FAQs
Index
What does Full Randomness mean?
Isn't Soft Shuffle just Pile Shuffling?
Note: i'll keep adding to these as they come up.
What does Full Randomness mean?
Think of Full Randomness as the ideal for randomness.
If we imagine a deck of 150 unique cards, there are a very, very, large number of different orders they can be in (150 factorial, which works out as 5.7e+262, or roughly 5-plus-262-zeros different orders).
There are three things we want to be true for an ideal randomisation:
-
All of the different orders the cards can end up in (know as permutations) should be possible.
-
Each of those different orders should be equally likely to occur.
-
The order of the cards we start with shouldn't affect the probability of each order being picked.
All three of those being true is what I mean by Full Randomness.
From our Full Randomness definition - that each possible order is equally likely, whatever the starting order - the equal likelihood for each output order is the difficult bit, and is what you don't get with regular physical shuffling.
But you *do* get Full Randomness with Soft Shuffle (it's the problem I wanted to solve and why I created it).
I discuss this more in the Overview page and video, and for a full description of how and why this works - see the Technical page.
Isn't Soft Shuffle just Pile-Shuffling?
In short - no, though I can see why initially it looks that way.
In fact Soft Shuffle gives you what pile-shuffling (and riffling/mashing) doesn't - Full Randomness.
What that means is every possible order of your cards being equally likely once you've Soft Shuffled, whatever the starting order of your deck (something riffling doesn't give you and pile-shuffling *really* doesn't give you).
It's fundamentally different from a pile shuffle because none of the randomness is coming from the actual act of dealing out the cards - that's why I refer to it as a 'pile deal'. Your phone generates a new, unbiased, ordering for the cards - and that's how we get all the possible combinations being equally likely. The dealing looks like a pile-shuffle but is in fact a clever way to sort to the order generated on your phone.
Pile-Shuffling is notorious bad at randomising cards - to the point it's banned in tournaments - but you could say Soft Shuffle is using pile-shuffling's non-randomness against itself by using it as a mechanism to make an actual random order (from your phone) a reality.
If you want a full description of how and why this works - see the Technical page, though it's touched upon in the Overview page and video.
Isn't 7 riffles / mashes enough?
My aim with Soft Shuffle isn't to gatekeep what level of randomness you're comfortable with - 7 riffles potentially gets you what I call 'good enough' randomness with 52 cards and that may be sufficient for you (and I discuss in the Usage page and video when I do and don't use Soft Shuffle myself). My aim with Soft Shuffle is to offer an alternative if you want more than what physical shuffling gives you.
So what do Soft Shuffle give you that riffling doesn't? Full randomness.
What that means is every possible order of your cards being equally likely once you've Soft Shuffled, whatever the starting order of your deck (something riffling doesn't give you and pile-shuffling *really* doesn't give you).
The 7 riffle statistic assumes a few things (primarily 52 cards and a good quality riffle).
7 riffles gets you to a point where you can't predict what order your 52 cards are in and every possible order being theoretically possible. That is very different to all the 52 factorial (which is a very large number) possible orders being equally likely from every starting order.
To think about it, imagine a Faro shuffle (or 'perfect' riffle - perfectly deterministic). You know the exact single output order you'll get, based on the input order. Adding in more Faro shuffles will still only end up with one possible (but different) order.
We add in more entropy (randomness) by them being 'imperfect' riffles, and that will change the possible orders you get out, but some will be much more likely than others.
As we add in more entropy with more riffles, overhands etc, we get to the point where we're have some confidence that all the output permutations are possible, but that's still not the same as them being equally likely.
For them to be equally likely we need to riffle it far far more than we'd ever do - technically an infinite amount (for them to be absolutely equally likely).
With Soft Shuffle I just wanted to be able to jump straight to 'they're all equally likely, whatever the input permutation' - and how that's achieved is explained in the Technical page and video.