WSOP 2008, Day 1 of the Main Event…Phillips beats Mabuchi’s Quad Aces with a Royal Flush… So there’s that. I was in a live tourney once and all of the 5’s were on the board at my table, so there is that. I’ve played at a half-dozen or more sites, and the gripe of the"non"-randomness of the program goes on everywhere, every site. People like to have something to complain about. We then have reason (we think) to be heard and voice our gripe. Behaviors, traits, and characteristics that validate us in the Universe, imho.
5 Likes
thank you for the sane well thought out post.good luck.
1 Like
Thank you spivak
Until I read you post I had decided to scrap replay, after the month I’ve just had
but you’ve convinced me to stick with it.
Not being that well educated means a lot of what you said went over my head
but I got the gist of it.
Also I think if you take the statistics away from the shuffle and look at the human
side if it . You are sitting at a table with eight other people in an environment where
there is probably a thousand other people playing. Now in life every one gets there
dose of good luck and bad luck and you just have to share .For someone to have
their share of good luck you just have to miss out.
Also the amount of poker we play is enormous . You could never play like that in
real life because you would have nothing to play with having lost it all because
as we are all aware it will get you in the end.
I think online free poker gives us a unique insight into how our luck ebbs and flows.
1 Like
I’m probably wrong but I think live play is less random… when you have a run of bad cards it lasts longer but on rpp (probably other online sites as well) turn arounds are quicker. If anything I think the way the shuffling is makes it advantageous to get players out more quickly because of the randomness.
Or I’ve lost my mind lol
1 Like
Live play is indeed less uniformly random. The lack of uniformity probably does not have that big of an effect on the strength of hands, but it is there.
1 Like
Obivously, many of you youngsters,newbies, and amatures have not read some of my observations. So here we go again.
First, my credentials. I am 72 years old. I was playing cards before Texas Hold’Em left Texas. I have played with the “old timers”,Doyle Brunson, Bill Preston(Amarillo Slim). before there was a World series of Poker. and I remember when there were only 24 of us playing. SO I have seen a few hands.
Now, remember, Everything that can’t possibly happen,does on a poker table will and does. The fact that sometimes the out come screws you,doesn’t figure into the equation. Of course, you never remember what you caught on the river,or flopping a Boat,or Quads, or the night you went home with 29K more than you started with. People always remember the bad stuff,and sluff off the good. Anyway,if you are going to gamble,which Replay is not. Remember, There is no all-knowing poker “god” who gives a rip about how you waste your time. If you bother to chart every hand played over a years time,you would find runs of good luck,bad luck,no luck and dumb luck. Finally as I have said before,Your bad beat was someone else’s coup, and when you river a boat,you probably handed somebody a “bad beat” worth crying over.
It is called gambling,if it was totally predictable,we wouldn’t do it.
.
3 Likes
Hear Hear!
spivak…how much did replay pay you to write this?
lol, I have no relation to replaypoker, financial or otherwise.
So how do you know anything about their RNG?
I’ll rephrase.
Do you have any specific knowledge of Replay’s RNG that any of us don’t?
Nope: No knowledge about the RNG of RPP specifically, but I know quite a bit about the topic of RNG in general.
I would be willing to give 100 to 1 odds that the RNG used by Replay is the Mersenne Twister. The reason that I am so sure about this is that nearly all computer programs written in the last 15 years use the MT (as mentioned in the original post). The MT is default in all modern programming languages, and you have to deliberately go out of your way to do anything else. There is very rarely any reason to do so and most programmers wouldn’t know how to if they wanted to.
Honestly, just read the post.
[quote=“Whittaker, post:50, topic:6651”]
So how do you know anything about their RNG?
[/quote]maybee cause SPG, reposted thier reply to another post, that came from gatsby.
That too 
Although that came after I wrote the initial post, and also depends on trusting gatsby, who has an incentive to make RPP look good.
Well yeah, Spivak… 
It also depends I guess on what the flip is being certified too…
I’m going to have to assume its 2fold, using a trusted MT/RNG plus demonstrating that no manipulation, during implementation, has occurred. Hell iono 
Lets think about this side of it, shall we … the difference between picking 1-52 , 500 times … and picking 1-x , 51 times ( where X starts @ 52 and decreases 1 each pick )… One of these we get to see the standard distribution over the 500 picks of one number (1-52), the other we get to see the 1st number in the data set, for 51 different numbers…sure its a dataset of its own, but in a different way…
There are many different “implementations” that all are “manipulation free” and use a trusted RNG … but I can tell you right now, different implementations create different patters over time… and when I say time, I’m talking 100m hands or more…
Spivak, I read your post. Your knowledge of RNGs is impressive. Your knowledge of Replay’s RNG is nonexistent. You don’t have a clue as to whether it is flawed or has been tampered with intentionally. Yet you said this: “I am pretty sure the RPP algorithm is perfectly fair.”
You are entitled to your opinion but without tests or specific knowledge of Replay’s system your opinion is no more reliable than a million others.
Assume you have a simulator that creates a uniform U(52) discrete distribution and you want to simulate x from a U(43) distribution. This can be done by simulating from u~U(52), and if u>43 you discard and draw again. If u < 44 you stop and set x=u. We can prove that this is so simply by considering n in {1,2,. . . ,43} realizing that P(x=n) = P(u=n|u<44) = P(u=n and u<44)/P(u<43) = (1/52)/(43/52) = 1/43 exactly, which is what we wanted.
Now given a way to simulate a continuous uniform U(0,1) distribution we want to simulate from U(43). This can be done in the following way: Simulate v~U(0,1) and let x=ceil(v43). Now as before consider P(x=n) = P(ceil(v43)=n) = P(n-1/43<v<n/43)= 1/43 exactly.
While these algorithms will not produce the same card, they produce the exact same probability distribution over the cards, provided in the first case that you correctly simulate from the discrete uniform U(52) and in the second case that you correctly simulate from the continuous uniform U(0,1).
In other words, these algorithms will produce the exact same stochastic behavior over as large a dataset as you want.
crap, ya lost me, and thats hard to admit, only because certain terms I’m not getting…
Why the 43 not 52, and from there no dead space ?? I’m used to @least having a clue what all the variables are, I also don’t work @ google in thier damn algorithm dept. , hahahahaha… Sorry, but poker, where its not u against the house ( need’n a way to skew odds 2-3% for the house ), only needs to replicate a real dealer and cards, not have some wierd formula it goes by for cards…
Even if the table has a auto-shuffler, live, how the dead pile is incorporated into the unused deck, does matter… as to the next hand…who folds and when, adds human randomization to the dead pile, and if the unused deck is still random, then a few fair shuffles from there, results in the easiest and most random new deck possible… to me its illogical, not to try and add in all the same variables you have on a live table, knowing at todays speeds of servers/computers/cons, almost anything you do will be almost instant…
If you wanna talk something NO OTHER poker site has ever had… talk about the advertizing potential, but anywho… Have a slider, just let the dealer button do it, and let the cards be cut live via that slider, or player can click for no cut… dealers’s choice…
Just 1 more layer of “randomness”, that I bet players would love…
This is fair I guess. I do not KNOW for sure whether RPP RNG has been tampered with intentionally, and I never claimed that I did.
I said “I am pretty sure” rather than “I know for certain”. I stand by that assertion.
I have looked hard at the accusations leveled at the system and have found that either they are insane or they did not stand up to a basic test.
43 is an example: If 9 cards are dealt you will want to simulate a random card from the remaining 43. You may replace 43 with any number of your choice. I just wanted to show that both methods produce the same probability for each card.
As for notation, u~U(52) means you draw u uniformly from the numbers {1,2,3, . . . ,52}. P(x=n) is the probability that the random variable x takes the value n. The vertical bar is “given” and the fromula is P(A|B)=P(A and B)/P(B) for any events A and B. “ceil” is the ceiling function. v~U(0,1) means you draw v uniformly in the interval [0,1] (or (0,1): It makes no difference because P(v=0)=P(v=1)=0). Does that clear up the notation?
You ARE correct in that the obvious way to randomize will not behave the same as physical shuffling. Most people, however, would not interpret this as a flaw in computerized randomization, but as a flaw in physical shuffling. I guess it depends on what the ideal of internet poker should be. Is it to emulate physical poker as closely as possible, or is it to achieve perfect fairness and uniform randomness? I would argue that it is the second, improving on the flaws in physical poker.
As for cutting the cards, you can give the players a slider and give the slider no back end at all: It’s just for show and does nothing. Nobody would ever know the difference and you can save yourself some time coding . . . Maybe some players might appreciate it 
Spivak,
Yes, as much as possible, only the 43 thing still bugs me, especially since 5+3+2+2= 12, then 40 would be the highest remaining number possible, even with only 2 players… so WTF on 43, or even on 40…
I’m looking @ this as abstractly and logically as possible… or trying to. Plus ya know I’m completely frustrated @ the moment (site), and not instantly understanding what you wrote piss’s me off more… and I thought I over analized stuff… hahaha.
I would never give the players a slider that was useless, thats deceptive as hell …
[quote=“spivak, post:59, topic:6651”]
Is it to emulate physical poker as closely as possible, or is it to achieve perfect fairness and uniform randomness?
[/quote]Since alot of what Replay does is not even close to “fair”, I would not say that that was one of thier main concerns… yes there’s a difference between “computer poker” and “online poker”, plz don’t confuse the two… and I know you haven’t yet, but we almost have…
[quote=“spivak, post:59, topic:6651”]
improving on the flaws in physical poker.
[/quote]This is why 9 outta 10 players will admit, this is why “internet poker” isn’t like playing live, once they have a decent grasp on the game itself… It also might be why every site I have ever played on, had ppl saying it was rigg’d… it almost begs the question Spivak… " Rigg’d in the normal conventional way, or Rigg’d in some bizzare mmorph like computer pattern of so called randomness "…
Why not mimmic the real thing, then be transparent ( as possible but protecting ur brand ) as to the fact you run the most realistic game in town.
I guess one of my questions to you was… due to variance and standard deviation ( if thats right )… is one more rnd than the other … picking 50 times from the same range, or picking 50 times from a different range … ie- 1 data set of 50, or 50 data sets of 1 … if that makes sence.
Laws of probability in my opinion, would favor 1 data set of 50 tries, to blend out the variance… rather than the 1st number in 50 data sets where variance would be huge…