So yeah… I’m not going to choose cards against him in the Online Memory League Championship.
It’s ridiculous that I’m in the same division as he is
Is the standard to round to the hundreds place in memory events? The impression I get from cubing (though I could be wrong) is that people tend to truncate to the hundreds place.
On a bad day, I sometimes struggle to remember 12.74 cards in 60 seconds!
That result is absolutely amazing. To watch him race through those cards was just breath-taking.
What an achievement. It makes you wonder where the limit is. Much like athletics, it’s eventually going to come down to hundredths or thousands of a second. Either way, we must be closing in on it.
Please explain what happens about a 1:39 in? He memorizes the cards, then please explain the method used to prove it. I am missing what happens after the memorizing. Thank you
There are three stages in Speed Cards with physical cards:
- Memorization - memorize a shuffled deck as fast as possible
- Recall - arrange a second deck to match the deck memorized, in a maximum of 5 minutes
- Checking - the original deck and recall deck are compared card by card to see how many cards were recalled correctly.
The video here only shows the memorization and the checking stages.
Does someone want to take a shot at telling me how you can encode 52 cards in 12.739 seconds. 4 cards per second. I assume he is using a multicard system, 2 or 3 but I dont understand how he can chain them so fast. Assuming a 2 card system with 26 pegs he is still linking in 500ish milliseconds. Seriously, how he do it?
Memory-Sports : Please describe us your number system and why do you use it.
Shijir : I use PAO (Person-Action-Object) in my number system. I use it because that is the first main system that our memory school teaches us and it feels comfortable and interesting to use.
Memory-Sports : Please describe your card system, with which you broke the Junior world record in Speed Cards at the US Open 2016.
Shijir : I use the PAO system in cards as well.
The “memory school” he mentions is once more the Mongolian Intellectual Academy, where most the top Mongolian competitors started out… I pointed that out before in this thread:
The record mentioned in the interview was at 23.22 sec from 2016, so not the one in the video. Maybe he has upgraded to a two-card system or maybe he’s still using PAO.
I’m not sure what you mean by chaining… if you mean from one image to the next, that doesn’t happen… you just place the compound image at the next location in the memory palace. Nobody at these speeds uses the linking method.
If you mean how he can combine P, A, and O into one image… consider a young child learning the alphabet and being amazed by how some adult can read really long and complicated words… sometimes even whole sentences in the same time that child puts together consonants and vowels to form a word.
The math in case of a PAO system: place one image per location (3 cards) for 17 locations. For the Junior record at 23.22 he spend just over a second at each location.
Assuming a 2-card system: place two objects (double-card each) in, on average, 13 locations. So that would be just under a second with the record in the video.
Usually you don’t encode the last three or so cards, but rather form the image after you put down your deck and stopped the timer. Also you get the remaining time of the 5 mins you have to memorize the cards that you can go over the whole memory palace again a few times before starting the recall.
Bear in mind that you can place very soft Images during the initial run with the timer and then strengthen them when you go over them again for the next 4 1/2 minutes. So it’ll take less time to place the image initially.
Thanks for talking this through. Reading through I can see how a 2-card system could get you to the sub-30 second range but the reaction time required for 12.74 seconds makes me suspect he has a better system. Doing the math, using a 2 card system is basically 2 images per second filed in 26 locations. theoretically he could ignore the last pair but just the act of sliding through the deck so you can identify each card in 12.74 seconds is a challenge. An efficient 3-card system might get you there but I have no idea what an efficient 3-card system could look like. 3-Card systems that i have seen either add overwhelming complexity or grow to an unreasonable size. He might have fighter pilot reflexes as well.
That’s not what a 2-card system is… there is an image for every 2-card combination; hence, 13 locations on average if you place 2 object (4 cards) per location. Have a look at posts tagged #shadow-system or refer to Ben system, etc.
Most people with a 2-card system actually only use the same location until they get to a double-card that starts with a red suit, using that as a signal to move to the next location. Therefore I’m saying on average above, because it could be anywhere between 1 and 26 locations; however, it is highly unlikely to end up in either of these extremes.
What you refer to as 3-card systems are in fact 1-card systems that place 3 items in one location. With PAO you’d have a person, action, and object for each card which you’d them combine into a new compound image. But really you only have one image per card for a total of 52 images.
But just like a 3-digit system where 357 could me MiLK using the major system a 2-card system uses two cards in combination… i.e., 52 x 51 images or 52 * 51 / 2 if you move to the next location every time something starts with a red suit, using the same image for a spade / heart combo as you would for a heart spade combo.
Great topic. A lot of information on speed cards. Thanks.
Found this video:
12.74 seconds is seriously impressive! I was amazed even when Alex Mullen brought the record down under 18 seconds.
One potential shortcut is to memorize the last few cards visually before stopping the timer, and then encoding them with your standard system. This could bring the number of card pairs down to 24 or even 23. Only a small improvement on remembering 26 pairs, but at this sort of speed, every little helps!
What’s your rational here exactly? If you can do 52 cards (or if you like to look at it as 26 card pairs), then that’s a quarter of a second per card (half a second per pair)… how’d switching from system, to visual, back to system help?
Ah apologies I didn’t explain this very neatly - what I meant was using the working memory (sound or visual components) as a buffer for the last few cards.
In fact, reading back on this it’s a point you have previous made, but more clearly:
If this is a technique that you use yourself Bjoern, do you encode the images of the cards, or the images of the objects (etc.) that they represent? And do you do it as images or by reading them aloud e.g. “dragon-spade”?
Fast card memorizers don’t really see the card images themselves, just as a good reader doesn’t notice the individual letters when they see a word like “walk” - they go straight to the meaning.
I think “grabbing” the last few cards in working memory is probably only useful to slower memorizers like me. But I would be interested to know whether any sub-30 memorizers treat the last few cards any differently.
Insane attempt, PAO is super good indeed. Lot of practice also required.
I think the reason why we don’t see these types of attemps often is because it is risky for a competitior to do in competition. If they don’t do well it ruins comp rankings