Gaming The "RNG"

[Dana_H]

Ever wondered why it is that 2 different slot machines very often come up with the same symbol patterns when they’re played at the same time?

The simple answer to why is pretty straight forward… computers can’t guess.
Random numbers can only be simulated by the software asking the system clock or software timer for its value at that exact moment, and then running that result through a bit of math to get a value that the software (game) can use. Once it has that value it compares it to rules, like if it’s between these 2 numbers… give this result, or if it’s between the next sequential set of numbers… give that result, and so on. Because these timers are not slow, the end result of all this looks like a randomly picked number.

But random they are not… because two of the exact same timer results will always give the exact same fake random number result, and because the timer runs in sequence like a clock, two nearly the same timer results will give you two fake random numbers that are also nearly the same. This is how we game the RNG… by throwing almost simultaneous requests at it to see what happens.

---

When you play a slot machine, after the result, you can see the point where it’s timer resets for the next play if you have your cursor on it. When your Eridium total pops up in the lower right corner of the screen and the ‘pay to play’ message comes up is when it resets. This can be used as a marker for when to start counting (or timing) to pin down one or more sweet spots in the timer results… sweet spots that yield favorable results, of course.

I know from previous observations that a sweet spot for 3 of a kind sits somewhere between 3 and 4 seconds after reset. I start one machine and then the second one very quickly afterwards (about the same tempo as a left mouse double-click). This is to reset both timers at close to the same timing, assuming there’s no long audio clip that delays the reset. It resets and I count to 3, then start both machines quickly…

Blocked… somewhere right around that 3 second time the results give fleshstick as 3rd symbol. The fact that both machines show him in the 3rd means I hit a deliberate miss at that time slot. I try 4 seconds next…

Same thing… Cherry block in the 3rd, but my timing is better… the first and second are only one symbol off from each other. I back off from 4 seconds to about 3.75 seconds…

That’s more like it, but the fact that I cant even tell how far that right machine under… or over rotated that last wheel means I’m still not there yet. I try just a tad more than 3.5 seconds…

Hit! double 3’s… that’s what I wanted. Looking at it, the cupcakes on the left machine over rotated by one symbol because that was the 2nd machine started… so I shave just a hair off that time, right around 3.5 seconds and start the left machine first this time…

Yep, exact opposite… this time teapots on the right went one symbol past. I stay at the same time but start the right machine first again…

Left side over rotated… this is getting predictable. I must be right on the sweet spot but not getting either the first or the second machine started within that little time window I’m shooting for.

Cleaned out backpack for room and tried again… a couple misses and after that…

My timings off now… still got both machines in the triple symbol time frame but I fumbled and the right one got started late. It’s ok this time because that’s a Storm Front

Nailed it! 3.5 seconds

OMG! What a time to miss. Right machine went first… just a microsecond too late for that left machine to hit all 3 in the triple time slot

Found the sweetspot again… not easy to do when merely counting to yourself.
Note: My 3.5 seconds may not be the same as other peoples, I’m counting 1 exactly when the Eridium total flashes. Who knows, it may be possible to stopwatch this stuff and peg specific symbol times as well…
Now THAT would be cool

[Nac_Mac_Feegle]

Oh wow, good work.

Just make sure you mask your i.p. address, don’t want to get traced down by the RNG police for breaking the code.

Going to try this in sanctuary in a bit

[Scootalewis (King of the Loot Midgets)]

So you’re saying it’s possible to guarantee three of a kind?

This could be veeeeeeery interesting. Rev up those watches!

[DeputyChuck (Sheriff)]

I wonder what you think of this @HeyCarNut

[3298]

[quote=“Dana_H, post:1, topic:1128312”]
Random numbers can only be simulated by the software asking the system clock or software timer for its value at that exact moment, and then running that result through a bit of math to get a value that the software (game) can use. Once it has that value it compares it to rules, like if it’s between these 2 numbers… give this result, or if it’s between the next sequential set of numbers… give that result, and so on. Because these timers are not slow, the end result of all this looks like a randomly picked number.
[/quote]You know, that’s not how random numbers are usually generated. Yes, time is used, but not directly.
Random number generators usually store a seed internally. When asked for a random number (in the C programming language: int my_var = rand();, needs the system header file stdlib.h included to work), they take the seed, mangle it with some math, then store the result of that as the new seed and return part of it as the “random” number. The clock is used once: to initialize the seed on startup. (C code: srand(time(NULL));, needs stdlib.h and time.h included.)
With a multiplayer game, there are a few more things to think about. I know of one game where the seed is sent across the network while connecting, then the random number generators on all computers work in parallel. That is very fragile, though, so other games just have one computer (the host) generate the random numbers, then they are all sent across the network.

That said, the slot machines might be specifically coded to rely on time alone, not random numbers.

[Dana_H]

Yes, you’re correct about seeded generation. I included the loop-timer-based description mostly because that is exactly what seems to be at work here.

[Dana_H]

Guarantee? no… up the odds a whole lot? yeah sure.
We’re talking about hitting something like a half second window with both machine starts… if you can guarantee that then I might put a few $$ on the fact that you’ll continue to get triples
I certainly had a fair share of misses during all this, but by the end I was probably hitting close to 80 - 85%, with a few perfect streaks in there once I got the feel for it down. The window for just triples seems to be a lot easier to hit than the one that gives you double-same.

[VaultHunter101 (So long, and thanks for all the fish)]

More likely that, since nothing else requiring the use of the random function is happening, you’ve managed to hit a happy coincidence in the random number stream (given the limited number of options for each wheel.) In fact, I did quite a bit of Tina slots myself trying to get seraph crystals, and I can safely say that I never got results like yours.

Try this: Quit the game completely, and try the same thing again tomorrow. If you get the same result, then you might be on to something. I’m dubious though - it’s hard to believe that the developers would make such a rookie mistake with the RNG of the slot machines as continually reseeding the RNG with the current clock time every time you pull the lever.

Edit: I will say that three-of-a-kind seems generally to be more common on the Tina slots than the regular ones in Sanctuary. What I get from them, though, is mostly green/blue/eridium/cash.

[Dana_H]

I WAS doing this yesterday
Also, this same method to track sweetspots by time works on Butt Stallion as well, but in that case, the time frame location that your trying to hit seems to reset (move) about once every minute or so. My best there was 6 gemstones in a row, after that I had to start over and find it again at a different wait time.

[VaultHunter101 (So long, and thanks for all the fish)]

You realise that, if the RNG is actually pretty close to true random behviour, you would expect to hit six in a row periodically regardless of how long you waited between “throws” of the RNG dice? You’ll get the same thing at the grinder in TPS, too: take a ton of blue-rarity gear and start grinding without using moonstones. You’ll get runs of purple, then runs of nothing but blue. Long runs will be rarer than short runs, but they’ll happen with predictably regularity if you run a large enough sample size.

[Dana_H]

I’m not disputing that at all. It very well could be a random burst… one the likes of which have never even been close to in my game… but ok, I can’t rule it out.

[Matrixneo42]

Do you ever open a chest and see two of the same item in there? I have. I’ve always found that amusing. It’s like the 10 or so numbers needed to generate the parts for the gun fell into a repeating groove in the RNG stream. That sentence is awesome.

[VaultHunter101 (So long, and thanks for all the fish)]

Agreed. Needs finger-clicks, candle-light, and a bottle of red wine though!

[Kurtdawg13 (Loot Whore)]

Because REASONS! and maybe SCIENCE!

[Carlton_Slayer]

I am certainly glad that you notice things like that and experiment with it. This is fascinating stuff…

[Dana_H]

Ok, so … people made me curious that it might’ve been some wild hot streak, so I powered everything off and went and ate some food. Came back and went to sanctuary. I was only there maybe 30 minutes or so…

Yeah it still works, and how bout this pile…

Not counting the 2 I still had in my backpack… every one of those was a triple 7. Tediore Customer Service skins for every character and a few Vladof Sickle skins.

Not done yet… time for Badass Crater Bar…

More Tediore Customer Service and Vladof Sickle skins…

What I noticed at these other machines: The Marcus machines are a bit more picky than the ones in Flamerock, not a lot of room for error, but they still like right around that 3.5 second mark. They like giving out skins!
The Torgue machines are different… I got a lot more triples when closer to 3 seconds instead of 3.5. They also sit further apart so it’s harder to set them both off at the same time. If you happen to see a psycho head stop in the first slot… run! Two of them nasty grenades at once is lot more difficult to get away from, trust me lol.

[HeyCarNut]

I think a lot of people will fall for this. Barring a video with rock-solid evidence no… “background helpers” are running, everyone should take such claims with a boulder of salt. There was zero statistical evidence of such behavior from the test in the good ol’ days, nor evidence of such behavior coded (so I heard, since, you know, reversing in a no-no…).

Could GBX have changed code to rip out perfectly fine RNG use and replace it with some “timer” - sure. Can’t imagine why they would, again easily checked.

That’s about it, gotta get back to my Flat Earth Society meeting, got another Bedford Level experiment to do…

[Dana_H]

Hold on a minute… “fall for”? I’m not trying to sell anything. I showed what I stumbled upon and did my best to describe exactly how I was doing it. You’re free to not believe or skoff at it all you want, but I’d appreciate it if you don’t fly in with your jaded glasses on and start insinuating things about me, my motivation, or try painting me as some kind of hack cheater simply because the precious “numbers” say it can’t be done. Not everyone is a glory-seeking 13 year old ‘gamer’ looking to impress people, and I didn’t deserve that.

[Wingsday]

He was making a valid point, and also joking. Refer to:

That’s about it, gotta get back to my Flat Earth Society meeting, got another Bedford Level experiment to do…

This game has been out for over 3 years now and has been played by many people with experience working with RNGs. I would think someone would have realised this a long time ago and mentioned it…somewhere, but to my knowlage, that is not the case. On top of that, this is your first major, potencialy gameplay influenceing post, we don’t really know you around here. You’ll have to excuse a certain amount of eyebrow raising.

End of the day, a few more people, (mabye myself included) will give this a test, and I imagine we’ll find you’re correct. But we can’t go around and just blindly accept everyting everyone says as hard fact, this is the internet after all.

[boombumr (Here comes the BOOOOM)]

To be fair, it took people almost 3 years to find out you could chain money shot outside of gunzerk, and that discovery was made by someone who wasn’t around the forums that much either.

Even the frequent posters here don’t know everything about the game, and to immediately discount something based solely on who said it first is pretty close minded. Last I saw, no where in the forum rules does it say you have to be at some undefined level of street cred to share the things you’ve found with the community. Anyone is free to share anything. And this took a lot of time on the OPs part. The least we can do (if its is incorrect) is to provide evidence contrary to his testing, in the same way he provided it to us, so that there can be no doubt that his results were a statistical anomaly. Same way the luck cannon theory posted by Glacier was revised when he was showed the thread by umihotaru.

And performing the test twice with picture evidence is enough to satisfy me. When I play slot machines, I also try to press the slots one after the other. I thought it would give me extra luck (yeah placebo) but if there is some truth to it I’ll be happy.