Wordle

[GinandTonic]

It looks like there is almost universal opening strategy, taking few initial guesses with same words with non-repeating letters.
Didn't try hard mode yet, may be more interesting.
Funny that English version seems easier than Russian even for me.

I flicked the hard mode toggle since that was how I played anyway. It's the Americanisms that throw me.

Spoiler eg for today :

THEIR H exists and R is correct
So...
---HR Don't think this exists outside of place names
--H-R Nothing springs to mind, but possibly
-H--R Known to be wrong
H---R Ahh, it's got to be HONOR... But that's only hitting one vowel, anything else? HUMOUR almost but not quite. *Googles how the Imperialist pig-dogs spell humour* 'K, two vowels, that'll do.

 

[Akka]

Yeah, others here have convinced me it is not hard. I think it is the word lists, I was working on a 24,000 word list, but it is only a 2,000 word list for the answers.

I think the millions was for the name, not the copyright of the code.

I don't think it would be significantly harder even with 24000 word lists. The trick is to filter which letters are used and try to score some exact places only as a bonus, and really try to find an answer only once you have narrowed down the choices enough. In fact the game I lost was precisely because I had found three letters at the right place from the get-go, so I became cocky and tried to find the right answer too soon. Ended up with far too many non-greyed letters in the last rows and had to fall back on guessing.

I tend to start every game with "beach", and then depending on how many vowels are found, "tours"/"prism"/"stylo" or something like that. It's a good way to make a wide pass at nearly every vowel and a few common consonants, and from then it massively narrows down the amount of possibilities.

It's funny you made this thread, because I've started playing this game not that long ago because of a community Discord I follow, and nearly every day I play four versions of it (one english and three french). Pretty funny and pretty short, so it's a good way to start the day

 

[Samson]

It's funny you made this thread

Just to be clear, I posted in the random thoughts thread that a multi million dollar bit of software was duplicated in a handfull of lines of bash. When it took off the conversation was moved to its own thread, with my random thought as a start.

 

[Akka]

Just to be clear, I posted in the random thoughts thread that a multi million dollar bit of software was duplicated in a handfull of lines of bash. When it took off the conversation was moved to its own thread, with my random thought as a start.

Ok, let's say it's funny that this subject came to then

 

[tjs282]

I flicked the hard mode toggle since that was how I played anyway. It's the Americanisms that throw me.

Nearly threw me, too.

Spoiler Today's guesses :

ASTER = R correctly placed, A + S + T + E wrong

OK, if no A or E, what's next most likely to precede R? If O, what other vowels and consonants can I try with that? Hmmm...

MINOR = O + R correctly placed, M incorrectly placed, I + N wrong

No I? Then the missing vowel is most likely U. Oh balls, I thought this Josh Wardle guy was British, has he included American (mis)spellings?

Spoiler And the winner is... :

HUMOR

 

[red_elk]

I bet it is todo with the word lists. I think the russian ones are embedded here, but with the words in unicode and my browser just displays codes, not that they would mean anything to me. If you want the word it is here anyway.

I used this one:
https://wordlegame.org/wordle-in-russian
Not sure what you mean by word lists, I just used the ones from my memory

The difference is that English alphabet has fewer letters (26 vs 33) and some of the letters in Russian are relatively rarely used.
So, you are more likely to find only 1-2 letters after first few tries.

 

[Samson]

I used this one:
https://wordlegame.org/wordle-in-russian
Not sure what you mean by word lists, I just used the ones from my memory

The difference is that English alphabet has fewer letters (26 vs 33) and some of the letters in Russian are relatively rarely used.
So, you are more likely to find only 1-2 letters after first few tries.

The english version uses 2 word lists, an answers list (presumably an "everyone knows these words" list) that is 2315 words long and an accepted guesses list which is 10,657 words long, excluding the answers list. I THINK your russian one has a 73,754 long word list. While I cannot say much about how the different strategies will be effected by this, it seems it would make it much harder as you have so much more inherant entropy in the system having so many more potential words.

The character number will also have a big effect.

 

[GinandTonic]

If the list accepts ASTER that would be better than THEIR. Knowing off the bat if the E was behind the R would be useful.

 

[red_elk]

Having the list of words and acceptable guesses, it shouldn't be difficult to write a program which will find list of first guesses which would maximally narrow down the search.
May be I'll do it myself later.
Overall optimal strategy is more tricky though. It should guarantee finding all possible words with minimal number of guesses.

 

[Samson]

Having the list of words and acceptable guesses, it shouldn't be difficult to write a program which will find list of first guesses which would maximally narrow down the search.
May be I'll do it myself later.
Overall optimal strategy is more tricky though. It should guarantee finding all possible words with minimal number of guesses.

I think it is easy, at least for the English word list. If there are 2,000 possible answers and 12,000 possible guesses there are only 24 million states. You just calculate the result of all of them, class them by guess/result, and then build a decision tree.

 

[red_elk]

The tricky part is that "locally optimal" word, which minimizes uncertainty may be not optimal in terms of minimizing number of guesses.
The game is so popular that obviously analysis has been done already:
https://www.poirrier.ca/notes/wordle/

 

[GinandTonic]

I think it is easy, at least for the English word list. If there are 2,000 possible answers and 12,000 possible guesses there are only 24 million states. You just calculate the result of all of them, class them by guess/result, and then build a decision tree.

Brute number crunching doesn't represent the language. If R is the last letter the penultimate letter is disproportionately likely to be E. Hence ASTER would be more useful than THEIR. S would be most common as a final letter, but would reveal nothing as to the rest of the word.

Language as maths (or truth or logic) sends a shiver down my spine from Philosophy A level twenty-something years ago.

 

[Gelion]

I suspect that all you need to do is find 4 words that cover the most common letters and make a guess from there, making every game largely the same?

 

[El_Machinae]

Lots of people will have the same words as their first two guesses, it seems. That takes away some of the fun for me, even if it makes me more likely to 'win'.

 

[red_elk]

I suspect that all you need to do is find 4 words that cover the most common letters and make a guess from there, making every game largely the same?

Yes, but there are nuances

 

[amadeus]

Lots of people will have the same words as their first two guesses, it seems. That takes away some of the fun for me, even if it makes me more likely to 'win'.

Before 1988 contestants on Wheel of Fortune were not given the letters RSTLNE for the bonus round, but since contestants always picked them for their five consonants and vowel they started giving it to them.

 

[Gelion]

Yes, but there are nuances

I know, I know, math

 

[Gori the Grey]

Game should be called "wordl" or "wrdle"

 

[Kaitzilla]

Would the game be easier or harder with 6 letters?

 

[red_elk]

The english version uses 2 word lists, an answers list (presumably an "everyone knows these words" list) that is 2315 words long and an accepted guesses list which is 10,657 words long, excluding the answers list. I THINK your russian one has a 73,754 long word list.

I parsed the file with Russian words, it indeed contains 73755 words, but only 4887 of them are 5-letters long.

Out of curiosity, found largest subgroups of English words which differ by one letter only. These groups are limiting the optimal strategy, according to analysis.
19 ['?ills']
15 ['?ight', '?acks', '?angs', '?ests', '?ines']
14 ['?ales', '?ares', '?arks', '?eals', '?ells', '?inks', '?ocks', '?ooks', 'co?ed']
13 ['?ails', '?ates', '?ears', '?ents', '?ings', '?oles', '?ores', '?owed']

In Russian:

Spoiler :

11 ['?есть']
10 ['?урка']
9 ['?айка', '?аять']
8 ['?кать', '?анка', '?етка', '?ерка', '?учка', 'ко?ка', 'ла?ка', 'ма?ка', 'те?ка']

What's interesting, in English the largest group covers almost all of the alphabet (19/26), while in Russian it's much smaller (11/33)
In both languages the letter which gets "replaced" is most often the first one, followed by the middle one.