I had a half-written post on optimizing Wordle guesses with Shannon entropy. Apparently Grant Sanderson had the same idea and presented it better than I could’ve hoped to, so I’ll just link to his video instead. Check it out!
I’ve posted my own code on Gitlab. I’m sure it’s suboptimal in many ways, but I had a lot of fun writing it. “SOARE” seems to be the highest-entropy first guess against the provided list of answers, followed closely by “ROATE” and “RAISE”.