(Close Window)
Topic: On Jordan's Premo
Message: Posted by: alecStephenson (Feb 17, 2016 12:33AM)
A quick comment on Jordan's "The Premo", from Thirty Card Mysteries (1919), as discussed in e.g. Magical Mathematics by Diaconis and Graham.

This basically goes: cut/shuffle once cut/shuffle twice, move card, cut/shuffle a third time, locate.
I've found a way to improve this.

Handling: You need to use the original handling from 1919. It is very specific and existed for a reason, but this has seemingly been lost in later literature.
If you follow the Magical Mathematics handling, the long-run success rate will only be around 50%. Which is exactly what Café member seneca77 found in this thread

Locate Card: I've come up with my own practical looking-through-spread id method which increases the chance of success.
Below is some simulation* results (100k simulations each) comparing mine to Jordan's original, and to the mathematical approach in the Bayer/Diaconis** Annals of Probability paper. The B/D approach is basically a "gold standard", but it cannot be used in performance.

% Success Rates:

Jordan's Method: 66.50%
My Method: 79.34%
B/D Mathematical: 82.79%

So not ground-breaking, but a good improvement of the success rate from 66.5% up to 79.3%.
And if that's too low you can always drop the first shuffle and then my method is 99.5%.

Maybe I'll write the method down in three years time, which would be the 100th anniversary.


The Technical Stuff:

*Uses GSR shuffling algorithm. The probs are fairly strongly related to the shuffling algorithm. The GSR can produce fairly blocky riffles; if the actual shuffles are better, then the probs will be a lot higher than those given here. For shuffles more similar to table faros, they will be in the high nineties. The probs also depend on the card displacement. For displacements of around 10-16 cards they will be fairly similar to the above, but outside of this they will decrease. Please see the original Jordan (1919) handling to understand this. If anyone is vaguely interested, I have a graph of probs vs displacement positions and you can pm.

**The Bayer/Diaconis paper used the distance metric d(i-1,i) + d(i,i+1). It is a good concept to consider this as a "gold standard", (even though there is no proof of optimality in the paper). To directly implement this within a reasonable time frame you would probably need a computer. Note that Sect 2 of the B/D paper is somewhat misleading: it does not actually simulate "The Premo", because the simulation therein is only for the case where all shuffles are before the card move, and this is a very different scenario requiring different handling.
Message: Posted by: alecStephenson (Feb 20, 2016 02:56AM)
I've looked a bit more into computer simulations of Tamariz's T.N.T and Jordan's Premo. There are three ways they can fail, and if you understand these, and how they are related to card displacement, it is fairly easy to create handling variants that minimize/remove these modes of failure, and also create proper procedures for ID-ing (can be done as spread or as dealing).

For example, I've come up with a version of Tamariz's T.N.T that works essentially 100% of the time (no out needed). Which is quite neat.

Historical Note: I said that the Jordan 1919 handling was lost in later writings. This is not correct - it was also given accurately by Annemann in The Jinx #40 p267, where he discusses a performance variant to ID-ing from the spread. The original handling minimizes one of the three modes of failure.
Message: Posted by: NotThatLarson (Jun 30, 2016 08:25PM)
Do you think there are advantages to this as opposed to just learning a card control?
Message: Posted by: alecStephenson (Jul 1, 2016 09:00AM)
[quote]On Jun 30, 2016, NotThatLarson wrote:
Do you think there are advantages to this as opposed to just learning a card control? [/quote]

No big advantages really. It's a single card location, and there are much easier ways of locating a single card. See Tamariz's T.N.T (the one in Mnemonica, NOT the unfortunately named TNT in CCLight) and read his comments on it though, which are more positive. It does have the advantage of shuffling by a spec. And for T.N.T type effects you will need to know your MD cold (unless your revelation is as slow as molasses). The best thing to do is try it and see.
Message: Posted by: alecStephenson (Jul 10, 2016 05:46AM)
Actually, since my OP I've read Aronson's "Some People Think" which is a location with just one shuffle, which simplifies things.

I agree with everything written in that effect except for note 2: this suggests immediately announcing a card if it is missing during the deal. If you do that, then you could be wrong!

The problem with this is that suppose your chain is 15 16 18 and you announce card 17. But suppose the chain continues 15 16 18 17 19 20. Then you know it is either 17 or 18, but you cannot possibly know which. So it may be card 18, rather than card 17. This "switching" possibility is one aspect is not considered in any of these types of effects (except for Jordan's original, so Jordan was aware of the problem, although he incorrectly thought he could solve it by specifying a particular direction for the card displacement).
Message: Posted by: Andy Moss (Oct 27, 2016 11:20AM)
Alec, Only today I read an effect by Jordan called "Long Distance Mind Reading" on page 194 of my 'revised edition' of Jean Hugard's encyclopedia of card tricks (not the Dover edition!:)

This effect sounds like it might be in the same genre as the one being discussed in this thread. Is this an early prototype for an idea that later became improved and more developed?

In summary the procedure is-
Cuts, one dove tail shuffle, cuts, two more or less equal piles made, selection transfer from one half to the other, one of the two piles shuffled in any way.

Of course I could be completely confused and Jordan's effect "Premo" might be a completely different beast.
Message: Posted by: alecStephenson (Oct 27, 2016 08:22PM)
Long Distance Mind Reading is one of my favourites. You have to try it. People are amazed when you email then the correct card about 5mins after they give you the half-deck. Particularly if you tell them it might take you a few hours.

The two effects use a similar principle; Fulves' book (Charles Jordan's best card tricks) has a chapter called "Shuffle Systems" with 15 effects with the same principle, including both long-distance mind reading and the Premo. You can also download Thirty Card Mysteries (1919) for free from the (Australian) state library of Victoria, which contains both effects. The Magical Mathematics book seems to suggest that the Premo is a sort of "modern version" of long-distance card reading, but that's not right. They are two different effects.

I like Jordan's stuff in general. He was basically a maths/theory person, and his stuff seems often to be incorrectly credited to performers. For example, the Ireland Shuffle is his, and the in-the-hand antifaro is his; see Double Prediction* in the keycard chapter of EOCT - Vernon made a very tiny change to the reveal and it was then published widely as Vernon's "up and down" (e.g. #81 in Scare on card tricks).

*There is an error in the EOCT write-up; it omits the fact that packet needs to have 20 cards (and therefore 21 cards after the thrust). You can use any even number but then you need to change the prediction. I'm guessing there was a sentence at the start that was accidentally removed.
Message: Posted by: Claudio (Oct 28, 2016 04:37AM)
Great post, Alec, especially about the crediting corrections.

If I am not mistaken in Pit Hartling's new book "In Order To Amaze", there's an effect called "Sherlock" that uses a double riffle shuffle. I believe you have commented about the same ambiguity issue that may arise in this effect.

I have not got the book, so I can't be sure whether it's the same concept as Jordan's Premo? Or is just the same effect but starting from a memdeck and adding some out-jogging actions instead of laying the cards down? Just thinking aloud.

I find Jordan's Long Distance Mind Reading very crafty. I would never use it as is, but I was thinking that over the phone, it would be a killer. I am a software developer (Android, Windows). It should not take too long to develop a laptop or phablet application, which, from a known deck order, would allow you to move the cards into new positions as the interlocutor says their names, slowly enough (a 2s interval would be ok). The algorithm to determine the selected card would be easy enough to code. Just an idea I may try out if I have time.
Message: Posted by: alecStephenson (Oct 28, 2016 08:27PM)
Great idea. A nice thing about long-distance mind reading is that they have to choose what half of the deck to give you. It really messes with them. I have tended to get the card-within-half rather than the card-without-half. Maybe reverse psychology at play.

Pit Hartling talks about where the concept is from in the effect, so I'll have to leave that one. Jordan used stacks for everything since memdeck wasn't really around then (Nikola card system was 1927).

For Sherlock (two shuffles) if you follow the effect and deal with the switching issue I mentioned in that thread then you'll be fine. Theoretically there are still two ways you could mis-identify the card, but these are very very unlikely, so while they could be accounted for in performance, it really isn't worth bothering about and I haven't said what they are. You would need to be a complete pedant to be interested in these cases.
Message: Posted by: Claudio (Oct 29, 2016 06:16AM)
Regarding Long Distance Mind Reading, I did not think it through. There's no need for any computer when pen and paper will do as it is a simple concept.

As I wanted to try it out without having to mail the deck first, I asked my partner to assist me. I handed her a (falsely) shuffled pack of cards (in fact in memdeck order to facilitate the procedure) and went to the next room. I gave her verbal instructions. Once she was finished she called out all her cards of the packet of her choosing. I instantly told her the card she was "thinking of". She was amazed.

All I had to do was to have a pre-written sheet will all the card names in the deck order. I used 4 rows of 13 cards to make it easy to follow. I scored out each card she called out. Because I used a memdeck, it was easy to locate its name. At the end there was a glaring gap in a sequence (she'd called out the half with the missing card) while everything else looks copacetic.

If I were to perform this effect regularly over the phone, or from an adjacent room, I would make up a sheet with the names of the 52 cards (in memorized deck order for easy retrieval) in a circle that I could print out at will. It's much easier to visual the two distinct chains that way.
Message: Posted by: ratitekeeper (Mar 7, 2019 10:45AM)
Alec - It has been three years since your original post. Any chance you could provide details on your method?