science side of tumblr what would happen if you drunk yoohoo chocolate liquid every day.
asking for a friend
DEAR READER

No title available

blake kathryn
Cosmic Funnies
"I'm Dorothy Gale from Kansas"

No title available

JVL

@theartofmadeline
Not today Justin
Stranger Things
Today's Document
Xuebing Du

oozey mess
Lint Roller? I Barely Know Her

Love Begins
KIROKAZE
dirt enthusiast
RMH
Alisa U Zemlji Chuda

Product Placement
seen from North Macedonia

seen from United States
seen from United States
seen from North Macedonia
seen from North Macedonia
seen from North Macedonia
seen from North Macedonia
seen from North Macedonia
seen from North Macedonia
seen from North Macedonia

seen from Singapore
seen from North Macedonia

seen from United States

seen from Australia
seen from Australia

seen from North Macedonia
seen from United States

seen from Malaysia

seen from India

seen from Singapore
@jdkm4
science side of tumblr what would happen if you drunk yoohoo chocolate liquid every day.
asking for a friend
I am devastated that André De Shields is leaving Hadestown as if I had even the slightest of chances (or money) to ever go to New York and see him perform.
"I'm not gonna be able to see him live :("
and you wouldn't have even if he stayed. you are broke.
i’m planning on seeing it this summer but... 😔😭
so i don't really know what to do with this blog but like a month ago i optimized wordle so i'll write about that
so the best first word in wordle is ROATE which according to dictionary.com means "the cumulative net earnings after taxes available to common shareholders, adjusted for tax-affected amortization of intangibles, for the calendar quarters in each calendar year."
i calculated this using python (my beloved). you want to eliminate the most words on average in your guess, so that's what i'm optimizing.
before i go deeper i want to clear up some terminology
a guess is a word that you can type in to wordle. there are more than 12,000 possible guesses you can make in the game.
an answer is a word that can actually be the answer, the word that you are looking for. there are only 2315 possible answers in the game.
a result is some collection of greys, yellows, and greens. when you type in a guess you get a result
by the way all the wordlists that i used in these calculations are the actual wordlists that they use in the wordle game
so, basically my program takes a word that you can guess and sees what the result would be if you guessed that word for every single possible answer (of which there are 2315). in this way, each possible answer is assigned a result.
so your goal is for these answers to be as spread out as possible between the results. if 99% of the time your guess gets all greys, than you're revealing very little information. ideally every possible answer is spread out perfectly between the different possible results. there are 243 possible results in wordle (3^5), and 2315/243 is around 9.5, so in your ideal world no matter what happens after your first guess you have narrowed it down to 9.5 different words.
we of course don't live in this ideal world, so we have to do with ROATE. how do i calculate how good a word is? take the number of answers assigned to a result, divide by the total number of answers (2315), and square that. repeat for every result and sum all the numbers you get.
let's say there were 3 possible results and 6 possible answers. a word that had the answers split evenly would get:
(2/6)^2 + (2/6)^2 + (2/6)^2 = 0.33
meanwhile, a word that had most of the answers assigned to only having one result would look more like:
(5/6)^2 + (1/6)^2 + (0/6)^2 = 0.72
lower numbers are better, with 1 being the worst possible score. these numbers actually tell you the average amount of elimination for the words. so the first example with 0.33 on average reduces the possible answers to 33% of it's original size (obviously), from 6 to 2. less obviously, the second example on average reduces the answers to 72% of it's original size, from 6 to 4.32.
SO, i did this calculation for each word that you can guess, and the best was ROATE. ROATE's score was 0.0262. this means that when you guess ROATE, on average the possible answers are cut to 2.62% of the original possible answers, which means on average after getting ROATE you are narrowed down to 60 words. that's pretty good for just one guess. (as I said earlier, the theoretical best would be 9.5). a very rough estimate tells me that a second guess would probably narrow you down to three answers, so with ROATE (if you play tactfully) you will hopefully usually get the wordle in three or four.
one other thing to keep in mind is that although ROATE is a valid guess, it isn't a valid answer. would a valid answer (such as the word RAISE) be better? i did the math and it is very close but ROATE is still better either way.
thanks for reading, i'm not even done because i calculated the best possible second words as well, ill talk about that later
wordle stuff: part 2
so, ROATE is the best first guess in wordle. the question is, what's the best second guess?
well, of course, it depends on what happens after your first guess. if ROATE gets all grey than you're gonna want to try new letters, but if you get some yellows you might want to keep them in your next guess to try and figure out where they are.
the calculations for this are actually pretty simple. I just used the same formula I had last time but switched it so that the possible answers was limited to the possible answers for the given result. so first it looks at the possible words if you got all greys from ROATE, and then it recalculates with this much smaller word list. and on and on for each result.
we do have problems though. let say you have narrowed your words down to STINK and STING. the obvious next guess is either one of these. if you're right, great. if you're wrong, you know what it is for your next guess.
but my program doesn't think like that. my program is designed to narrow down your results as much as possible, not to actually get the answer. so my program would just tell you to do a random word with a G or a K, perhaps a word like YIKES. if the K is yellow than it's stink, if it's grey it's sting.
YIKES is what I was thinking when i figured this out. fortunately I coded in something to account for this. it was annoyingly difficult, but now it will sometimes actually tells you to guess a possible answer instead of a good-at-narrowing-it-down answer.
so, these are the answers. I wrote it down for the ones with a greater than 1% chance of happening, if it's less than that it's really not that useful and also i have a slight worry that the code might not even work at that point😅
I DO NOT USE THIS CHART WHILE PLAYING WORDLE. it's cheating. what I DO do is memorize the words beforehand, so when I'm actually playing the game I can still play perfectly without any guilt. I recommend anyone reading this to do that as well.
a few comments: there was some surprising stuff here. when I'd played wordle in the past, whenever I got a yellow I moved it around and tried it somewhere else. but the program is lax on that here. it happens sometimes, but not always, and sometimes only for some of the letters. I guess sometimes guessing new letters is more important than moving old ones around.
another thing that surprised me was that the program actually repeated letters in some of the guesses. usually I think that repeating letters is bad because it's wasteful, but in some cases it actually makes sense. if you have a yellow E and want to know where it is, LENES will find it faster than LINES. also, of course, it's useful if your word actually does repeat a letter.
another thing that surprised, no DELIGHTED me, was that after a yellow R and green A the best word is DICKS. hehehee
okay. I did not particularly have time on my hands, but i still decided this was a good idea, I guess.
I played today’s wordle using ROATE, and admittedly I already had played it so it wasn’t completely fair, but the wordlebot analysis of the starting word is still accurate.
spoilers for Wordle #326 under the cut:
omg, the wordle bot! i don’t have a nyt subscription so ive never been able to use it, but this is really cool. nice to get at least some validation for my math heh.
again, spoilers for wordle 326:
since roate narrows 🟨⬛️🟨⬛️🟩 down to 10 it was so unlikely i didnt include it in my chart (also the math gets dicier with lower numbers). hopefully tomorrows is better.
Regardless, i got this one in 3, so im happy
you don’t have to have a subscription to use the bot! you do need an account but it’s free, i completely get it if you don’t want to have to sign up for one though. it’s kind of fun to see what it makes of my games. might start using ROATE, just to see what it does- mine is STEAM, but it’s more sentimental than anything else.
oh wow, I don't know why i thought that. THIS BOT IS SO COOL it's very nice to see validation in all the calculations that I came up with on my own. honestly idk why they use CRANE because they say themselves that ROATE on average narrows it down more. perhaps because CRANE sounds like a word and people would get angry if it was ROATE? idk. anyways thank you for showing me this lol.
oh I know why it’s CRANE! it’s because they don’t only consider the number of solutions eliminated, they’re also considering what “groups” the results will give.
(spoilers again)
ok... i need to think about this
the screenshots you had showed BLADE and not CARVE so i’m slightly confused but i think i see what you’re saying?
average group size sounds pretty good... but in some instances i don’t think it makes sense.
average group size doesn’t work because larger groups are more likely to be the correct group.
let’s say you had 15 words left and two possible guesses. one guess would split those 15 into 5 groups: one group with 11 words and the other 4 groups with 1 word. the OTHER guess would split those 15 words into 3 groups, each with 5 words.
the average group size for the first guess is (11 + 1 + 1 + 1 + 1) / 5 = 3, but for the second guess it is (5 + 5 + 5) / 3 = 5. but the second guess is clearly better. That’s why I did a weighted average for my calculations.
My calculations do (11/15)^2 + (1/15)^2 + (1/15)^2 + (1/15)^2 + (1/15)^2) = 0.556, with the second guess being (5/15)^2 + (5/15)^2 + (5/15)^2 = 0.333
this is just weighting it. 11/15 is far more likely to be the correct group than one of the 1/15s, so it should be given more weight.
maybe I didn’t understand what you said but I feel like average group size isn’t as good of a metric? if I’m wrong please tell me I am not fully sure about any of this.
about hard mode-- that’s really cool and maybe i’ll try to code something for that that next, that could be very fun.
hear me out guys. ny times crossword should be free. i love crosswords, i do the la times one every day. but i refuse to subject myself to a monthly payment to the new york times just for the crossword when i could just do a free one. that's my precious caffeine money, i cant afford to waste it
yeah that’s why they bought wordle they’re want EVERYONE to have a nyt subscription by taking over a bunch of random stuff like that. one of the few remaining successful newspapers other than like the washington post and WSJ. they just chose subscription over ads and it’s been (shockingly) successful for them.
also vox, probably my favourite news site, has a crossword so there’s that.
so i don't really know what to do with this blog but like a month ago i optimized wordle so i'll write about that
so the best first word in wordle is ROATE which according to dictionary.com means "the cumulative net earnings after taxes available to common shareholders, adjusted for tax-affected amortization of intangibles, for the calendar quarters in each calendar year."
i calculated this using python (my beloved). you want to eliminate the most words on average in your guess, so that's what i'm optimizing.
before i go deeper i want to clear up some terminology
a guess is a word that you can type in to wordle. there are more than 12,000 possible guesses you can make in the game.
an answer is a word that can actually be the answer, the word that you are looking for. there are only 2315 possible answers in the game.
a result is some collection of greys, yellows, and greens. when you type in a guess you get a result
by the way all the wordlists that i used in these calculations are the actual wordlists that they use in the wordle game
so, basically my program takes a word that you can guess and sees what the result would be if you guessed that word for every single possible answer (of which there are 2315). in this way, each possible answer is assigned a result.
so your goal is for these answers to be as spread out as possible between the results. if 99% of the time your guess gets all greys, than you're revealing very little information. ideally every possible answer is spread out perfectly between the different possible results. there are 243 possible results in wordle (3^5), and 2315/243 is around 9.5, so in your ideal world no matter what happens after your first guess you have narrowed it down to 9.5 different words.
we of course don't live in this ideal world, so we have to do with ROATE. how do i calculate how good a word is? take the number of answers assigned to a result, divide by the total number of answers (2315), and square that. repeat for every result and sum all the numbers you get.
let's say there were 3 possible results and 6 possible answers. a word that had the answers split evenly would get:
(2/6)^2 + (2/6)^2 + (2/6)^2 = 0.33
meanwhile, a word that had most of the answers assigned to only having one result would look more like:
(5/6)^2 + (1/6)^2 + (0/6)^2 = 0.72
lower numbers are better, with 1 being the worst possible score. these numbers actually tell you the average amount of elimination for the words. so the first example with 0.33 on average reduces the possible answers to 33% of it's original size (obviously), from 6 to 2. less obviously, the second example on average reduces the answers to 72% of it's original size, from 6 to 4.32.
SO, i did this calculation for each word that you can guess, and the best was ROATE. ROATE's score was 0.0262. this means that when you guess ROATE, on average the possible answers are cut to 2.62% of the original possible answers, which means on average after getting ROATE you are narrowed down to 60 words. that's pretty good for just one guess. (as I said earlier, the theoretical best would be 9.5). a very rough estimate tells me that a second guess would probably narrow you down to three answers, so with ROATE (if you play tactfully) you will hopefully usually get the wordle in three or four.
one other thing to keep in mind is that although ROATE is a valid guess, it isn't a valid answer. would a valid answer (such as the word RAISE) be better? i did the math and it is very close but ROATE is still better either way.
thanks for reading, i'm not even done because i calculated the best possible second words as well, ill talk about that later
wordle stuff: part 2
so, ROATE is the best first guess in wordle. the question is, what's the best second guess?
well, of course, it depends on what happens after your first guess. if ROATE gets all grey than you're gonna want to try new letters, but if you get some yellows you might want to keep them in your next guess to try and figure out where they are.
the calculations for this are actually pretty simple. I just used the same formula I had last time but switched it so that the possible answers was limited to the possible answers for the given result. so first it looks at the possible words if you got all greys from ROATE, and then it recalculates with this much smaller word list. and on and on for each result.
we do have problems though. let say you have narrowed your words down to STINK and STING. the obvious next guess is either one of these. if you're right, great. if you're wrong, you know what it is for your next guess.
but my program doesn't think like that. my program is designed to narrow down your results as much as possible, not to actually get the answer. so my program would just tell you to do a random word with a G or a K, perhaps a word like YIKES. if the K is yellow than it's stink, if it's grey it's sting.
YIKES is what I was thinking when i figured this out. fortunately I coded in something to account for this. it was annoyingly difficult, but now it will sometimes actually tells you to guess a possible answer instead of a good-at-narrowing-it-down answer.
so, these are the answers. I wrote it down for the ones with a greater than 1% chance of happening, if it's less than that it's really not that useful and also i have a slight worry that the code might not even work at that point😅
I DO NOT USE THIS CHART WHILE PLAYING WORDLE. it's cheating. what I DO do is memorize the words beforehand, so when I'm actually playing the game I can still play perfectly without any guilt. I recommend anyone reading this to do that as well.
a few comments: there was some surprising stuff here. when I'd played wordle in the past, whenever I got a yellow I moved it around and tried it somewhere else. but the program is lax on that here. it happens sometimes, but not always, and sometimes only for some of the letters. I guess sometimes guessing new letters is more important than moving old ones around.
another thing that surprised me was that the program actually repeated letters in some of the guesses. usually I think that repeating letters is bad because it's wasteful, but in some cases it actually makes sense. if you have a yellow E and want to know where it is, LENES will find it faster than LINES. also, of course, it's useful if your word actually does repeat a letter.
another thing that surprised, no DELIGHTED me, was that after a yellow R and green A the best word is DICKS. hehehee
ALL the best second guesses
on further consideration (and checking of my calculations) I'm going to put all of the best second guesses for ROATE here, not just the top 1%. i don't really want this to become something that just tells you exactly what to do in every situation because then what's even the point? but i'll put them here just in case someone wants them.
https://pastebin.com/0wcxdKKs
so i don't really know what to do with this blog but like a month ago i optimized wordle so i'll write about that
so the best first word in wordle is ROATE which according to dictionary.com means "the cumulative net earnings after taxes available to common shareholders, adjusted for tax-affected amortization of intangibles, for the calendar quarters in each calendar year."
i calculated this using python (my beloved). you want to eliminate the most words on average in your guess, so that's what i'm optimizing.
before i go deeper i want to clear up some terminology
a guess is a word that you can type in to wordle. there are more than 12,000 possible guesses you can make in the game.
an answer is a word that can actually be the answer, the word that you are looking for. there are only 2315 possible answers in the game.
a result is some collection of greys, yellows, and greens. when you type in a guess you get a result
by the way all the wordlists that i used in these calculations are the actual wordlists that they use in the wordle game
so, basically my program takes a word that you can guess and sees what the result would be if you guessed that word for every single possible answer (of which there are 2315). in this way, each possible answer is assigned a result.
so your goal is for these answers to be as spread out as possible between the results. if 99% of the time your guess gets all greys, than you're revealing very little information. ideally every possible answer is spread out perfectly between the different possible results. there are 243 possible results in wordle (3^5), and 2315/243 is around 9.5, so in your ideal world no matter what happens after your first guess you have narrowed it down to 9.5 different words.
we of course don't live in this ideal world, so we have to do with ROATE. how do i calculate how good a word is? take the number of answers assigned to a result, divide by the total number of answers (2315), and square that. repeat for every result and sum all the numbers you get.
let's say there were 3 possible results and 6 possible answers. a word that had the answers split evenly would get:
(2/6)^2 + (2/6)^2 + (2/6)^2 = 0.33
meanwhile, a word that had most of the answers assigned to only having one result would look more like:
(5/6)^2 + (1/6)^2 + (0/6)^2 = 0.72
lower numbers are better, with 1 being the worst possible score. these numbers actually tell you the average amount of elimination for the words. so the first example with 0.33 on average reduces the possible answers to 33% of it's original size (obviously), from 6 to 2. less obviously, the second example on average reduces the answers to 72% of it's original size, from 6 to 4.32.
SO, i did this calculation for each word that you can guess, and the best was ROATE. ROATE's score was 0.0262. this means that when you guess ROATE, on average the possible answers are cut to 2.62% of the original possible answers, which means on average after getting ROATE you are narrowed down to 60 words. that's pretty good for just one guess. (as I said earlier, the theoretical best would be 9.5). a very rough estimate tells me that a second guess would probably narrow you down to three answers, so with ROATE (if you play tactfully) you will hopefully usually get the wordle in three or four.
one other thing to keep in mind is that although ROATE is a valid guess, it isn't a valid answer. would a valid answer (such as the word RAISE) be better? i did the math and it is very close but ROATE is still better either way.
thanks for reading, i'm not even done because i calculated the best possible second words as well, ill talk about that later
wordle stuff: part 2
so, ROATE is the best first guess in wordle. the question is, what's the best second guess?
well, of course, it depends on what happens after your first guess. if ROATE gets all grey than you're gonna want to try new letters, but if you get some yellows you might want to keep them in your next guess to try and figure out where they are.
the calculations for this are actually pretty simple. I just used the same formula I had last time but switched it so that the possible answers was limited to the possible answers for the given result. so first it looks at the possible words if you got all greys from ROATE, and then it recalculates with this much smaller word list. and on and on for each result.
we do have problems though. let say you have narrowed your words down to STINK and STING. the obvious next guess is either one of these. if you're right, great. if you're wrong, you know what it is for your next guess.
but my program doesn't think like that. my program is designed to narrow down your results as much as possible, not to actually get the answer. so my program would just tell you to do a random word with a G or a K, perhaps a word like YIKES. if the K is yellow than it's stink, if it's grey it's sting.
YIKES is what I was thinking when i figured this out. fortunately I coded in something to account for this. it was annoyingly difficult, but now it will sometimes actually tells you to guess a possible answer instead of a good-at-narrowing-it-down answer.
so, these are the answers. I wrote it down for the ones with a greater than 1% chance of happening, if it's less than that it's really not that useful and also i have a slight worry that the code might not even work at that point😅
I DO NOT USE THIS CHART WHILE PLAYING WORDLE. it's cheating. what I DO do is memorize the words beforehand, so when I'm actually playing the game I can still play perfectly without any guilt. I recommend anyone reading this to do that as well.
a few comments: there was some surprising stuff here. when I'd played wordle in the past, whenever I got a yellow I moved it around and tried it somewhere else. but the program is lax on that here. it happens sometimes, but not always, and sometimes only for some of the letters. I guess sometimes guessing new letters is more important than moving old ones around.
another thing that surprised me was that the program actually repeated letters in some of the guesses. usually I think that repeating letters is bad because it's wasteful, but in some cases it actually makes sense. if you have a yellow E and want to know where it is, LENES will find it faster than LINES. also, of course, it's useful if your word actually does repeat a letter.
another thing that surprised, no DELIGHTED me, was that after a yellow R and green A the best word is DICKS. hehehee
okay. I did not particularly have time on my hands, but i still decided this was a good idea, I guess.
I played today’s wordle using ROATE, and admittedly I already had played it so it wasn’t completely fair, but the wordlebot analysis of the starting word is still accurate.
spoilers for Wordle #326 under the cut:
omg, the wordle bot! i don’t have a nyt subscription so ive never been able to use it, but this is really cool. nice to get at least some validation for my math heh.
again, spoilers for wordle 326:
since roate narrows 🟨⬛️🟨⬛️🟩 down to 10 it was so unlikely i didnt include it in my chart (also the math gets dicier with lower numbers). hopefully tomorrows is better.
Regardless, i got this one in 3, so im happy
you don’t have to have a subscription to use the bot! you do need an account but it’s free, i completely get it if you don’t want to have to sign up for one though. it’s kind of fun to see what it makes of my games. might start using ROATE, just to see what it does- mine is STEAM, but it’s more sentimental than anything else.
oh wow, I don't know why i thought that. THIS BOT IS SO COOL it's very nice to see validation in all the calculations that I came up with on my own. honestly idk why they use CRANE because they say themselves that ROATE on average narrows it down more. perhaps because CRANE sounds like a word and people would get angry if it was ROATE? idk. anyways thank you for showing me this lol.
so i don't really know what to do with this blog but like a month ago i optimized wordle so i'll write about that
so the best first word in wordle is ROATE which according to dictionary.com means "the cumulative net earnings after taxes available to common shareholders, adjusted for tax-affected amortization of intangibles, for the calendar quarters in each calendar year."
i calculated this using python (my beloved). you want to eliminate the most words on average in your guess, so that's what i'm optimizing.
before i go deeper i want to clear up some terminology
a guess is a word that you can type in to wordle. there are more than 12,000 possible guesses you can make in the game.
an answer is a word that can actually be the answer, the word that you are looking for. there are only 2315 possible answers in the game.
a result is some collection of greys, yellows, and greens. when you type in a guess you get a result
by the way all the wordlists that i used in these calculations are the actual wordlists that they use in the wordle game
so, basically my program takes a word that you can guess and sees what the result would be if you guessed that word for every single possible answer (of which there are 2315). in this way, each possible answer is assigned a result.
so your goal is for these answers to be as spread out as possible between the results. if 99% of the time your guess gets all greys, than you're revealing very little information. ideally every possible answer is spread out perfectly between the different possible results. there are 243 possible results in wordle (3^5), and 2315/243 is around 9.5, so in your ideal world no matter what happens after your first guess you have narrowed it down to 9.5 different words.
we of course don't live in this ideal world, so we have to do with ROATE. how do i calculate how good a word is? take the number of answers assigned to a result, divide by the total number of answers (2315), and square that. repeat for every result and sum all the numbers you get.
let's say there were 3 possible results and 6 possible answers. a word that had the answers split evenly would get:
(2/6)^2 + (2/6)^2 + (2/6)^2 = 0.33
meanwhile, a word that had most of the answers assigned to only having one result would look more like:
(5/6)^2 + (1/6)^2 + (0/6)^2 = 0.72
lower numbers are better, with 1 being the worst possible score. these numbers actually tell you the average amount of elimination for the words. so the first example with 0.33 on average reduces the possible answers to 33% of it's original size (obviously), from 6 to 2. less obviously, the second example on average reduces the answers to 72% of it's original size, from 6 to 4.32.
SO, i did this calculation for each word that you can guess, and the best was ROATE. ROATE's score was 0.0262. this means that when you guess ROATE, on average the possible answers are cut to 2.62% of the original possible answers, which means on average after getting ROATE you are narrowed down to 60 words. that's pretty good for just one guess. (as I said earlier, the theoretical best would be 9.5). a very rough estimate tells me that a second guess would probably narrow you down to three answers, so with ROATE (if you play tactfully) you will hopefully usually get the wordle in three or four.
one other thing to keep in mind is that although ROATE is a valid guess, it isn't a valid answer. would a valid answer (such as the word RAISE) be better? i did the math and it is very close but ROATE is still better either way.
thanks for reading, i'm not even done because i calculated the best possible second words as well, ill talk about that later
wordle stuff: part 2
so, ROATE is the best first guess in wordle. the question is, what's the best second guess?
well, of course, it depends on what happens after your first guess. if ROATE gets all grey than you're gonna want to try new letters, but if you get some yellows you might want to keep them in your next guess to try and figure out where they are.
the calculations for this are actually pretty simple. I just used the same formula I had last time but switched it so that the possible answers was limited to the possible answers for the given result. so first it looks at the possible words if you got all greys from ROATE, and then it recalculates with this much smaller word list. and on and on for each result.
we do have problems though. let say you have narrowed your words down to STINK and STING. the obvious next guess is either one of these. if you're right, great. if you're wrong, you know what it is for your next guess.
but my program doesn't think like that. my program is designed to narrow down your results as much as possible, not to actually get the answer. so my program would just tell you to do a random word with a G or a K, perhaps a word like YIKES. if the K is yellow than it's stink, if it's grey it's sting.
YIKES is what I was thinking when i figured this out. fortunately I coded in something to account for this. it was annoyingly difficult, but now it will sometimes actually tells you to guess a possible answer instead of a good-at-narrowing-it-down answer.
so, these are the answers. I wrote it down for the ones with a greater than 1% chance of happening, if it's less than that it's really not that useful and also i have a slight worry that the code might not even work at that point😅
I DO NOT USE THIS CHART WHILE PLAYING WORDLE. it's cheating. what I DO do is memorize the words beforehand, so when I'm actually playing the game I can still play perfectly without any guilt. I recommend anyone reading this to do that as well.
a few comments: there was some surprising stuff here. when I'd played wordle in the past, whenever I got a yellow I moved it around and tried it somewhere else. but the program is lax on that here. it happens sometimes, but not always, and sometimes only for some of the letters. I guess sometimes guessing new letters is more important than moving old ones around.
another thing that surprised me was that the program actually repeated letters in some of the guesses. usually I think that repeating letters is bad because it's wasteful, but in some cases it actually makes sense. if you have a yellow E and want to know where it is, LENES will find it faster than LINES. also, of course, it's useful if your word actually does repeat a letter.
another thing that surprised, no DELIGHTED me, was that after a yellow R and green A the best word is DICKS. hehehee
okay. I did not particularly have time on my hands, but i still decided this was a good idea, I guess.
I played today’s wordle using ROATE, and admittedly I already had played it so it wasn’t completely fair, but the wordlebot analysis of the starting word is still accurate.
spoilers for Wordle #326 under the cut:
omg, the wordle bot! i don’t have a nyt subscription so ive never been able to use it, but this is really cool. nice to get at least some validation for my math heh.
again, spoilers for wordle 326:
since roate narrows 🟨⬛️🟨⬛️🟩 down to 10 it was so unlikely i didnt include it in my chart (also the math gets dicier with lower numbers). hopefully tomorrows is better.
Regardless, i got this one in 3, so im happy
so i don't really know what to do with this blog but like a month ago i optimized wordle so i'll write about that
so the best first word in wordle is ROATE which according to dictionary.com means "the cumulative net earnings after taxes available to common shareholders, adjusted for tax-affected amortization of intangibles, for the calendar quarters in each calendar year."
i calculated this using python (my beloved). you want to eliminate the most words on average in your guess, so that's what i'm optimizing.
before i go deeper i want to clear up some terminology
a guess is a word that you can type in to wordle. there are more than 12,000 possible guesses you can make in the game.
an answer is a word that can actually be the answer, the word that you are looking for. there are only 2315 possible answers in the game.
a result is some collection of greys, yellows, and greens. when you type in a guess you get a result
by the way all the wordlists that i used in these calculations are the actual wordlists that they use in the wordle game
so, basically my program takes a word that you can guess and sees what the result would be if you guessed that word for every single possible answer (of which there are 2315). in this way, each possible answer is assigned a result.
so your goal is for these answers to be as spread out as possible between the results. if 99% of the time your guess gets all greys, than you're revealing very little information. ideally every possible answer is spread out perfectly between the different possible results. there are 243 possible results in wordle (3^5), and 2315/243 is around 9.5, so in your ideal world no matter what happens after your first guess you have narrowed it down to 9.5 different words.
we of course don't live in this ideal world, so we have to do with ROATE. how do i calculate how good a word is? take the number of answers assigned to a result, divide by the total number of answers (2315), and square that. repeat for every result and sum all the numbers you get.
let's say there were 3 possible results and 6 possible answers. a word that had the answers split evenly would get:
(2/6)^2 + (2/6)^2 + (2/6)^2 = 0.33
meanwhile, a word that had most of the answers assigned to only having one result would look more like:
(5/6)^2 + (1/6)^2 + (0/6)^2 = 0.72
lower numbers are better, with 1 being the worst possible score. these numbers actually tell you the average amount of elimination for the words. so the first example with 0.33 on average reduces the possible answers to 33% of it's original size (obviously), from 6 to 2. less obviously, the second example on average reduces the answers to 72% of it's original size, from 6 to 4.32.
SO, i did this calculation for each word that you can guess, and the best was ROATE. ROATE's score was 0.0262. this means that when you guess ROATE, on average the possible answers are cut to 2.62% of the original possible answers, which means on average after getting ROATE you are narrowed down to 60 words. that's pretty good for just one guess. (as I said earlier, the theoretical best would be 9.5). a very rough estimate tells me that a second guess would probably narrow you down to three answers, so with ROATE (if you play tactfully) you will hopefully usually get the wordle in three or four.
one other thing to keep in mind is that although ROATE is a valid guess, it isn't a valid answer. would a valid answer (such as the word RAISE) be better? i did the math and it is very close but ROATE is still better either way.
thanks for reading, i'm not even done because i calculated the best possible second words as well, ill talk about that later
wordle stuff: part 2
so, ROATE is the best first guess in wordle. the question is, what's the best second guess?
well, of course, it depends on what happens after your first guess. if ROATE gets all grey than you're gonna want to try new letters, but if you get some yellows you might want to keep them in your next guess to try and figure out where they are.
the calculations for this are actually pretty simple. I just used the same formula I had last time but switched it so that the possible answers was limited to the possible answers for the given result. so first it looks at the possible words if you got all greys from ROATE, and then it recalculates with this much smaller word list. and on and on for each result.
we do have problems though. let say you have narrowed your words down to STINK and STING. the obvious next guess is either one of these. if you're right, great. if you're wrong, you know what it is for your next guess.
but my program doesn't think like that. my program is designed to narrow down your results as much as possible, not to actually get the answer. so my program would just tell you to do a random word with a G or a K, perhaps a word like YIKES. if the K is yellow than it's stink, if it's grey it's sting.
YIKES is what I was thinking when i figured this out. fortunately I coded in something to account for this. it was annoyingly difficult, but now it will sometimes actually tells you to guess a possible answer instead of a good-at-narrowing-it-down answer.
so, these are the answers. I wrote it down for the ones with a greater than 1% chance of happening, if it's less than that it's really not that useful and also i have a slight worry that the code might not even work at that point😅
I DO NOT USE THIS CHART WHILE PLAYING WORDLE. it's cheating. what I DO do is memorize the words beforehand, so when I'm actually playing the game I can still play perfectly without any guilt. I recommend anyone reading this to do that as well.
a few comments: there was some surprising stuff here. when I'd played wordle in the past, whenever I got a yellow I moved it around and tried it somewhere else. but the program is lax on that here. it happens sometimes, but not always, and sometimes only for some of the letters. I guess sometimes guessing new letters is more important than moving old ones around.
another thing that surprised me was that the program actually repeated letters in some of the guesses. usually I think that repeating letters is bad because it's wasteful, but in some cases it actually makes sense. if you have a yellow E and want to know where it is, LENES will find it faster than LINES. also, of course, it's useful if your word actually does repeat a letter.
another thing that surprised, no DELIGHTED me, was that after a yellow R and green A the best word is DICKS. hehehee
so i don't really know what to do with this blog but like a month ago i optimized wordle so i'll write about that
so the best first word in wordle is ROATE which according to dictionary.com means "the cumulative net earnings after taxes available to common shareholders, adjusted for tax-affected amortization of intangibles, for the calendar quarters in each calendar year."
i calculated this using python (my beloved). you want to eliminate the most words on average in your guess, so that's what i'm optimizing.
before i go deeper i want to clear up some terminology
a guess is a word that you can type in to wordle. there are more than 12,000 possible guesses you can make in the game.
an answer is a word that can actually be the answer, the word that you are looking for. there are only 2315 possible answers in the game.
a result is some collection of greys, yellows, and greens. when you type in a guess you get a result
by the way all the wordlists that i used in these calculations are the actual wordlists that they use in the wordle game
so, basically my program takes a word that you can guess and sees what the result would be if you guessed that word for every single possible answer (of which there are 2315). in this way, each possible answer is assigned a result.
so your goal is for these answers to be as spread out as possible between the results. if 99% of the time your guess gets all greys, than you're revealing very little information. ideally every possible answer is spread out perfectly between the different possible results. there are 243 possible results in wordle (3^5), and 2315/243 is around 9.5, so in your ideal world no matter what happens after your first guess you have narrowed it down to 9.5 different words.
we of course don't live in this ideal world, so we have to do with ROATE. how do i calculate how good a word is? take the number of answers assigned to a result, divide by the total number of answers (2315), and square that. repeat for every result and sum all the numbers you get.
let's say there were 3 possible results and 6 possible answers. a word that had the answers split evenly would get:
(2/6)^2 + (2/6)^2 + (2/6)^2 = 0.33
meanwhile, a word that had most of the answers assigned to only having one result would look more like:
(5/6)^2 + (1/6)^2 + (0/6)^2 = 0.72
lower numbers are better, with 1 being the worst possible score. these numbers actually tell you the average amount of elimination for the words. so the first example with 0.33 on average reduces the possible answers to 33% of it's original size (obviously), from 6 to 2. less obviously, the second example on average reduces the answers to 72% of it's original size, from 6 to 4.32.
SO, i did this calculation for each word that you can guess, and the best was ROATE. ROATE's score was 0.0262. this means that when you guess ROATE, on average the possible answers are cut to 2.62% of the original possible answers, which means on average after getting ROATE you are narrowed down to 60 words. that's pretty good for just one guess. (as I said earlier, the theoretical best would be 9.5). a very rough estimate tells me that a second guess would probably narrow you down to three answers, so with ROATE (if you play tactfully) you will hopefully usually get the wordle in three or four.
one other thing to keep in mind is that although ROATE is a valid guess, it isn't a valid answer. would a valid answer (such as the word RAISE) be better? i did the math and it is very close but ROATE is still better either way.
thanks for reading, i'm not even done because i calculated the best possible second words as well, ill talk about that later
i don’t even have tiktok but this https://www.tiktok.com/@blueakrasia/video/7069457159179767083 edit is so good and i literally just sporadically watch it every other day
oh also massive-shit-luvr was the one that showed it to me thanks hehe
i’ve got noooo idea what to do on this website eheheh