The score you see in the solutions is based on the frequency of the letters at various positions (x1000). It is x1000 as some of the frequencies are very small. An article on this is coming soon!
In an attempt to brute-force the solution, I naively decided that the best way was to take this approach. Seeing my efforts, Marton Trencseni got his Python-fu & Data Science on and solved the problem in minutes. I'll link to his blog post here when he's done with it. You can find the playgrounds here, here and here. My weak attempts are here .
If this helped you, a shoutout on Twitter will be great!
You can try the following words: 'brick', 'glent', 'jumpy', 'vozhd', 'waqfs' These 5 words cover 25 out of the 26 alphabets (the remaining one being X) The assistant will give you possible solutions.
Curious about how it works?