Considering the amount of cards they tend to produce these days, makes you wonder if they aren't already using some kind of AI to generate a baseline of cards.
What amazing control in that video.
From the very start, spreading the cards so evenly that every single card can be shown to be in order, then even more the precision needed to riffle the cards together perfectly eight times.
Yep, I'm up to about 100 cards at this time, working on the basic cards first. Besides the cards, there needs to be more work on the AI modeling... currently, it won't handle any card with elements of randomness to it. I think I have an idea on how to treat it (without the AI cheating), so that implementation is coming up... soon.
That is more involved than I thought. I was looking for a covert information channel. For instance the assistant could hold the cards in 2 different ways, 4 cards represents 16 bits of information, plus 2 more bits for which card is the high card.
Here is a simpler one that impresses many people. Deal out 3 cards across, return to the top, 3 more cards, and so on until you've dealt out 27 cards in 3 columns. Have the other person pick a card but don't indicate what it is. Ask them to indicate which column has the card. Stack the 3 columns with that in the middle. Now redeal go through the same procedure. Repeat one more time. Now count the cards out. Card #14 is the picked card.
Why does it work? Well the first time you deal it could have been anywhere, the second time it is down to the 3 middle rows, the third time it is in the middle row, then you stacked it in the middle of the deck.
Author here. You are right, that was a lot of cards, but doing it in French is several times easier and faster than doing it in English due to the massive number of cognates and near-identical alphabet. The 5k deck was also, as you say, a massive time saver, if not less effective than making your own deck.
Yes - it boils down to “there are 10 cards. Pick some number of cards, give me the rest, and I’ll tell you how many you picked”. The addition of “more cards” very thinly disguises the basic fact that you are subtracting a number from 10. It’s not obvious to me how this could be done in a way that seems impressive.
The tools exist for people to experiment as if they had infinite copies of every card ever printed, and the people who most enjoy that sort of thing definitely make use of those tools!
Maybe you could use a handheld scanner type device to load the next image and send power to each card, or an electronic Rolodex where you stack the cards and it flips through a set amount reconfiguring them. Both of those ideas would require some hardware development though.
I forgot to add: very interesting project and amazing to see the fruition of your efforts!
> In one version of the task, one subject (always one—he spurned testing subjects in groups) is presented with four cards lying flat on a table, each with a single-digit number on one face and one of two colors on the other.
From the article. All the needed information is present.
It wasn't just me, you see. If it was, I probably could have gotten this done in one or two. But, because we are a team, we had planning sessions where I needed my team to commit. There, the team had long heated debates about some of the cards. You see, several had a great deal of uncertainty surrounding them. No matter how far you'd zoom in, some of the requirements were not fine-grain enough to be understood. Ultimately, we came up with a few prototypes and a/b tested to get at the best solution.
The photo of the card drawer open is really spectacular. I am curious how they identified faults with that number of individual cards. Any insights into that?
I'm sure with a little creativity one could figure something out. Figuring that out would certainly be easier than figuring out the order to put the cards in ;) Anyway, this is less about actually doing it and more about the coolness factor in knowing it's possible, and in seeing how to figure it out.
reply