My dream version of Scrabble would have a score system somehow based on word infrequency, so more interesting words would gain higher scores, exactly to mitigate the phenomenon you describe here.
It’s a fun thought but how do you define infrequency and what is an interesting word?
ANESTRI is a very uncommon word by most “normal” measures (it’s the plural of anestrus, a period of sexual dormancy) but it’s (among) the most likely (by probability) 7-letter word played in tournament Scrabble games — there’s something like 9 anagrams of ANESTRI.
Keeping a rolling social history via some sort of merge-friendly approximate frequency data-structure (e.g., count-min sketch - https://en.wikipedia.org/wiki/Count%E2%80%93min_sketch ) would allow you to score the game dynamically based on the play history of the current players, their friends, or others with similar or different characteristics (e.g., "We're playing Brooklyn Scrabble now!").
Yes it’s certainly not a trivial problem. Assuming the requirement for a digital component in order to compute word scores, perhaps frequency could be weighted by previous (competition?) word plays.
That would just reward people who spend years memorizing entire dictionaries. Scrabble is already too skewed in favor of the elite players, IMO.
As paulcole says about chess, "the better player nearly always wins," and that's a problem when the only way to win is to turn the game into an obsession or a career. A big part of Wordle's appeal is that a reasonable vocabulary, and not a savant-level one, is all you need to bring to the table.
