Netbook-Run Dice Robot Can Rack Up 1.3 Million Rolls a Day

stevel writes "The owner of games site GamesByEmail.com created Dice-O-Matic, 'a machine that can belch a continuous river of dice down a spiraling ramp, then elevate, photograph, process and upload almost a million and a half rolls to the server a day. ... The Dice-O-Matic is 7 feet tall, 18 inches wide and 18 inches deep. It has an aluminum frame covered with Plexiglas panels. A 6x4 inch square Plexiglas tube runs vertically up the middle almost the entire height. Inside this tube a bucket elevator carries dice from a hopper at the bottom, past a camera, and tosses them onto a ramp at the top. The ramp spirals down between the tube and the outer walls. The camera and synchronizing disk are near the top, the computer, relay board, elevator motor and power supplies are at the bottom.' While not called out in the article, the pictures clearly show a Dell Mini 9 running the show (and performing the optical recognition of the dice values.) No, it's not running Linux."

Comments

Why?

[Publikwerks]

Why would you need this? And how is this better than a RNG?

Re:Why?

[Domint]

Because any developer worth their weight in salt will tell you that RNGs are not truly random.

Re:Why?

[gnick]

Trust.

People who know better will trust a good RNG just as well (as long as it's open source) - They're not perfectly random, but probably just as random as the dice roll. But if you're dropping $$ on the roll of a couple of dice (especially if you're remote), people will put more faith in a couple of pieces of bouncing plastic than they will a computer telling you that you just lost your $100 with no explanation.

Of course, that's purely speculation - Why RTFA when you can just glean through the comments.

Re:Why?

[CheddarHead]

From TFA:

"To generate the dice rolls, I have used Math.random, Random.org and other sources, but have always received numerous complaints that the dice are not random enough. Some players have put more effort into statistical analysis of the rolls than they put into their doctoral dissertation."

So, basically it was to quiet complaints about the randomness of the computer generated dice rolls. I question whether it's really better, but the players think it's better and in this context I guess that's all that matters.

Re:Why?

[gnick]

That's true RNGs are not truly random. But, then again, neither is anything else. Just sufficiently random to be indistinguishable from an actual random event.

Re:Why?

[Anonymous Coward]

And yet, there is probably some bias in the way that the dice are thrown. This project is cool, but as a source of random numbers it is quite silly.

If this does not qualify as random, what does?

A dicerolling machine external to and slightly more complex than the universe itself?

I normally don't mind the arguments or objections that this or that is psuedo-random rather than truly random, especially when it comes to computer generated numbers designed to do X or Y.

But really these are physical dice. If this doesn't cut it for you, what will?

Re:Why?

[SCPRedMage]

"Dices"? Really? "Dice" is the plural, "die" is the singular.

Re:Why?

[CherniyVolk]

I would be interested in seeing him run this machine for 30 days and then compute the Shannon entropy [wikipedia.org] on the results and then compare this to popular RNGs out there.

After reading the article, I think he's less focused on mathematical accuracy and more focused on appeasing customers. He only went to these lengths due to complaints against his RNGs he used to use.

This way, those wannabe math majors can't so easily complain.

Oh, and regardless of Shannon entropy, it is a bit more obvious that this approach will more satisfy a feeling of randomness. Unless the die are weighted, their effect will be random. Or as random as one will possibly get, or as random enough for those mathematicians who work for Casinos across the world who use real die for a variety of Casino games and are willing to shell out hundreds of millions in winnings should the roller win.

It's already agreed that if we knew the entire state of the universe, that we could see past present and future (just as we can determine numbers either side of a target number within a known sequence or set). But, it's also agreed that it's impossible to know the entire state of the universe. While the universe is big, it's probably agreeable that you don't know the entire state of this machine, nor is it possible. So as each dice has vastly different atomic landscapes on it's edges, and as the surfaces vary greatly with wind currents randomly blowing through, variations in humidity, vibrations, electro magnetics and all sorts of subtle forces and their variations and effects on one another and the die... those dice rolls are as random as they can get.

Please note, the machine can be manipulated. But, this has nothing to do with the fact that throwing dice upon a surface is sufficiently random.

Re:Why?

[asdf7890]

Why would you need this? And how is this better than a RNG?

Erm, it is an RNG. A proper one at that, not a PRNG.

OK so there might be a little bias somewhere in the system (a slight manufacturing defect in some of the dice making the chance of getting a six 1 in 5.99999999 instead of 1 in 6, or perhaps some oddity in the optical processing code that makes it fail to recognise the colour representing four more often that it fails to recognise threes) but only a perfect RNG would not have a little bias like this and there is no such thing as a completely perfect RNG. There are statistical analysis and filtering techniques designed to detect and filter/reduce such bias in systems.

And on the subject of "why would you need something like this?": sometimes wanting something is enough. Sometimes the fun of creating something and the joy of a successful project completed are the whole point.

Re:More Like Color Recognition

[SQLGuru]

In the "good old days," we called that a clever hack. Solving a problem is about simplifying the problem space in any and every way possible. I've made similar "OCR" hacks when everything was going to be in a known font and size.

how long do the dice last?

[Dan Ost]

The article mentioned that the dice get beat up pretty bad at the bottom of the machine. I have three questions:

1. how long do the dice last before needing to be pulled out of the machine and replaced?

2. how are damaged dice identified to be removed?

3. does the software recognize when damaged dice are causing errors (for example, when the paint from a pip has been completely chipped off)?

Ig Nobel prize!

[Megane]

This guy deserves to be nominated for an Ig Nobel prize in statistics!

Re:More Like Color Recognition

[Patch86]

Why bother with numerical "dots" at all?

If it can do colour recognition (obviously it can) why not just have a single coloured circle on each side of the dice. When it's a blue circle, it's a "1". A yellow circle, it's a "2". Two yellows and three blues, 7 total.

Surely the need for a symbolic representation of the number is only necessary for us feeble humans, with our tendency to forget abstractions. For a computer, which need never forget that green means "6", actually drawing a picture or making a pattern seems pointless.