Okay so it seems my intuition about intuition might be wrong, but there is still this step:
I’m not sure the order of the words is completely random. As in, even though nobody could predict what the next word could be due to the algorithm being too complex, it’s still completely deterministic based on the original word list, and so far that list hasn’t produced the kinds of output I’d expect from a truly random source.
Again, this is my own intuition, but I’d expect more occasions where “sting” is followed by “stung” or another too-similar word.
I’m not sure if that not-too-similar-ness can be derived from the original word order, but until we know for sure how and why the original word list order is as it is, I don’t think we can assume complete randomness.
Of course, my own beating-the-odds could be completely random, and I just happen to feel like it’s skill and intuition, but there is a lack of real randomness that is bugging me.
If the creator of wordle selected and ordered 2500+ words by hand in any way, they are completely mad. I can assure you there is a 99.999% chance the original word list was taken from a publicly available dictionary, then shufled a few times with typical computerized RNG.
What you are experiencing is a well understood cognitive bias along the lines of the clustering illusion.
Clustering illusion - Wikipedia
What I’m saying isn’t that it was ordered by hand, but maybe something like “random order but no two words in a row beginning with STA”.
I know about the clustering illusion, and based on the 30 or so games I’ve played so far, there is an absence of clustering, which is what I’d expect if was a modifier to the complete randomness.
I’m happy to be wrong about this, it’s just I’ve been relying on my intuition to guide my play and I’m coming out ahead of the curve, and at about as optimal as I could hope for, and I was hoping for a more interesting explanation than “Luke is good at words”.
I did some more checking.
Something I didn’t realize is that the list of valid guesses is much larger than the list of correct answers. For example, you can legally guess something like PRAHU, but that will never be a correct answer.
People are saying, but I can’t find any source, that the 2500~ word list of correct answers was indeed hand-curated by the creator’s girlfriend from the larger list of permitted guesses. There is no known criteria for those selections beyond “words normal people actually know.”
As for ordering, everything suggests the list was just randomly shuffled by a computer.
There really isn’t anything more interesting going on because Wordle itself isn’t that interesting. Most of the interesting stuff is examining the larger problem space that Wordle represents, but the reality of Wordle (a pre-selected relatively short-ish list of words) is much less exciting.
Of all these wordle remixes out there, I wonder what it would look like if we simply had a normal wordle, but harder. Instead of a short list, include all possible valid words, or at least a much longer list. Also, maybe go for 6 letter words and allow more guesses before failure.
The video I shared mentioned the list was hand-curated, so maybe that’s why I made the jump to hand-ordered, or at least not totally random. It probably wouldn’t take much to see if it has any appearance other than totally random, but that’s too much for me to bother with.
There’s also no easy way to do it besides asking the creator. Imagine you had a list of items. You had a person order the list to their liking. Then you had a pseudo random number generator shuffle it (your computer) then you had an actual RNG shuffle it (random.org). It’s entirely possible that they all end up with the same ordering. You can calculate various things about the resulting list, for example how far away it is from being a uniform distribution (alphabetical?), but that’s about it.
The only thing I’m going on that leads me to be so sure it isn’t hand-ordered is that it’s unbelievable that a person would sit there with 2500+ words and hand-order them.