Sure, after I finish playing Wordle I do not have to check WordleBot’s egotistical analysis of my daily attempts but I do anyway. Call it hopeful (maybe WordleBot will praise me) or maybe self-destructive (WordleBot often shames me for not using the same guesses as it does), but I check in with WordleBot about 95% of the time. The only times I don’t check is when I’ve made a very stupid guess.
The NYT WordleBot introduction page has this to say about WordleBot:
“WordleBot is a tool that will take your completed Wordle and analyze it for you. It will give you overall scores for luck and skill on a scale from 0 to 99 and tell you at each turn what, if anything, you could have done differently — if solving Wordles in as few steps as possible is your goal.”
Josh Katz and Matthew Conlen, New York Times
The thing is, WordleBot never loses. But WordleBot doesn’t play the hard version either. It’s always using words that do not have the letters previously uncovered. I always play the hard version.
Another gripe I have with WordleBot is that it’s happy to give me positive reinforcement if I do worse than it, but if I get the correct word in fewer tries than WordleBot it tells me I was very lucky. Not “Congratulations, you beat me!” but more like “Eh, lucky guess.”
Today, for instance, I guessed the correct word in two tries. WordleBot guessed it in three. Here’s what WordleBot had to say:
“You got it! But, with 10 solutions still to choose from, this was a very lucky guess.”
WordleBot July 19, 2022
I am not alone in my criticism of WordleBot. Back in April (before I knew about WordleBot), Christopher Livingston over at PC Gamer wrote an amusing article with an equally amusing title, The official Wordle companion bot is here to tell you how bad you are at Wordle. Also in April, Mashable’s Cecily Moran wrote NYT’s new ‘WordleBot’ will passive-aggressively insult your strategy. Finally, Alice O’Connor, associate editor at Rock Paper Scissors wrote an article about WordleBot called Wordle’s official WordleBot analysis make me feel even more foolish.
Clutch Cargo Lips 
Bot has a extremely reasonable algorithms for determining luck and skill, based on the stated criteria. I don’t know exactly what these are, but when I tap “explain” it usually becomes apparent.
My more intuitive neighbor, with whom I share results, often outscores me by luck when I’ve played [usually slightly] more skillfully, and I do it to Bot occasionally. The usual scenario is when there’s a choice of, say, ten possible answers and one or two are better “deciders”, producing the lowest expected score, but the correct answer is one of the others.
One place I sometimes disagree with Bot is about the likelihood of a word being an answer, where Bot’s algorithm is obscure. It thinks ELATE is very unlikely [I suppose because it’s usually seen in its past-participle-as-adjective form] but that FREED, a past verb formed by adding -[e]d, is as likely as most other than words.
Also, Bot always works by making an exhaustive list of all possible answers and trying them against thousands of possible guesses. On second turn, unless the first guess gave an unusual amount of information, there are dozens or hundreds of possible answers and this method is beyond reasonable human ability; at this stage, I’m looking to pin down positions of CV yellow letters and to test likely as-yet-untried letters.
Overall, I find Bot to be extremely reasonable.
LikeLike
this article is so right on I felt like I wrote it myself. The collective indignations of “the bot” are infuriating, but great inspiration for my efforts to beat it. As was the case today when I got it in two with 45 possibilities remaining. The “bots” response. “Good Job”. By the way. What percent of the time does “bothead “ get it in two? Maybe if it played in the hard mode like about 100% of daily players I would have a bot (sic) of respect. Currently running at %17 twos. In over1000 games. Great article. Hats off to CCL, glad to know others feel similarly.
LikeLike