Show HN: I built a Wordle decoder that guesses your guesses | Hacker News Reader

Show HN: I built a Wordle decoder that guesses your guesses

github.com

90 pointsmattlock21 comments

Open articleView on HN

If you want to try playing this as a game, you can play the crosswordle https://crosswordle.vercel.app/

Ha, I built a crosswordle solver and didn't even know it!

appreciate both of you turning this question into practical app! https://news.ycombinator.com/item?id=30569209

I’m not the author of crosswordle, just a fan!

the most interesting part of this imo is the heuristic it uses to find the most likely word:

    # Greatly penalize words that have multiple of the same black letters
    # Greatly penalize words that have the same black letters as any seen word
    # Greatly penalize words that have the same yellow letter/index pair as any seen word
    # Reward words that have yellow letters that match yellow letters in seen words, but in different positions
    # Reward words that have yellow letters that match green letters in seen words
    # Penalize words that have yellow letters that don't appear in seen words
    # Penalize words that have green letters that don't appear in seen words
    # Rewards words based on commonality
    # Reward/penalize words based on line indexes and common letters

I built something similar to this that just lists all the possible words, and is nowhere near as smart -- but it lets you draw pretty pictures! https://jse.li/wordle-reverse/

I honestly thought I removed those comments lol. They might be a little outdated, but yeah, I spent a lot of time thinking about what kinds of information I could glean. There's definitely some bias in there towards a certain kind of strategy/play.

I like your project, very creative!

Looking at the parent run, my first thought was that its initial guess is a relatively uncommon word and is (I assume) probably not a very good choice as an initial guess. There are various lists of the best starting words in order going around. I wonder if you could get to the best results by picking the highest ranked word consistent with the other lines?

I thought about using lists like that, but it felt a little too much like cheating. My attempt at something like this is, if the word is one of your first two guesses, I reward it for having more common letters. And actually penalize common letter "misses" in later lines. I'm sure it's not 100% optimal, but I saw some improvement on my admittedly very limited set of test data. Here's the relevant code - https://github.com/mattruzicka/wordle_decoder/blob/0bfd7eaac...

Of course, doing what I wrote also makes the assumption that the person playing is using one of the top 10 (or whatever) starting words which may or may not be correct.

That aside, I started playing a 6 letter variant and, while I haven't done any sort of analysis with access to the word list, I did do a quick Word Hippo analysis of words with common letters, especially vowels, and while I'm sure my results aren't optimal they're probably pretty good.

ADDED: Which I think is probably not all that different from what your code is doing. I fed Word Hippo 4 or 5 letters and told it to come up with some common words.

But that still requires an assumption that a player has given some thought to good starting guesses.

I briefly had the thought of creating a game where you get a bunch of those Wordle color square messages from in-game "friends", and maybe some side info - "Tim always guesses adieu as his first word", "Jane always uses one of her kids' names - Olive or Hazel", etc and then you have to guess the word. Haven't had time to experiment with it though so not sure how fun it'd be. I don't retain any rights to the idea if anyone wants to run with it and try it out.

I tried that in a single game with a friend's Octordle result. You have to solve it yourself first, then work out their initial guesses. Here's my solution to today's Octordle:

  O = incorrect (white)
  ~ = wrong place (yellow)
  # = correct (green)

  Daily Octordle #40 / octordle.com

  OOOOO ~~~OO
  OO~OO OOOOO
  ~O~O~ OOOOO
  OOOOO O~#OO
  OO~OO OOOOO
  #OOOO O##OO
  OOOOO #####
  #OOOO ·····
  OOOOO ·····
  OOOOO ·····
  OOOOO ·····
  ##### ·····

  ~~O~O O~~OO
  OOOO~ OOO~O
  OOOOO OOOO~
  ###O# OO#OO
  #OOOO OOOOO
  OO#OO #####
  OO#O~ ·····
  O~OOO ·····
  O~~~# ·····
  O~O~O ·····
  O~O~O ·····
  OOOOO ·····
  ##### ·····

  O#OOO OOOOO
  OOO~O ~OOO~
  OOOO~ O~#OO
  OO~OO #OOOO
  OOOOO #####
  #O~O# ·····
  ~O~OO ·····
  ##### ·····

  O#~#O ~#O~O
  OOOOO OOOO~
  OOOOO OOOOO
  OO~O~ ~O~##
  OOOOO ~OOOO
  O~~OO OO~~O
  #~~OO OO~O~
  O#~OO O#OOO
  O#OO~ #####
  ##O## ·····
  ##### ·····

(This one was difficult and/or arbitrary, I normally try and complete Octordle in 11 guesses, i.e. the 8 correct answers plus three extra.)

My first three guesses, base64 encoded: VEFSRUQKTE9JTlMKQ0hVTVAK

That could be a lot of fun. Someone did a "guess the word based on people's tweets a while back: https://news.ycombinator.com/item?id=30101456

Throwing in those extra hints about the players make it feel more like a zebra puzzle, but that's not a bad thing, especially if you want mere humans to solve it. I can't easily process a thousand tweets, so giving more compact data is key.

Even using the example on the github page, it gives me a crash -

/Library/Ruby/Gems/2.6.0/gems/wordle_decoder-0.1.2/lib/wordle_decoder/wordle_share.rb:8:in `block in final_line?': undefined local variable or method `_1' for WordleDecoder::WordleShare:Class (NameError) from /Library/Ruby/Gems/2.6.0/gems/wordle_decoder-0.1.2/lib/wordle_decoder/wordle_share.rb:8:in `any?' from /Library/Ruby/Gems/2.6.0/gems/wordle_decoder-0.1.2/lib/wordle_decoder/wordle_share.rb:8:in `final_line?' from /Library/Ruby/Gems/2.6.0/gems/wordle_decoder-0.1.2/bin/wordle_decoder:14:in `<top (required)>' from /usr/local/bin/wordle_decoder:23:in `load' from /usr/local/bin/wordle_decoder:23:in `<main>'

Ah, that "_1" numbered parameter syntax was added in Ruby 2.7

I actually hadn't used that syntax before this project, was interested in trying it out. I'll document that it only works with 2.7 or later.

Thanks for sharing

Let's give the secret word more context, try to guess a keyword on this day: https://peotik.com/otd/

Could you post some examples of the results?

https://www.pastery.net/xfmbvv/

Contains SPOILERS for today's word.

I love it, thanks for sharing. Is the confidence a percentage? How is it determined?

Yeah, you can think of it as a percentage. The min is 1 and the max is 99. It could be 100, but I don't trust my mostly untested side project that much, so I cut it off at 99 :)

The base confidence score is determined based on how many words a guess technically could be, that's here - https://github.com/mattruzicka/wordle_decoder/blob/0bfd7eaac...

Then, I sort of just willy nilly add/subtract from that based on the "intrinsic" score of the word which is computed here - https://github.com/mattruzicka/wordle_decoder/blob/0bfd7eaac... and then the "contextualized" score AKA how well it works with other words, which is computed here - https://github.com/mattruzicka/wordle_decoder/blob/0bfd7eaac...

Fun, but it left me feeling jealous of someone with this amount of free time.

Ha, it ended up taking a bit more time than I thought it would, but once I started I couldn't stop.

Reply on news.ycombinator.com