The method of loci (MOL)/memory palaces are a widely known mnemonic devices and enable memory artists impressive tasks (like memorizing several decks of cards, memorizing numbers etc.). Further MOL is featured in pop culture e.g. Sherlock Holmes, Hannibal Lector etc…
There is, to the best of my knowledge quite some research, which shows that MOL is working/useful for improved retention, especially when combined with spaced repetition.
It seems I have never seem real world examples of long term memory palaces/method of loci applications. It always seems like a short term crutch for cards, numbers, speeches, grocery lists, phone numbers, vocabulary or for test/exam preparation. For example it seems that in language learning, the MOL is for encoding some vocabulary and visiting it regularly, until it is committed to long term memory.
All examples I find in books about the method of loci are again only about having one location, a route of 10-N stations, and never about building/using mnemonic devices to organize bigger amounts of knowledge.
Are there any examples of people using the method of loci/memory palaces to organize for example their professional knowledge? Or of polyglots having/keeping memory palaces for language learning? Is there any research about long term usage of the method of loci?
Professionals have large networks of neurons. They are sturdy and efficient from repeated use. Memory palaces help to start the construction of these large networks of neurons. Afterwards, as another commenter noted, the knowledge is deeply processed. Mnemonics are replaced by networks of meaning. It is no longer “This algorithm rhymes with tomato”, but “This algorithm is faster if the data is stored in faster hardware, but our equipment is old so we better use this other algorithm for now”.
Broadly, the progression of learning is: superficial learning, deep learning, and transfer. Check out Visible Learning: The Sequel by John Hattie for more on this.
Edit: To directly answer your question, experts have so many sturdy neural hooks on which to hang new knowledge that mnemonics become less and less necessary. Mnemonics may be particularly helpful when first learning something challenging, but are less necessary as people learn.
You could also check out a paradox called the expert paradox. We used to think memory is boxes that get filled. This idea was directly challenged by Craik and Lockhart’s Levels of Processing. Levels of processing supports the idea that “the more you know, the faster you learn”. Note that this is domain-specific. In other words, an expert in dog training won’t learn quantum mechanics faster than anyone else.