# Data-addressed loci: has anyone tried a linked-list memory palace? **URL:** \ **Category:** General Memory Chat\ **Created:** [August 19, 2026, 12:13pm UTC](https://forum.artofmemory.com/t/data-addressed-loci-has-anyone-tried-a-linked-list-memory-palace/113721 "2026-08-19T12:13:49Z")\ **Posts on this page:** 7\ **Page:** 1
**Author:** ![tumbleweed](https://forum.artofmemory.com/letter_avatar_proxy/v4/letter/t/43a26b/32.png) [@tumbleweed](https://forum.artofmemory.com/u/tumbleweed)\ **Post date:** [August 19, 2026, 12:13pm UTC](https://forum.artofmemory.com/t/data-addressed-loci-has-anyone-tried-a-linked-list-memory-palace/113721/1 "2026-08-19T12:13:50Z")
Hello from Almaty, Kazakhstan! Has anyone tried a memory palace where the data decides which locus comes next? I am new to memory sports, so sorry if this idea already has a name. I found this older topic on data structures (A CS perspective on memory palaces), but there the journey method is described as an array and the linking method as a linked list, as two separate techniques. What I describe is a mix of both, so I did not find my answer there. In most systems the route is fixed. You walk the same path every time, and the digits fill the loci one by one. I wanted to try something different: a route that is built by the data. **My setup.** I use Ben System encoding (3 digits = 1 CVC image), but with two dictionaries on the same codes: - Dictionary 1: objects only (things and animals, never people). These go inside the scenes. - Dictionary 2: people only. These are the loci. So one code gives me both an object and a person. For example, 141 is “t-o-t”, so the object could be a tote bag and the person could be Tolkien. The consonants also tell me where that person sits, so I do not need to learn a seating list. It comes from the code. **How it works:** - Every locus is a person with a 3-digit code. - At each locus I put a scene of 3 or 4 objects. - The LAST object in the scene is also the code of the next person. So the pointer to the next locus is free. It is already part of the data. And because objects are never people, I always know if an image is part of a scene or is a locus. **Example with pi:** 141 592 653 589 793 | 238 462 643 383 … - I go to person 141 (Tolkien). His scene: lawn (592), gum (653), library card (589), comb (793). They interact in a chain, each one causing the next. - The last object is the comb, 793, so the next person is 793 (Cobain). - I go to Cobain. His scene: 238, 462, 643, 383. - The last object is 383, so the next person is 383. And so on. (My real dictionary is in Kazakh, so the images above are just English examples to show the structure.) **My questions:** 1. Does this already exist? Does it have a name? 2. **Collisions.** There are only 1000 people, so codes repeat. In the first 9000 digits of pi, 120 people get more than one scene. That is 43% of all scenes. But I also remember where I came from, so the pair (previous person, current person) is 6 digits. I checked the real digits of pi: there are zero repeated pairs in the first 9000 digits, and only 8 repeated pairs even at 70,000 digits. Is this correct, or am I missing something? 3. **The weak point:** this is a linked list. If I forget one scene, I do not know where to go next, and I lose everything after it. A fixed route only loses one locus. Is this why nobody uses this idea, or is there another reason? 4. Is the second dictionary worth the extra work? Or would you use one mixed dictionary and separate the roles in some other way? I am not competing. This is a hobby, so I can accept some risk. But I would like to know if people have tried this before and stopped, and why. Thank you. ---
**Author:** ![Stefos](https://forum.artofmemory.com/letter_avatar_proxy/v4/letter/s/e95f7d/32.png) [@Stefos](https://forum.artofmemory.com/u/Stefos)\ **Post date:** [August 19, 2026, 7:16pm UTC](https://forum.artofmemory.com/t/data-addressed-loci-has-anyone-tried-a-linked-list-memory-palace/113721/2 "2026-08-19T19:16:16Z")
Hi and Welcome 🙂 I will only answer Question #1 and #3 as I have no experience in this particular memory system: 1. Memory palaces are a series of linked items, concepts and/or everything else including numbers. There is no special name for it other that a “memory palace.” 2. You can make up earmarks like placed a special item like a Star at every 10th place OR whatever you’d like 🙂 I hope this helps you! Stefos ---
**Author:** ![tarnation](https://forum.artofmemory.com/letter_avatar_proxy/v4/letter/t/eada6e/32.png) [@tarnation](https://forum.artofmemory.com/u/tarnation)\ **Post date:** [August 19, 2026, 8:05pm UTC](https://forum.artofmemory.com/t/data-addressed-loci-has-anyone-tried-a-linked-list-memory-palace/113721/3 "2026-08-19T20:05:29Z")
Maybe this link will help. [Review: Advanced Memory Palaces by Joe Reddington - General Memory Chat - Art of Memory Forum](https://forum.artofmemory.com/t/review-advanced-memory-palaces-by-joe-reddington/76032) ---
**Author:** ![Ganxel](https://forum.artofmemory.com/user_avatar/forum.artofmemory.com/ganxel/32/11903_2.png) [@Ganxel](https://forum.artofmemory.com/u/Ganxel)\ **Post date:** [August 20, 2026, 6:45pm UTC](https://forum.artofmemory.com/t/data-addressed-loci-has-anyone-tried-a-linked-list-memory-palace/113721/4 "2026-08-20T18:45:19Z")
Hi tumbleweed! 1. As tarnation points out, Joe Reddington’s book discusses quite a few data structures translated into mnemonic techniques. Worth reading even if they’re not all intended as practical methods. 2. Yes there will be collisions. This happens with all methods. I find that some sort of hidden context/encoding tends to automatically bring back the correct item – probably when you retrieve “123 → 456 → 789”, each association has a sort of temporal trace which prevents it from getting mixed up with “555 → 456 → 222” which you memorised last week. 3. Yes, broken links are a weak point of the link method. However, in practice you can usually recall some random item later in the list and work backwards to the “break”. Since all items (except the first and last) have two separate encoded links (e.g B has both A → B and B → C), you have a fallback strategy for retrieving any forgotten item. This is an advantage that memory palaces don’t have, but the tradeoff is that memory palaces allow you to mark an item as forgotten and immediately move on without throwing off the whole sequence. 4. Not sure and I don’t think there’s much value in theorising about that. I think you’d have to try with and without it a bunch of times to establish a meaningful benchmark and answer whether the overhead is worth it – maybe the failure rate with a mixed dictionary is very low in practice, for the same reason collisions aren’t as big a problem as one would expect. ---
**Author:** ![wesleytwiggs](https://forum.artofmemory.com/letter_avatar_proxy/v4/letter/w/439d5e/32.png) [@wesleytwiggs](https://forum.artofmemory.com/u/wesleytwiggs)\ **Post date:** [August 23, 2026, 4:03pm UTC](https://forum.artofmemory.com/t/data-addressed-loci-has-anyone-tried-a-linked-list-memory-palace/113721/5 "2026-08-23T16:03:52Z")
Brad Zupp talks about this in “Ben Franklin blowing bubbles at a sword.” Basically you would use 3 card pao to memorize a deck, but the object always dictates the location of the next image. ---
**Author:** ![Felixvolata](https://forum.artofmemory.com/user_avatar/forum.artofmemory.com/felixvolata/32/31723_2.png) [@Felixvolata](https://forum.artofmemory.com/u/Felixvolata)\ **Post date:** [August 27, 2026, 1:57am UTC](https://forum.artofmemory.com/t/data-addressed-loci-has-anyone-tried-a-linked-list-memory-palace/113721/6 "2026-08-27T01:57:27Z")
My own system does not use set loci or set actions. I call it FAFL (free action free location) and I am looking over yours to see if I can steal any of it for my own system! I thought you might be interested since it fits the criteria of free loci - ![image](https://forum.artofmemory.com/uploads/default/original/3X/b/8/b8c084cbb775348a55442d11e8fd3d232e12d289.jpeg) ---
**Author:** ![Felixvolata](https://forum.artofmemory.com/user_avatar/forum.artofmemory.com/felixvolata/32/31723_2.png) [@Felixvolata](https://forum.artofmemory.com/u/Felixvolata)\ **Post date:** [August 27, 2026, 1:58am UTC](https://forum.artofmemory.com/t/data-addressed-loci-has-anyone-tried-a-linked-list-memory-palace/113721/7 "2026-08-27T01:58:37Z")
I find no problem with breakages, or strength of chain - only in the speed of encoding. Recall is as fast as traditional loci.