I love this game, but the 'similarity' score is worthless or frequently is even actively detrimental to trying to figure out the word. I think if it used Levenshtein distance instead it would be more helpful. As an example, without really spoiling anything for others, for (Oct 15th, 2025), the word "Food" is given a 46% similarity to the target word. Which is...useless; worse than useless. It has a single letter in a correct position, but that's it! The word length is radically different and it shares no other letters. How is that 46% similar!?
I think a better formula would be something like:
Similarity = 1 - (LevenshteinDistance / TargetWordLength)
For my prior example this would give ~0.14 (14%) which makes way more sense than the current 46%.
I think a better formula would be something like:
Similarity = 1 - (LevenshteinDistance / TargetWordLength)
For my prior example this would give ~0.14 (14%) which makes way more sense than the current 46%.


Sign in