[HN Gopher] Things we learned analyzing 515M Wordles
___________________________________________________________________
Things we learned analyzing 515M Wordles
Author : jakubmazanec
Score : 30 points
Date : 2023-12-17 17:59 UTC (5 hours ago)
(HTM) web link (www.nytimes.com)
(TXT) w3m dump (www.nytimes.com)
| platelminto wrote:
| https://archive.ph/NnLq8
| neonate wrote:
| https://web.archive.org/web/20231217194006/https://www.nytim...
| pstuart wrote:
| Interesting that SLATE is the apparent optimum opener. I found a
| youtube that "showed" that CRANE is mathematically optimal:
| https://www.youtube.com/watch?v=v68zYyaEmEA
|
| The second guess (if a low hit count) fleshes out the remaining
| vowels.
|
| My average is 4, followed by 3.
| stoneman24 wrote:
| My usual is SALET (high frequency vowels in 2nd and 4th place),
| followed by CURIO to provide more hints if needed. Then review
| the information acquired. Usually it's solved on the 4th
| attempt, occasionally on the 3rd. Sometimes there are too many
| options and I bust out. Just my opinion, probably not optimal
| strategy.
| agubelu wrote:
| I do something similar, I always start with RAISE then CLOUT
| which cover the 10 most common letters. It usually gives good
| results.
| dekayed wrote:
| I start with RAISE as well, but follow up with POUTY to get
| the Y if I blank on the first round.
| flir wrote:
| I go with ATONE GIRLS. Gets me E, T, A, I, N, O, S, R, L,
| which according to the first link in Google (https://www3.nd.
| edu/~busiforc/handouts/cryptography/letterfr...) is 9 of the
| 11 most common letters in English.
| bhandziuk wrote:
| They have confirmed that the answer with never be plural so
| if you have the hope of getting it in 1 guess GIRLS ain't
| your best guess.
| Eisenstein wrote:
| I use CURIO but use YEAST instead of SALET to cover all
| vowels+y, and YEAST is much more likely (IMO) to pop up as a
| chance instawin, since I have no idea what SALET means.
| stoneman24 wrote:
| By the way, Googling SALET says that it's piece of
| historical armour. (a light round helmet extending over the
| back of the neck)
| mplewis9z wrote:
| I optimize for finding vowels first (just better for my brain
| to figure out the next guesses), so my opening is STOAE. I
| largely follow hard mode so my next guess is highly dependent
| on what hits, but GUILD is the default if the first
| completely blanks out.
|
| I normally solve in 3-4, about equally split between the two.
| thehours wrote:
| 3Blue1Brown later issued a correction video (due to a minor bug
| in the initial code) that showed it should have been SALET [1].
|
| [1] https://www.youtube.com/watch?v=fRed0Xmc2Wg
| jakubmazanec wrote:
| I didn't actually look into any analysis, but IMO the different
| results are caused by what words you include in the analysis,
| and what words you consider as potentially the solution.
|
| I use this [0] for choosing the second word, but NYT Wordle bot
| usually chooses something different.
|
| [0] https://real-statistics.com/wordle-strategy/best-second-
| gues...
| prawn wrote:
| I use SOARE, then GLINT and CHUMP. Most often get it in three
| (comfortable lead over the fours), then four, and then two or
| five are currently tied.
| mike503 wrote:
| I use STERN to start as it covers as much of RSTLNE as you can
| and I'm a Wheel fan :)
|
| It's amazing that even if I have 100% miss I'm almost always able
| to still solve it by the end. I don't know what it is with 5
| letters and 6 chances.
|
| There was a version I played that had 6 letters and it was
| terrible.
| furyofantares wrote:
| I also like to start with a single vowel. I feel consonants are
| more important for getting it in 3. My goals are mainly to get
| it in 3 as often as I can, and to never lose.
|
| Also nice to not have the E in position 5, where it's more
| likely to be green - early greens can put you into a "trap"
| where you can get unlucky and lose in hard mode. In standard
| mode you can get out of traps with a filler word, but in hard
| mode there can just be more possibilities than you have guesses
| available.
| forinti wrote:
| I start with SNAIL ROUTE.
|
| I've been thinking about my results because they make a curve
| around 4, the most common result. 5 comes next and then 3 is a
| few numbers behind (I'm trying to make 3 catch up).
|
| Shouldn't it just be a slope going up?
| aardvark179 wrote:
| It should only be a simple slope going up if you count puzzles
| solved in n turns or less.
| counterpartyrsk wrote:
| I start with a different random word almost every time. I look
| around the room for inspiration.
| gonzo41 wrote:
| I do this also. I don't get why people start always using the
| same word. I wish it would score you on unique attempts / words
| over time.
| BurningFrog wrote:
| One reason to start the same way is that you get familiar
| with many possible continuations.
|
| STERN + AUDIO
| tmnvix wrote:
| I always start with the previous day's solution
| neilk wrote:
| I learned a lot from Wordlebot analysis, though I hardly look at
| that now. According to Wordlebot I scored 99 on skill on today's
| puzzle.
|
| I always start with SLATE, and then if there are no matches at
| all, CRONY. I'm not sure exactly why, but the bot highly approves
| of this. Perhaps because the initial C and terminal Y will yield
| more information about the solution space, even if the letters
| are uncommon.
| thrwaydle wrote:
| I tried to come up with an obscene language strategy, which have
| been using since:
|
| FUCKS -> BONER -> MIGHT -> PLEAD / PLAID (Depending on the
| results before)
|
| Usually this is enough to guess on the 5th word or before,
| however sometimes the sixth guess is required. I have been able
| to guess all of the words so far.
___________________________________________________________________
(page generated 2023-12-17 23:00 UTC)