We know Wordlebot's solving algorithm, but do we have any insight how it assigns numeric values to "skill"?
One of my biggest beefs with Wordlebot is not in the algorithm that it uses, but in the quantitative numeric values it assigns to the skill of different guesses. Yesterday was a good example as this one seemed really disproportionate to me.
- "RIVAL", Wordlebot's suggestion here, of course ranked 99, would give 9 groups, the largest having 14 options, but importantly, included a letter "A" which had already been eliminated.
- "CIDER", my guess, ranked 76, gave 8 groups, the largest of 15.
To me, learning about letters (which are included and which are not) is the most critical element of getting to the solution, so repeating a letter I already eliminated does not give me new information, whereas trying a new letter does.
Anyway, given two guesses with nearly equal outcomes in the "group splitting" algorithm (usable by computers) and with one demonstrably worse in the "letter knowledge" algorithm (usable by humans), does anyone understand the calculus by which Wordlebot assigns the conclusion, "Mine is an A+ (as always) while yours is a middling C"? That is to say, what is the math by which Wordle bot landed on a 76 for CIDER here instead of, say, an 89 or a 65 or a 34?