Wordle Lovers - Play the NYT Daily Game

[gmax137]

Wordle 1,345 2/6

 

[edwardII]

Good grief it was the first word that popped into my mind but I quit playing those because there was no logic involved.

Wordle 1,345 5/6

 

[Orodruin]

Wordle 1 346 3/6

 

[OmCheeto]

Wordle 1,346 4/6

 

[jack action]

Wordle 1,346 3/6

[ALERT] 18 left
[SNICK] filter - 1 left
[GLAND]

 

[gmax137]

Wordle 1,346 4/6

SLIME
CLOUD
BLAND
GLAND

 

[edwardII]

Wordle 1,346 3/6

 

[Orodruin]

Wordle 1 347 3/6

 

[OmCheeto]

Wordle 1,347 4/6

 

[jack action]

Wordle 1,347 4/6

[PARSE] 49 left
[TUMID] filter - 1 left: DECOR DEFER DRYER ODDER REEDY
[DECOR] possible answer + filter - 1 left
[DRYER]

I would have it in three with ARISE as a seed word.

 

[Orodruin]

I would have it in three with ARISE as a seed word.

I did

 

[gmax137]

Wordle 1,347 4/6

 

[edwardII]

Wordle 1,347 5/6

 

[Orodruin]

Wordle 1 348 2/6




Use of the sheet has definitely increased my rate of solutions in 2 … In this case ARISE left 7 possible solutions where two would resolve all degeneracies except a single one with two words left. ADORN or AWARD, take your pick!

 

[OmCheeto]

Wordle 1,348 4/6

 

[kuruman]

In case anyone is wondering about my absence, I decided to stop posting my assured solution method to the daily NYT Wordle offering. After doing the whole 2331 words on the limited list twice, once starting with the same seed, SLATE, and then again with the fixed triple seed ORATE LIMNS DUCHY, I gradually lost interest.

Instead, I undertook the more challenging task of exploring what can be done with the extended list of all 5-letter words, assuming random selection, i.e. no discarding already used words. This is a daunting task, and I am slowly chipping away at it. The list has 12972 words. I decided to use a quadruple filter, CARED BEWIG LUMPY KNOTS. I also increased the number of attempts to 7 because it quickly became apparent that a triple filter results in groupings of many candidates with high degeneracy.

My goal now is to explore and see if I can find a path to all 12972 words with no more than 7 guesses and without consideration of previously used words. The thrill of success in the face of failure when the outcome is uncertain is what keeps my interest alive. It is an ambitious but cumbersome process. I have completed 2252 (17.4%) paths up to now. My hope is to complete 1% a day, but that is not always feasible.

The screenshot of my spreadsheet shows the distribution of scores. The minimum possible score with 4 guesses already spent is a 5. Half-integer scores are awarded to 50-50 pairs. Thus, a score of 6.5 means that a 50-50 pair was produced by guess 5.

Note that, so far, there are only 3 scores of 7. The first word on the list is AAHED (as in "I oohed and aahed with wonderment"). Now there are 130 words on the extended list that fit the pattern _ A_ED. The 4-word filter reduced that number to 10. It was possible to find a path for each of the 10 words using the remaining three guesses. Would it be better to use a three-word filter and have 4 guesses left? My guess is probably not because that would increase the number of candidates fitting the pattern.

I will report back from time to time, especially if something monumental or catastrophic happens. The worst case scenario is that, when I am done, I will have to conclude that the filter I chose was inadequate.

 

[jack action]

Wordle 1,348 4/6

[SIREN] 53 left
[QUOTA] filter - 12 left
[WELCH] filter - 1 left
[AWARD]

 

[gmax137]

Wordle 1,348 4/6

SLANT
GRAPE
CHARD
AWARD (I saw BOARD as another possibility)

 

[edwardII]

Wordle 1,348 5/6

 

[Orodruin]

Wordle 1 349 3/6





Super-vanilla scenario. Seed leaves 29 possibilities. Best second guess (BULGE) with EER 0.98 pushes it down to one remaining.

I am starting to doubt the 4th and 5th rows of my sheet will ever be used …

 

[OmCheeto]

Wordle 1,349 3/6

 

[jack action]

Wordle 1,349 3/6

[ARISE] 30 left
[GLUED] filter - 1 left
[LODGE]

 

[gmax137]

Wordle 1,349 3/6

SLANT
GROPE
LODGE

 

[edwardII]

Playing by the seat of my pants and I'm just about threadbare.

Wordle 1,349 5/6

 

[OmCheeto]

Wordle 1 349 3/6





Super-vanilla scenario. Seed leaves 29 possibilities. Best second guess (BULGE) with EER 0.98 pushes it down to one remaining.

I am starting to doubt the 4th and 5th rows of my sheet will ever be used …

One of these days I'm going to figure out how to implement your method. I think I've spent at least 40 hours scratching my head over what exactly you're doing.

In the meanwhile can you tell me if the numbers in the below spreadsheet look right. This will tell me if I'm on the right track.

Game 1,347
Solution: DRYER
Guess 1: PLIER
Guess 2: GOATS
leaving 6 words as possible solutions:

next guess	Average number of words left after play. Lowest score is best choice for next guess
dryer	1.400
defer	1.800
freer	1.800
buyer	2.200
queer	2.200
newer	3.400

 

[Orodruin]

One of these days I'm going to figure out how to implement your method. I think I've spent at least 40 hours scratching my head over what exactly you're doing.

It is quite simple actually. I described it in a post I don't know how long ago ...

I compute the expected entropy remaining (EER) for a possible guess $w$ by checking what pattern each remaining word wold result it if the guess was made. I end up with information about how many ($k$) patterns are possible and how many words $x_i$ would result in each pattern $i$. The remaining entropy if you have $N$ words left is defined as $\ln(N)$ and so the entropy remaining of you have $x_i$ words left is $s_i = \ln(x_i)$. The probability of ending up in bin $i$ is equal to $p_i = x_i/n$, where $n$ is the number of remaining words before the guess*. The EER for a guess is therefore given by$$
EER(w) = \sum_{i = 1}^k p_i s_i = \frac 1n \sum_{i = 1}^k x_i \ln(x_i).
$$
I compute the EER for every remaining word and return a sorted list with the guesses having the lowest EERs on top. I will then pick a guess that has a low EER. I will sometimes forego the lowest EER guess if another word with low EER sounds much more likely and the EER difference is not too high.

The backside of this is that there are way too many words left before making the first guess (computation time in Google sheets becomes prohibitive) so I am using ARISE as a permanent seed. It is the word I have found has the lowest EER currently. This may change over time though and I recheck this every now and then.


* This is assuming all words are equally probable - which they are not, but for the purpose of EER I am not bothering to make any correction for that. It is meant to be used as a tool. If two words get similar EER I will make a judgement call.

 

[OmCheeto]

It is quite simple actually. I described it in a post I don't know how long ago ...

Yes, I've read all those posts. Unfortunately, I don't understand your notation. But it doesn't matter. I'm pretty sure I get the gist of the idea.

Anyways, back to the game:

Wordle 1,350 4/6

 

[jack action]

Wordle 1,350 5/6

[ARISE] 73 left
[LOUSY] filter - 11 left: BUNNY DUMMY DUMPY FUZZY JUMPY MUDDY NUTTY PUDGY PUFFY PUPPY PUTTY
[DEMPT] filter - 2 left: BUNNY FUZZY
[BUNNY] 1 left
[FUZZY]

Will @Orodruin do this one in less than 4?

 

[Orodruin]

Will @Orodruin do this one in less than 4?

Thanks for the reminder ... Had a lot to do today so might have lost that streak otherwise ...

I guess we'll know soon.

 

[Orodruin]

Wordle 1 350 4/6






Nope, but the streak is alive!