%20Sub%20Req%20V1.2...

Thanks, this is pretty thorough. Do you by chance know how these verification systems arose -- goverment intervention or, somehow, from within the industry?

It's remarkable to me that good software verification practices (well, trusting this document and knowing only a little about verification in general) is up and running for casinos, not banks or other important software users.

From my experience, which is a very small intersection of a software validation background combined with work in the gaming industry, labs like GLI sprung up from demand. Gaming laws were written to require independent verification from the start which seems smart.

That said, labs like GLI don't really comb the entire code for bugs. They're looking for obvious things like backdoors, flaws in random number generation, payout table problems...blatant things that would make a machine fail against the player.

Bugs still do happen. Google for "$42 million jackpot denied". You'll find multiple instances over the years, and you'll also quickly see that the actual amount in question is almost always $42,949,762.95.

Why is that number special? =)

Heh. Overflow? Funny (maybe not to "winners").

Technically a 32-bit underflow (2^32-1, or 0xFFFFFFFF), but you've got it.

Remember, kids: don't mix error states and return values.

It is highly regulated. The software has to go through stringent gaming reviews every year, the RNG has to be re-certified, the software cannot be updated without prior regulatory approval, the game math has to be certified.

You cannot change rules based on prior winning strategy. Theoretically two spins can payout the jackpot (but the probability is very very low), and the casino has to pay them out as long as they were valid spins.