Wordle Lovers - Play the NYT Daily Game

[sbrothy]

Wordle 1,198 3/6





I think that separating words as "weird" and "blah" is not a reliable criterion. I looked at the list of used words and I picked the first 5 that I would place in each category according to my sense of usage frequency in spoken English. Of course a thorough job would require one to go through the entire list. I am not inclined to do that at this point.

"Weird" "Blah"
ADOBE ABOUT
AORTA ANGRY
ASKEW AWAKE
BEGET BELOW
BIOME BEING

I'm pretty sure you're right that "blah" is not a valid criterion for anything at all!

EDIT: Oh, you said "reliable". Well, I think my point still stands.

 

[dwarde]

Absolutely no comment. :(
Wordle 1,198 X/6

 

[OmCheeto]

Wordle 1,199 3/6

 

[jack action]

Wordle 1,199 2/6

[CLASH] 8 left - CLIFF CLINK CLOUD CLOUT CLOVE CLUED CLUMP CLUNG
[CLOUD]

If the word selection is purely random, this is luck. But I still don't feel it is.

Looking at the word list, I could see that CLOUD was the most probable answer. CLIFF and CLOVE would have been among my second choices, but the rest of the list doesn't give you that cozy feeling that CLOUD does.

If I was wrong, CLOUD was not filtering everything but I could have made it in 4 at worst. The filter word proposed by the program was UNION which was not better than CLOUD at filtering the list. (If not worst.)

I felt like I had no choice but CLOUD. I didn't even check the other words to see if they were better filters. (I checked it now, they're not.)

 

[kuruman]

Wordle 1,199 3/6

 

[gmax137]

Wordle 1,199 5/6

 

[kuruman]

However, it is my theory that human intervention plays a role in the word selection for NYT Wordle. I'm testing that theory and sharing my results. It seems to give good results up until now and certainly better than always picking the first one at the top of the list.

Sounds like an interesting theory but I think it's a bit flawed because the "human" that presumably makes the word choice is very unlikely to have been faced with the same 2-3 choices that you are faced with when you solve the puzzle. It looks like this "human" is actually you. You make the choice but you shift the credit or blame to some other person for it.

You say you test your theory. That's as it should be but what is your testing protocol? In my mind a good test would be to make a prediction and see it borne out by data collection. You could use your theory to that effect for the NYT Wordle and the other versions. Here is one protocol that I thought of.

Initialization

• Go through the list of candidate words and assign a priori weights of +1, or -1 to each depending on whether they are, respectively, "More likely" or "Less likely" to be the answer according to your criteria of likelihood.
• Initialize the running total (RT) to zero. This will be modified daily as the data collection proceeds.

Data collection

• Use your favorite method of solving the daily puzzle.
• Use filters to bring the number of candidates down to $n\leq 3.$

When $n=2$,

• If (+1, -1), choose the word with +1. If correct add +1 to RT (Running Total); if incorrect add (-1) to RT.
• If (-1, -1) or (+1, +1), choose randomly. If correct add +½ to RT; if incorrect add (-½) to RT.

When $n=3$,

• If (+1, -1,-1), choose the word with +1. If correct add +2 to RT; if incorrect add (-1) to RT.
• If (+1, +1,-1), choose randomly one of the +1 words. If correct add $(+\frac{1}{3})$ to RT; if incorrect add $(\text{-}\frac{2}{3})$ to RT.
• If (-1, -1,-1) or (+1, +1, +1), choose randomly. If correct add +⅓ to RT; if incorrect add (-⅓) to RT.

After a sizable number of daily RT entries the RT should be close to zero when the selection is random (assuming that I figured out the weights correctly.) It will be positive if you have guessed the underlying human intervention mechanism correctly and negative if you have not.

To @sbrothy : If the idea crosses your mind that we are overthinking this, you're probably right $~\dots~$ but it's fun to do so.

 

[jack action]

@kuruman You have no clue how much I wanted to do such a protocol!

I wanted to sort the possible answers in order of importance, giving each of the answers a numerical value. I do not have enough time to do so.

Sounds like an interesting theory but I think it's a bit flawed because the "human" that presumably makes the word choice is very unlikely to have been faced with the same 2-3 choices that you are faced with when you solve the puzzle.

It doesn't matter. No matter which list of words I have, I still rank them overall, comparing them to all possible words. If I have:

• HEAVEN
• HELL
• BANANA

My order is HEAVEN, BANANA, HELL. But if I get:

• HEAVEN
• ANGEL
• BANANA

Then my order is ANGEL, HEAVEN, BANANA. But if I get:

• BRUTE
• CORRUPT
• BANANA

I switch to BANANA, BRUTE, and CORRUPT as my most likely words.

If the chosen word is BANANA, then only the last scenario will work. But if my sorting criteria - even if only in my mind - are similar to the one from the Wordle editor, HEAVEN and ANGEL will already have been chosen and thus never appear on my possible choices for a solution (I track past solutions). Of course, not having a list per se means that as the solutions appear, my sorting order evolves as well. I would love to put that in an algorithm just to understand how my mind works. But time is a factor and I must prioritize the projects I select and how I execute them.

 

[dwarde]

Wordle 1,199 3/6

 

[ChaseLess]

Wordle 1,200 4/6

 

[Orodruin]

But … why?

 

[Orodruin]

Wordle 1 200 4/6

 

[OmCheeto]

But … why?

I didn't even notice that! I refer to that type of thing as 'before my first cup of coffee in the morning' brain. I can totally relate.

Wordle 1,200 4/6

 

[sbrothy]

[...] To @sbrothy : If the idea crosses your mind that we are overthinking this, you're probably right $~\dots~$ but it's fun to do so.

That was exactly my point. Also, it was - to put it mildly - ironic.

I may have a "heurestic" MO but I'm hardly innocent of overthinking things.

EDIT: Heck, I wrote several thousand lines of C++ just to discover that cheating doesn't really pay off.

Fortunately (or disturbingly), most of the code was lying around from my cheating at Scrabble.

EDIT2: I think I already linked to this, but if I need a ticket to the nerd-club I hope this will do

 

[sbrothy]

Wordle 1,200 4/6*

Spoiler

 

[kuruman]

Wordle 1,200 4/6

 

[jack action]

Wordle 1,200 4/6

[CLASH] 228 left
[IRONE] filter - 3 left: DETOX DOPEY MODEM
[DOPEY] filter & most probable answer by a human
[MODEM]

 

[gmax137]

Wordle 1,200 6/6








I used @sbrothy 's seed word. I don't think I will do that again, lol

 

[gmax137]

But … why?

I figured that's why he said

Absolutely no comment. :(

 

[dwarde]

Wordle 1,200 6/6

 

[dwarde]

Wordle 1,200 6/6








I used @sbrothy 's seed word. I don't think I will do that again, lol

That solution pattern looks just like familiar. :)

 

[ChaseLess]

Wordle 1,201 4/6






The beach is calling

 

[Orodruin]

Wordle 1 201 4/6

 

[OmCheeto]

Wordle 1,201 4/6

 

[jack action]

Wordle 1,201 2/6

[CLASH] 2 left: SHELF SHELL
[SHELL] more romantic than SHELF

 

[kuruman]

Wordle 1,201 4/6

 

[gmax137]

Wordle 1,201 4/6

 

[sbrothy]

Wordle 1,201 6/6*








Eh.

 

[dwarde]

Wordle 1,201 5/6

 

[ChaseLess]

Wordle 1,202 5/6

 

[OmCheeto]

Wordle 1,202 4/6

 

[sbrothy]

Wordle 1,202 4/6*

Spoiler

That was pretty dumb. Got there eventually though.

 

[jack action]

Wordle 1,202 3/6

[CLASH] 115 left
[TENOR] filter - 2 left: AXION WAGON
[WAGON] Like, anyone would select AXION ... not!

 

[kuruman]

Wordle 1,202 4/6

 

[gmax137]

Wordle 1,202 5/6


vowel filter (something I don't normally do)

guessed wrong