> 2315 was the original number of Wordle secret words. After the New York Times acquired Wordle, it was revised down to only 2309 secret words.
The NYT puzzle is no longer beholden to any specific list for solutions. There's now a larger list (of 3200 words) that Wordlebot thinks are now likely to be solutions based on NYT word frequency data, but some recent solutions (like KEFIR and LORIS) have come from outside even this set. (The NYT did however add those words into the Wordlebot dictionary just before those puzzles went live.)
> 12972 possible 5-letter words
Likewise, the number of legally guessable words has been increased to 14855.
Loanwords spread from their point of adoption into the wider language and can shift from the original meaning, just like other words.
"Kefir" has definitions in every major dictionary and is often used in contexts where all the other words are also english. That makes it pretty conclusively an English word.
There is no English word for "pleasure derived by someone from another person's misfortune". Yes, we can borrow from German; that doesn't make it suddenly become English.
> For example, we can all agree "schadenfreude" is not an English word, but it's in the dictionary.
I don't agree.
> Yes, we can borrow from German; that doesn't make it suddenly become English.
It doesn't suddenly become English. It happens gradually as more English speakers encounter the word and start using the word.
Dictionaries start adding the word as usage becomes widespread. If a borrowed word is frequently used in an English context by English speakers and is understood by English speakers and that is widespread enoythat it is listed in English dictionaries...it is pretty safe to say that the word is part of the English language.
While I do speak German, I learned the English word "schadenfruede" well before I started studying any German. Similarly, I don't speak any Russian but I know what "kefir" means. I didn't learn it from a Russian, I learned it from other English speakers in the course of speaking English about 20 years ago. In fact, I didn't even know that word originally came from Russian until very recently.
Part of why English is so hard to learn for non-native speakers is because it is full of loan words from lots if languages. Many of them you probably don't even realize originally came from other languages.
Yet, it's on bottles that are sold in the grocery store. Yes, it's not really a common word as I commented at the time, but it's not really a word that I'd consider a proper noun or otherwise not legitimate.
My question was rhetorical. The point is that the English word for "kefir" is "kefir". So English does have a word for that.
shkkmo stated the situation well and nothing you wrote in response refuted anything they said. You say "Just because a word is "in the dictionary" doesn't make it English" but that's not true, as they explained. Virtually the entirety of the English language is "loaned" from somewhere. From the Merriam-Webster definition of schadenfreude: "In English, the word was used mostly by academics until the early 1990s, when it was introduced to more general audiences via pop culture."
You say "that doesn't make it suddenly become English" -- indeed, it wasn't sudden. The word did not always appear in English language dictionaries. It was added when it became recognized as having been incorporated into English--lexicographers are conservative and lag behind actual usage.
Anyway, I'm not going to recreate the entire prescriptive/descriptive debate in linguistics here, especially since "kefir" and "schadenfreude" are English words on both grounds. Enough said.
Not sure why pupal or agora would be removed. Fibre isn't common US spelling though that doesn't really seem to stop Wordle for other purposes. The other three I can see the Times removing them just because they may trigger controversy.
Fun analysis! I run https://squareword.org, which has 5x5 letters featuring words both down and across. Guesses are applied to the horizontal words, but the vertical words can be used as support, a bit like in Sudoku.
I wonder how the analysis would work here - since the verticals form words as well I think you get a lot more information from each guess. Many players can now solve them consistently in 6 or 7 guesses.
I could see it being handy to estimate probability where each word is to better inform each guess. The tricky thing is I'm not sure if each run of wave function collapse (with randomness injected) would be an accurate sampling method for the real distribution of possible permutations. While doing this writeup I tried to find ways to analyze permutations with restriction, but it turns out most general methods are pretty intractable: https://en.wikipedia.org/wiki/Permanent_(mathematics)#Enumer...
I enjoy wordle and wouldn’t automate it to replace my play. I also enjoy techie people demonstrating that you can automate X with Y tool. That’s another form of problem solving. Can’t those enjoyments exist simultaneously?
No you see chatgpt is doing the thinking for us now, instead of using your own brain to struggle with a puzzle that is trivial to solve for a simple algorithm you can let a language model struggle with it for you! You get all of the enjoyment from solving a puzzle with an inefficient overly complicated general purpose solver, without needing to put in any work yourself!
Same as the point in watching dog play with ball. You can also play with ball for enjoyment but sometimes the enjoyment is in seeing the limits of something else.
Is playing a variation of hangman really the limits of an LLM as if so we have vastly over estimated it's usefulness? It isn't a particularly challenging game? for humans, let alone a model that can use data and stats to analyse possibilities and probabilities.
> the list of 2315 valid secret words
> 2315 was the original number of Wordle secret words. After the New York Times acquired Wordle, it was revised down to only 2309 secret words.
The NYT puzzle is no longer beholden to any specific list for solutions. There's now a larger list (of 3200 words) that Wordlebot thinks are now likely to be solutions based on NYT word frequency data, but some recent solutions (like KEFIR and LORIS) have come from outside even this set. (The NYT did however add those words into the Wordlebot dictionary just before those puzzles went live.)
> 12972 possible 5-letter words
Likewise, the number of legally guessable words has been increased to 14855.
Good catch! I didn't know about this, I'll update the footnote. Anecdotally everyone I know was very surprised by KEFIR the other day.
Because it's not a f*ckin' English word!
Is "salsa" an english word?
In english, "salsa" means a spicy dip made from peppers and other vegetables. In spanish, "salsa" just mean a sauce.
Loan words are a well estabished mechanism by which words enter a language from another langues. https://en.m.wikipedia.org/wiki/Loanword
Loanwords spread from their point of adoption into the wider language and can shift from the original meaning, just like other words.
"Kefir" has definitions in every major dictionary and is often used in contexts where all the other words are also english. That makes it pretty conclusively an English word.
Just because a word is "in the dictionary" doesn't make it English.
For example, we can all agree "schadenfreude" is not an English word, but it's in the dictionary.
https://www.merriam-webster.com/dictionary/schadenfreude
There is no English word for "pleasure derived by someone from another person's misfortune". Yes, we can borrow from German; that doesn't make it suddenly become English.
> For example, we can all agree "schadenfreude" is not an English word, but it's in the dictionary.
I don't agree.
> Yes, we can borrow from German; that doesn't make it suddenly become English.
It doesn't suddenly become English. It happens gradually as more English speakers encounter the word and start using the word.
Dictionaries start adding the word as usage becomes widespread. If a borrowed word is frequently used in an English context by English speakers and is understood by English speakers and that is widespread enoythat it is listed in English dictionaries...it is pretty safe to say that the word is part of the English language.
While I do speak German, I learned the English word "schadenfruede" well before I started studying any German. Similarly, I don't speak any Russian but I know what "kefir" means. I didn't learn it from a Russian, I learned it from other English speakers in the course of speaking English about 20 years ago. In fact, I didn't even know that word originally came from Russian until very recently.
Part of why English is so hard to learn for non-native speakers is because it is full of loan words from lots if languages. Many of them you probably don't even realize originally came from other languages.
Yet, it's on bottles that are sold in the grocery store. Yes, it's not really a common word as I commented at the time, but it's not really a word that I'd consider a proper noun or otherwise not legitimate.
So what is "kefir" in English?
English doesn't have a word for that, hence the loanword.
My question was rhetorical. The point is that the English word for "kefir" is "kefir". So English does have a word for that.
shkkmo stated the situation well and nothing you wrote in response refuted anything they said. You say "Just because a word is "in the dictionary" doesn't make it English" but that's not true, as they explained. Virtually the entirety of the English language is "loaned" from somewhere. From the Merriam-Webster definition of schadenfreude: "In English, the word was used mostly by academics until the early 1990s, when it was introduced to more general audiences via pop culture."
You say "that doesn't make it suddenly become English" -- indeed, it wasn't sudden. The word did not always appear in English language dictionaries. It was added when it became recognized as having been incorporated into English--lexicographers are conservative and lag behind actual usage.
Anyway, I'm not going to recreate the entire prescriptive/descriptive debate in linguistics here, especially since "kefir" and "schadenfreude" are English words on both grounds. Enough said.
You can say fuck on the Internet. It's fine
Which 6 words did NYT remove?
You can check the diff on the file wordlist_hidden here: https://github.com/alex1770/wordle/commit/85c4a7e6
agora fibre lynch pupal slave wench
Not sure why pupal or agora would be removed. Fibre isn't common US spelling though that doesn't really seem to stop Wordle for other purposes. The other three I can see the Times removing them just because they may trigger controversy.
Fun analysis! I run https://squareword.org, which has 5x5 letters featuring words both down and across. Guesses are applied to the horizontal words, but the vertical words can be used as support, a bit like in Sudoku.
I wonder how the analysis would work here - since the verticals form words as well I think you get a lot more information from each guess. Many players can now solve them consistently in 6 or 7 guesses.
I wonder if model synthesis (wave function collapse) would be more or less efficient.
https://en.m.wikipedia.org/wiki/Model_synthesis
https://www.rserra.it/solving-hardest-sudoku/
I could see it being handy to estimate probability where each word is to better inform each guess. The tricky thing is I'm not sure if each run of wave function collapse (with randomness injected) would be an accurate sampling method for the real distribution of possible permutations. While doing this writeup I tried to find ways to analyze permutations with restriction, but it turns out most general methods are pretty intractable: https://en.wikipedia.org/wiki/Permanent_(mathematics)#Enumer...
ChatGPT Agent is pretty good at playing Wordle. It's banned from the official version, but it'll play the knock-off ones just fine for you.
What's the point? You play these games to improve your own mind, or for your own enjoyment.
I enjoy wordle and wouldn’t automate it to replace my play. I also enjoy techie people demonstrating that you can automate X with Y tool. That’s another form of problem solving. Can’t those enjoyments exist simultaneously?
No you see chatgpt is doing the thinking for us now, instead of using your own brain to struggle with a puzzle that is trivial to solve for a simple algorithm you can let a language model struggle with it for you! You get all of the enjoyment from solving a puzzle with an inefficient overly complicated general purpose solver, without needing to put in any work yourself!
Same as the point in watching dog play with ball. You can also play with ball for enjoyment but sometimes the enjoyment is in seeing the limits of something else.
Is playing a variation of hangman really the limits of an LLM as if so we have vastly over estimated it's usefulness? It isn't a particularly challenging game? for humans, let alone a model that can use data and stats to analyse possibilities and probabilities.
I watched a cat fall over. It was amusing. This is clearly the limits of cats and we have vastly over estimated their usefulness. I am truly wise.
It's like breeding a dog to have fingers so it can unplug your life support
Enjoyment comes from it autonomously choosing to do so.
Look at little jeepeety, so cute.
It was just a bit of fun to test the model, see what it could do :) But still, more people watch football than play football, right? ;)