What Would It Take???

[Frank Kneeland]

Thank you, backsider.  Let me give you an analogy of what I think is going on with Frank's "utility." He is coming up with a system to measure how much water is in a can of peas, because some people suspect there is too much water in the cans they buy which violates what is written on the label.  Frank's utility will measure the water and the peas in a can.  But if you use this to measure your can of peas it doesn't mean that all of the other cans of peas are not packaged as it says on the label.OK I just love this analogy. And with due credit to its originator I'd like to embellish.Some people believe the water to pea ratio is just fine, others are convinced that there's far too much water and they're getting swindled...but on both sides of the "pea water" debate you have people that have: 1. Never measured how much water there is.2. Never read the label to see what the correct ratio is supposed to be.3. Only shaken the can without opening it to guess at the water content.4. Listened to a friend of a friend that told then not to buy a certain brand of peas and taken whatever was told them on "faith".Even with a methodical pea water separation technique and the best digital scale all that measuring a single can of peas will tell you is if that particular can of peas meets, or fails to meet the standards printed on its own label. For all its faults, that would still be far better than opening the can, tossing the peas in your pot and convincing yourself that you'd gotten the correct or incorrect amount of peas based solely on whatever you wanted to believe.Faith is the primary domain of people too lazy to do proper research.  ~FK

[Frank Kneeland]

Progress Report and Request for New2VPI'm nearing completion of the first of the three modules of the utility.The part I did first is the Every Card Test, where one records all their dealt and drawn cards.Just to keep the sheet symmetric I ended up with 12 sets of 104 hand trials.@New2VP: What will the approximate confidence level be with 1248 trials dealing at least 5 cards out of 52, and drawing from 0 to 5 out of the remaining 47?1. Do I need more trials or could we do it with less trials?2. Remember, I was hoping for at least 95% confidence and 99% would be preferable.Assuming it takes people about 30 min to record their results from 104 hands. This would make the current test completable in about 6 hours. That's totally doable, but is it good enough???The longer it takes to complete the better the test becomes, and the less likely it will be that anyone ever uses it. I need opinions.~FK

[BOOPSAHOY]

6 hours in one sitting or on multiple days?

[Frank Kneeland]

6 hours in one sitting or on multiple days?
Whatever you like. That's up to you.One person suggested taking phone video. If you did it that way you could get the raw data in about 1.5 hours and then it would be just a matter of data entry.~FK

[moneyla]

Frank, Is this a total of 1248 hands that you are using to determine if all 52 cards are dealt/drawn equally? Just playing with some numbers... If you don't hold any cards from the initial deals, then you will see 10  X 1248 = 12,480 cards. 12,480 / 52 = 240 is the average number of times each card shows. Am I following you correctly?

[Frank Kneeland]

Frank, Is this a total of 1248 hands that you are using to determine if all 52 cards are dealt/drawn equally? Just playing with some numbers... If you don't hold any cards from the initial deals, then you will see 10  X 1248 = 12,480 cards. 12,480 / 52 = 240 is the average number of times each card shows. Am I following you correctly?You aren't going to be drawing 5 cards all the time so the numbers will very slightly but yes I think you have the general idea.This is only one of the three tests planned. The others tests are for ongoing play. One is strategy dependent, and the other is strategy independent. The other tests will be something one could do over months or years.Oh and thanks for the pea analogy. That was perfect!~FK

[moneyla]

Actually it is possbile to do a test of "test #1" in the purest sense which is to deal five, and then draw 5 cards.   If I were to do it on a machine playing pennies (one cent per hand) the out of pocket cost would be .01 X 1248 hands = $12.48. There are 50 play machines where you can play one line for one cent.  I'll splurge and pay the $12.48. Rincon has two of these machines.  I'm sure there are some in Vegas. (Just as a side note: one night I saw an elderly couple playing one cent on one line of a fifty play machine at Rincon... they were dealt a royal on the one line.  it made their night.  It made everyone else in the video poker area sick.  LOL)

[Frank Kneeland]

Actually it is possbile to do a test of "test #1" in the purest sense which is to deal five, and then draw 5 cards.   If I were to do it on a machine playing pennies (one cent per hand) the out of pocket cost would be .01 X 1248 hands = $12.48. There are 50 play machines where you can play one line for one cent.  I'll splurge and pay the $12.48. Rincon has two of these machines.  I'm sure there are some in Vegas. (Just as a side note: one night I saw an elderly couple playing one cent on one line of a fifty play machine at Rincon... they were dealt a royal on the one line.  it made their night.  It made everyone else in the video poker area sick.  LOL)Also a good idea. This was actually discussed on vpFREE. The one problem is that for some people it wouldn't test what they wanted to test, because they are concerned only about the max coin play and they believe that the RNG works differently if you play max or less than max coins.When I started this project I made it absolutely clear I had no intention of talking people out of their beliefs, so if this is a concern for someone, they wouldn't be able to use your 1 cent test method.For those that aren't concerned about 1 coin play "being different" it would be the fastest cheapest way to get the largest sample really fast.Thanks.

[moneyla]

Of course I beleive that the RNGs are random and it doesnt matter to me if the test is made with one coin or five coins. I just want to define a way that the test can be conducted. What I would like to see is someone who finds a machine or machines that are not random, because that would be the "man bites dog story."

[Frank Kneeland]

Of course I beleive that the RNGs are random and it doesnt matter to me if the test is made with one coin or five coins. I just want to define a way that the test can be conducted. What I would like to see is someone who finds a machine or machines that are not random, because that would be the "man bites dog story."I think that would be a fascinating story as well. The great part (as I see it) about this utility is that people will be able to share exact results done in an exact way and then others will be able to replicate the experiment, a condition for good science. I still don't expect people to believe what others tell them, but I expect people to be able to test it for themselves, under the exact same conditions of the original test. The creation of a standard test makes this possible.In other words, if you tell me that you experienced a 5 standard deviation negative swing in pay cards at a particular casino on a particular machine, I will be able to go to that exact same place and that exact same machine and run the exact same test again to confirm your results. No longer will people simply claim something that might as well be a Russel's tea pot.http://en.wikipedia.org/wiki/Russell%27s_teapotI see this as a very good thing. I'm 75% done with the first of the three tests.~FK