Simulation program flawed random?

Good day. I'm one of the lurkers here and finally make my post.

So as I understand our dear Wizard once make the challenge of the betting system. He used C++ (Which I believe is Windows?)
Now here is my question. Random numbers from computer are known as Pseudo Random Number. These 'random' number are got from some sort of algorithm.

This is the main point. Assuming he use Windows, there're some trouble here. Windows are known for being bad a generating number, which can be seen on this page here The random number turns out not to be really random, instead make a pattern.

I understand that the experiment use PHP, but no matter your language, when it comes to random, the language always takes benefits of the OS. Since the OS is pretty much flawed, then it means the simulation are also flawed right? (I will be very happy if somebody proof me wrong, seriously. I'm also a guy who benefits PSRN)

Thanks

Unless it gives a player an opportunity to use the information about the pattern, that there is a pattern may not matter for simulations. That would be my take, assuming you are right about this.

I could! Assuming that our wizard using Windows. I can starting making a strategy that beats PRNG! (I'm referring to the $20,000 gambling system challenge.) (Of course this is just Illustration, but since PRNG is a serious business (especially in internet security), I would like to hear what our Wizard will have to say)

Quote: Woohoo

I understand that the experiment use PHP, but no matter your language, when it comes to random, the language always takes benefits of the OS.

Nope. I'm pretty sure Wiz uses the Mersenne Twister in his simulations. MT is both platform-independent and very good for Monte Carlo simulations.

http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/emt.html

Quote: Woohoo

He used C++ (Which I believe is Windows?)

Point of order: C++ is a programming language which can be used across many platforms to write simulations.

I would guess that the Wiz uses Windows, although I don't remember it coming up before and I wouldn't be surprised if he uses Linux or some other flavor of *nix for his simulations.

As to the RNG question, MathExtremist hit the nail on the head with respect to the Mersenne twister being "good enough" for these simulations.

I tried to convince the Wiz to integrate with random.org to get true random numbers. There's even a C++ client already built! But he was not interested, which I totally understand given that the Mersenne twister code is already integrated into his programs.

Still, random.org is pretty cool. If I ever do a gambling simulation I'll probably use it to get random numbers.

Quote: AcesAndEights

As to the RNG question, MathExtremist hit the nail on the head with respect to the Mersenne twister being "good enough" for these simulations.

I tried to convince the Wiz to integrate with random.org to get true random numbers. There's even a C++ client already built! But he was not interested, which I totally understand given that the Mersenne twister code is already integrated into his programs.

The problem with using random.org is the speed. You've got network latency to deal with. A local implementation of the MT can generate 10 million 32-bit integers in less than half a second. The random.org API allows you to pull down 100 integers at a time (but not in the full range of 64 bits), so you'd need to do 100,000 requests in half a second in order to keep up. I don't think that's close to the expected responsiveness of the API.

Quote: MathExtremist

Nope. I'm pretty sure Wiz uses the Mersenne Twister in his simulations. MT is both platform-independent and very good for Monte Carlo simulations.

I do.

Quote: MathExtremist

Quote: AcesAndEights

As to the RNG question, MathExtremist hit the nail on the head with respect to the Mersenne twister being "good enough" for these simulations.

I tried to convince the Wiz to integrate with random.org to get true random numbers. There's even a C++ client already built! But he was not interested, which I totally understand given that the Mersenne twister code is already integrated into his programs.

The problem with using random.org is the speed. You've got network latency to deal with. A local implementation of the MT can generate 10 million 32-bit integers in less than half a second. The random.org API allows you to pull down 100 integers at a time (but not in the full range of 64 bits), so you'd need to do 100,000 requests in half a second in order to keep up. I don't think that's close to the expected responsiveness of the API.

You know, I had considered that briefly while writing my post but my thoughts didn't make it into the text box. The latency would definitely be an issue if you are doing a large simulation. "You win this round," he said jokingly.