is it 2card and 3digit, no one has experimented w a 2digit pao, or like a 2card po ?, is 2card/3digit the ultimate arsenal ?
There have been improvements and updates and different ways to structure and read a 3-digit or 2-card set, like finding simpler more consistent ways to determine how to read it or by figuring out how to minimize subvocalization, but at the core, its still “figure out a way to translate a card pair or a 3-digit number into a single unique mnemonic element.” I don’t know that there is really much more to it.
Ultimately the system is usually less important than the effort and commitment to learning it.
That said, there are objective advantages to using a 2-card and 3-digit system. Given the same recognition and scene building speed, there are simply fewer intentional mnemonic elements needed compared to systems that only encode one card or less than three digits per element. I only need to read and work with 26 intentional elements to get though a deck and someone else needs to consider 52, then there is an advantage. If someone can get these large systems to the point of instantaneous fluency, I believe they will have an edge.
That is a big IF, however. The tradeoff, as you know, is time to onboard and train these systems to fluency, and time needed for maintaining sharpness via periodic drilling. The time commitment is steep initially, but someone can fairly quickly reach a point with a large system where they can rip through practicing recognizing every element in their system in a pretty short amount of time. I’m nowhere close to world champ level speed and I can run through all 2704 elements of my card/number system in around 40 minutes. An elite level competitor could likely review them all in 15-20 minutes. But thats only seeing each element once. But in 15 minutes at that same speed, you could review every element in a single card PAO 17 times. So the advantage in building a system to fluency definitely goes to the simpler ones.
I’m not sure that a 3-Digit or 2-Card PAO where there are 3000 unique associations for numbers and 8112 for card pairs is viable. Even taking the near impossibility of finding that many actions out of the equation, it is still a massive lift for likely minimal if any gains. Its just as fast to link two of the single associations as it is to like a specific P and O. There is still huge variety with just a single association per element.
I think extending to a true 4-digit system or a true 3-card system is unrealistic in a practical sense. There are just too many elements to ever be able to pre-define and pre-learn them all fluently. If you needed to make something up on the spot, you’d likely slow down enough to lose out over a fluent 3-digit / 2-card user. Its interesting to consider from a though experiment perspective but I think thats all it really will be. Usually I’m not one to say “we’ve reached the limits of possibility” but I think for practical purposes we probably have with the pre-determined 3digit / 2card systems.
3 Likes
If I could just chime in here for a mo, you might have played around with this at some point; but I got this idea from Buzan back in 2004, the idea that if you already have a 3digit system, you can expand it by having some sort of environment corresponding to the 10 usual digits to enhance it. For instance, let’s assume that 0 = fire, 1 = ice, 2 = forest.. (and you can keep going but for the sake of this example it’s superfluous). And, you have, let’s say your person for 654 is Julia Roberts. If you had to encode 1654, you could imagine Julia in a block of ice. For example. Or dressed as a tree, if you wanted to use the forest element if the number was 2654.
Now, the thing is, I never actually do this in practice. But have you? Or do you know of anyone who has tried it?
Rereading that I can see that it is somewhat poorly written out, but also, I don’t think I can really explain it much clearer. The problem with talking about a system you don’t actually use!
1 Like
This idea of using modifiers comes up pretty often. I posted about it a while back.
Here’s my issue with it. By using a modifier in the way you describe, Buzan isn’t actually creating a true “4-digit” system. He’s creating a “1+3 digit” system. He needs two intentional elements to encode that 1+3 structure. By doing that, he’s reducing the data compression to 4 digits across two elements, equivalent to the 2:1 compression ratio that you get from a 2-digit system. This “4-Digit System” is actually a DOWNGRADE from just using a 3-digit system where the three digits are represented by a truly single intentional element at a 3:1 compression.
(There is an important difference when talking about how many digits are compressed into each single mnemonic ELEMENT vs. how many elements and/or digits a SCENE contains, and how we communicate that when we talk about what a SYSTEM is capable of. I think people often confuse these things when talking about creating “a new x-digit system.” Here’s an older post about trying to clarify some of those terms to try to help folks communicate a bit more clearly.)
Think about a PAO where each element encodes 2 digits. It isn’t a “6-digit system.” It is a structure that allows you to encode 6 digits across 3 elements. A 2-digit PAO gives you a 2:1 compression ratio with the elements used and the scenes that are constructed. A 2-digit PO gives you the same 2:1. Even a single association system where every 2-digit number only gets one element associated with it gives you a 2:1 compression.
Yes, you can fit more digits into one SCENE by adding categories or modifiers like “person-action-adjective object”, or “environment-person”, or by repeating structural elements like making a “person-action-object-person-action” scene, but you need to consider the total number of elements that must be combined intentionally in order to create that scene and that ratio of digits:elements. Person-Action-Object-Person-Action isn’t a 10-digit system. It’s a 2-digit system with a 10-digit scene structure across 5 elements per scene. To me, the digits-per-scene number is basically irrelevant and not a good metric to use to compare systems.
When I talk about a 4-digit system as being theoretically superior to a 3-digit one, I mean a system that can take 4 digits and compress them into a single non-compound mnemonic element. So something like 1274 being represented by a “TaNKeR” truck would be a single element in a “true” 4-digit system. Compare this to the modifier system where 1274 is represented by a door kNoCKeR being encased in a block of Ice. The structure of that association isn’t really 1274, it’s actually 1-274, a compound image made from two distinct intentional elements. The Ice is an intentional single-digit modifier element that needs to be actively considered and incorporated into the scene when looking at the 1, then the 274 needs to be similarly translated so that you know what is being affected by the Ice. You get the same value (4 digits, 2 elements) from just doing a PO (ToNy CaR), PA (ToNy CaRRy) or even a 2-digit “adjective-object” system (TiNy-CaR.) In those cases, you only need to learn 200 associations, much easier compared to the 1000+10 needed with a Modifier+3digit system. If you’re only getting a 2:1 overall compression ratio from your scene, you might as well stick with a system with less associations to learn. Or, since you’re using two elements anyway with a modifier system, just combine two 3-digit elements to keep your 3:1 ratio: 127-471 = TaNK RoCKeT.
Lets look at examples of each type of system with 2 intentional elements per scene:
- 1-digit - 12 - Tea Neo - 2 digits per scene using 2 elements, 1:1 compression
- 2-digit PO: 12-74 = Tony Car - 4 digits per scene using 2 elements, 2:1 compression
- 3-digit + Modifier: 1-274 = Ice kNoCKeR - 4 digits per scene using 2 elements, 2:1 compression.
- True 3-digit: 127-471 = Tank Rocket - 6-digits per scene using 2 elements, 3:1 compression.
- True 4-digit: 1274-0274 = Tanker Sinker - 8 digits per scene using 2 elements. 4:1 compression.
The true 4-digit is clearly superior for data compression and scene capacity, but there are 10,000 associations to learn fluently, making it impractical. The sweet spot usually ends up being 3-digits with 3:1 compression where 1000 total fluent associations is realistically manageable.
1 Like
Nice, very nice. Some quality reasoning there Tim. At the end of the day (really, any time of the day) it’s going to come down to mental efficiency. And the ratios are a nice way to determine ahead of time how much work you are going to be doing.
My apologies for being off topic, the question probably would have been better suited as a personal message.
Much appreciate your thoughts though.
3 Likes