Hacker News Reader

Comment by apjaffe | Hacker News Reader

Seems like one way to do the trick would be the following: Choose some pre-determined sequence of the cards and memorize it. Buy 52 decks of cards, and put them all in that same sequence, but shifted by various offsets, ranging from 0 to 51. Then hide all the decks in various places. Now when someone names a card and a number, you just need to figure out which deck is shifted by the right amount to put that card in that position and point them to the corresponding deck.

Anyone see any obvious flaws in this approach?

View on HN

Fits with the Teller quote about magic: "Sometimes magic is just someone spending more time on something than anyone else might reasonably expect"

The main flaw is practicality. The first thing you could do is cut it down to 26 decks, all you need to do then is have the participant count off the top with the face up instead of the bottom to replicate the other 26.

Then if you allow “off by one” or 2, or 3, you can shrink it down further.

Thanks for this. Just want to add, the "off by up to two" seems like a hedge for the author having his pick of three decks. Berglas couldn't remember or describe succinctly enough which of the three decks had the named card at the named index, but knew any of the three would be close.

If I can name any card and any position don't you need 52 ^ 2 decks? Or at least that many divided by 2 for counting off the top or bottom.

If you want to frame it that way, then each physical deck can handle 52 distinct queries. So you have 52^2 / 52 = 52 decks required.

No, because a single deck can answer for 52 queries. and 52 decks (offset 0-51) will be sufficient for any card and number combo

Oh, I see. For what it's worth, here is someone else performing the trick and it doesn't look like that's what's happening.

https://www.youtube.com/watch?v=mdXIVQ-asqU

It still requires memorising the positions of 52^2 cards though.

No, because all decks have the same order (just rotated).

Reply on news.ycombinator.com