I've been using visuospatial techniques like the method of loci (i.e. memory palaces, but I don't call them that) to enhance my memory for about 10 years - here are a few reasons why I think this particular solution won't really improve your memory, at least compared with normal diagramming/outlining techniques.<p>1. How you navigate through the environment matters. A lot. Let me use an example to try to illustrate how navigation plays a role in the method of loci. You're tasked with remembering how to solve a maze you must walk through. Consider how your brain ultimately end up encoding the solution. The brain only retains information that's (likely) relevant to solving some problem it is motivated to solve, and, also noteworthy, it tries to encode this information as sparsely as possible. Thus, what you end up remembering is not the precise spatial layout of the maze. Instead, you remember a stack of instructions/decisions associated with certain visual cues that contain a very small handful (say 1 to 3) of precise identifying details framed within a very undetailed visual "gist".[0] How does this relate to the method of loci? Well, as the encoding of your memory palace matures in your brain it is precisely these pictorial memories that end up solidifying in long term memory and becoming the "loci" in which you place the items you want to remember.<p>This entails some important things: first, to actually use a "method of loci" it is important that the environment change little. The solution here does not appear to afford that; you generate the environment, you can change the environment, etc. Well, learning is ultimately a generative activity, so this is okay. Chunking is a generative activity, and that is a highly effective memory technique. But I am going to guess that this, as a generative activity, is not significantly more effective than the generative activities of spider diagramming or outlining in a word document. Worse, if it's implemented poorly or lacks certain features it's probably worse. Second, to actually use a "method of loci" it is important that you have a clearly defined path to navigate. Again, the solution here does not afford this, as you seem to be constantly moving around wherever you like in the environment. Exploration is fine as a learning activity but if you never boil it down to a specific path (like you do in the maze example), you're not creating a true "mind palace" and getting the benefits thereof. Third, each loci must have specific and fairly apparent details that distinguish it from the others, or else your images will tend to get confused. Doesn't look like this solution offers much in the way of that with the little models you can place on the islands. I assume that will change over time.<p>As a side note, I also once tried to use the method of loci with such a top down perspective as you see in the demo (specifically I was using the Sims 2, ah how I love that game...). I found it didn't work that well.<p>Some years ago my experience was further validated by reading Mary Carruthers' work on the medieval idea of _ductus_, a core concept in medieval memorytechnique. Ductus doesn't have to do with a defined navigation of your body through an environment, though, but a defined navigation of your eyes through an image.<p>To sum up point #1, this solution does not help you produce strong loci like you get with a true method of loci. The "loci" it helps you produce are, if I had to guess, not significantly more effective than the "loci" you produce in your mind when you create, view, and encode the spatial relationships in, say, a draw.io diagram.<p>2. The loci are not being populated with imageable contents, and the method of loci works best for imageable content. I remember reading some cog psyche studies showing this somewhere but CBA to go find them. Also, the more specific the images the more memorable. E.g. apple is worse than Macintosh is worse than "that one Macintosh apple I saw on the floor at the store last week".<p>3. Related to #2, converting non-imageable concepts to images is expensive and, when you don't have at least a partly preconceived "memory language" of images, this almost always offsets the advantages of using the method of loci. I think the problem with non-imageable content is probably the biggest obstacle to using the method of loci for real work (There are some exceptions to this, e.g. if you're learning Art History or Medicine) and this does nothing to solve that big problem. Again, I think this puts it in the same league as other diagramming/outlining techniques.<p><long tangent>
The solution to this problem I found was creating a bunch of rules for creating such a "memory language". Most of these images (I call them "esographs") are compiled and a few can be made JIT when you need them. Another advantage of using these rules is that creating and using the esographs can be a mnemonic/memory exercise in of itself that can be used for real work. An example. I might use a black chicken to represent the concept of a decimal point. The connection comes from a story that I heard from watching the TV show QI[1]. The story goes that John Napier, the inventor of the decimal point, had a thieving servant. Now, to figure out which was the thief he persuaded his servants that he owned a magic black chicken that could identify thieving hands when touched. He placed this chicken in an empty room and had each servant go in and touch it. In reality, the story goes, the chicken was actually covered in soot and he identified the thief by seeing which servant had clean hands.<p>Now you can see that this image packs a whole parable-like story (which is itself a memory techniqe), and every time I use this image it reinforces my knowledge.<p>The trick to creating a "memory language" like this is to get into the habit of creating these images whenever possible and eventually you will build up enough images to have some sort of fluency in the language. Admittedly, this is a difficult habit to pick up. But after a while it has been paying off in my case, and can be enhanced with the method of loci. YMMV.<p>Digging through some old notes I found a list of good "rules" for converting non-imageable concepts into imageable ones. I'm not sure if I should share them since this post is already really long.<p>In any case, such a framework is actually more important than the method of loci since, without it, it's very difficult to use the method of loci for real work.<p>Funnily enough, such images representing abstract concepts is also core to medieval memorytechnique and also seems to be a major underlying motivation for religious iconography. And of course you can use the iconography as readymade images for certain concepts.
</long tangent><p>I guess the last thing I will mention is that the method of loci seems to work quite well for some people, especially when they are highly motivated and interested in content that is more imageable, but it does not seem to be all that effective for others. So YMMV.<p>[0] Anecdotally, I have a long commute that doesn't require a lot in the way of cognitive effort to navigate. I noticed that it took a very long time for me to remember the landmarks on long stretches of road where I don't have to turn.<p>[1] Season 3, Episode 3