I designed this machine to win the jackpot everytime on our kid games like these, where you have to hit the button at the exact moment. The light comes by So the plan was for me to be a hero amongst my buddies and a veritable Chuckie Cheese. Robin Hood, giving out free tickets to all The little kids, but that’s not What actually happened. It actually turned out way more interesting and the more I tested the more I realized this was turning into a repeat of when I use physics to expose the rigged carnival games, because by building this super precise Little robot, I discovered arcade games like this are pretending To be one thing when in fact, They’re total scams now in order to explain just how bad they’re scamming you, let me first explain how my machine works and then walk through the results of our testing for the past two months.
Fundamentally, the way this works is you have a photoresistor right here And it senses the light five or six lights Upstream of where you need to hit the button to get a jackpot and then, after a pre-programmed, delay, the plunger comes down and hits The button To Build this, my buddy John And I started by designing the main base in CAD, and then we 3D printed it overnight, And then we attached the photoresistor at the end of the articulating arm. And then we took a solenoid with a plunger and attached it to the base. It’S powered by 20 volts worth of double-A batteries, And this is the same power pack I use to trigger the Super Soaker and the brains of the operation. Is this wireless Arduino board and it connects via Wi-Fi to a Super ghetto app that we made an app is critical because it lets you adjust the delay between Sensing the light and pushing the plunger.
So if you miss short, You just add a few milliseconds to the delay and try again And then to disguise the whole thing. We put it in a backpack and cut a hole in the bottom, so the plunger could access the big button on the machine So to set it up. You simply walk in set the backpack down and then pull out your phone like you’re checking dankmemes. As you can see, the disguise worked really well And no one ever suspected anything. So at this point We were ready for our first real-world test and the results were disappointing. We were getting jackpots, but only about every 30 tries and what was so frustrating is we got the timing dialed in so well.
We would miss short by one light and then not adjust the timing at all and then the very next run we This long by one light. So for 30 attempts without changing anything about half were missing one light short and the other half were missing. One light long, so this could mean only one of two things: either we had some errors in our machine that was introducing some variability or there was something shady going on with the game itself And troubleshooting, because things don’t go as planned is incredibly common with doing Any engineering task so to fix it, you have to Systematically do some detective work to figure out where the error is coming from. In the last interview Carl Sagan ever gave, He said science is more than a body of knowledge. It’S a way of thinking. So here was our thought process and if I can prove that my Machine was super precise, then the only reason we weren’t getting a jackpot would be because the game was rigged.
So I needed to start By looking at any potential sources of variability error. In my machine there were only three, so it could be the Photoresistor, the solenoid or the arduino. There was one External source we considered, and that was light pollution from all the flashing lights in the arcade could mean the input to the light sensor Was too noisy and therefore not repeatable. Now, There’s a trick when it comes to troubleshooting, where you test several Potential error sources. At the same time, and if I can get really precise results within this loop, Then I know that none of those components are the source of the error without having to take time to test each individually.
So to test all three of these at once, we made a simple app that flashed a white screen and started a stopwatch at the Exact same time, and then we set the trigger for exactly one second later and sure enough. We were getting results with one millisecond precision. We also repeated the test with an incandescent bulb just like the game and got the same. Super accurate results Now to appreciate that one millisecond of precision, one flap of a hummingbird wing, takes 12 Milliseconds, so one millisecond is literally this long.
If you pull out your phone and trying it the stopwatch to land on four point: Zero zero seconds – that’s really hard – And this is just a two decimal places. I had some friends of mine use a timer app that went out to three decimal places, which is milliseconds, and then I Challenged them to land exactly on four seconds and in one hundred tries these mere humans, with their pitiful non machine reaction times Were only able To do it once, whereas my robot can do it pretty much every time. So there’s definitely no error coming from the hardware components, But maybe the sensor is being told to trip at different times, because there’s so many additional Flashing lights in an arcade and to investigate that. We looked at the signal data from the sensor, Which looked really noisy And low like this.
Until you see a spike in Intensity every time, the light would come around, which is when the delay countdown would start, because we have such a narrow field of view On the sensor when we compared the time between each of the peaks. It was also precise down to the millisecond, so light pollution Wasn’t the issue either. With this in mind, I headed to a different arcade with some family friends and their kids to see if things would be any different And we did win a few jackpots, But for the most part, once again, We would alternate missing by one light on either side Of the jackpot, without making any Adjustments to my machine which, as a reminder, can repeatedly hit the same mark within 1/12 of a hummingbird wing flap.
And so at this point it was pretty clear that these games just randomly hand out rare jackpots, but just to be sure I did some digging and was able to locate a 25 year old manual for the cyclone arcade game and Sure enough. If you flipped all the way to the back to page 31, it states the machine owner can specify How often they want a jackpot to be won Which actually makes sense. Because in our testing a lot of times, We’d hit a jackpot right away and then go through a dry spell until we would finally get a jackpot again on average, Maybe 30 games later, and so in conclusion, My issue here isn’t that this arcade game makes money For the arcade owner, My issue is this: arcade game presents itself as a winnable game of skill when in reality, It’s a random dice roll That is heavily stacked against you And if you recall, from the carnival scam video, the most lucrative games for the carnival owner Are those where people overestimate their chances of winning? That is Exactly what happens in this game, especially when you might play a few games And get it to stop one light short or one light long thinking You are so close to getting a jackpot when, in reality, you weren’t close at all in gambling psychology.
This is known as the near miss effect, and people will spend much more money to try and win because they think they can just Do it on the next one. And for that reason it should come as no surprise That this scam machine is considered the highest earning redemption arcade game ever made. So if I were you, I’d save your tokens for skee-ball There’s a plunger in the backpack and then it pushes down on the thing. Is that cool, So you’re, basically cheating? Well I mean cheating is a strong word, I’m using an advantage. I might have created myself.
It’S a new year, Which means it’s time to make new goals, and one of my new goals for this year is learning how to make really cool motion. Graphics, animations, like the guys from the YouTube channel Kurzgesagt, And to do that. I’M using skillshare.com, Who were also kind of to support this video in this class. They cover all the basics as well as how to animate their cool little bird And there’s a bunch of free downloadable files.
So you can follow, along with the instructions, And skillshare has a bunch of different courses from photography to marketing to tutorials on that. You think you just bought – or maybe you want to learn, to make videos like this. It’S less than ten bucks a Month for their premium membership, and that gives you full access to all their classes and the first 300 people to use the link in the video description Get two months for you to try it out. So if you appreciate the work I put into these videos – and you want to start the new year off right by expanding your mind and setting a goal, you can Actually achieve click the link in the video description to get started. Thanks for reading!