I use 3-digit decimal numbers converted to 9 digits of binary. (List created via Major System.) A single association for each. Placed in sets of two via a memory palace.
18 digits per scene. With just 12 scenes (24 mnemomic elements) you can do 216 digits.
I use this to memorize the red-black order of a deck of cards in under 20 seconds, only uses 6 total elements across just 3 loci to encode a full deck of cards plus jokers. Its like magic.
You only need to get fluent in recognizing 8 three-digit binary sequences.
Here they are (with their associated traditional Major sounds)
000 = S
001 = T/D
010 = N
011 = M
100 = R
101 = L
110 = J/Sh/Ch
111 = K/G
So putting together 9 digits like 000-000-011 would give me S-S-M, which I read as SeSaMe, which I visualize as Big Bird.
101-011-010 is L-M-N, LeMoN.
So 000-000-011-101-011-010 is SeSaMe LeMoN… Big Bird bites a massively sour lemon and his beak curls up. Thats 18 digits in one extremely quick and simple scene.
The tough part… Building and learning a 3-digit system with 1000 entries so that its fast enough to not hesitate when reading the sequence. But the cool thing is, if you don’t have a 3-digit number system already, you can kill two birds with one stone and learn both at once.
If that seems too daunting, a 2-digit PAO works well too, you can do 6 digits per element, each scene will give you 18 digits, you just have more total elements to read and encode. But less to learn can be easier to build and train to fluency.
As for realistic number of digits that can be memorized… A ton. Depends on how quickly you can convert binary to imagery and how quickly you can build scenes and how many loci you can navigate through. I’d say 500+ in 5 minutes is reasonably attainable given a few months of work. 200 in 5 mins is definately doable.