Show HN: Can you lose at Wordle if you tried?(dontwordle.com) |
Show HN: Can you lose at Wordle if you tried?(dontwordle.com) |
I'd personally like an "easy setting" where there's another key color that indicates if a letter is viable. So when there are only say 10 possible words left it'd highlight the letter the word could start with. Maybe it's just me though.
It's a great alternative for anyone looking for more challenge out of their word guessing game.
My team and I build fun instant games. Many of them are actually word games. We saw your post and found your thoughts very interesting!
We would love to chat and explore how we could maybe work on new games together! Lmk what's the best way to connect:)
Don't Wordle 8 - SURVIVED Hooray! I didn't Wordle today! ⬜⬜⬜⬜⬜6438 ⬜⬜⬜⬜470 ⬜⬜⬜⬜132 ⬜⬜⬜⬜15 ⬜⬜⬜⬜4 ⬜1 Undos used: 0
Spoiler for today's word: https://imgur.com/a/kNvhK0q
I then looked up the word list and used the first word that fit the conditions. And that was the word.
Does it work on mobile without physical keyboard?
WATCH HATCH BATCH CATCH LATCH MATCH
There, I made 6 near-perfect guesses that would have cost me the game if I started with any one of them.
This is so infuriating. I've lost I think 3 times already because of this type of word choice.
Note: This is only relevant if you play on the hard-mode, which enforces you re-use guessed letters that have already been signaled as correct.
Once you know every letter is correct except the W after the first guess, you can think of the missing letter in the possible correct words (HBCLMP), and put together a word like "CHOMP" to eliminate a lot of potential guesses.
If the word was PATCH, like most other times I would have played
S(T)ONE
L[A]IRD
(C)(H)UM(P)
I did once fail to guess after six tries. The word was FOYER.
The new game looks interesting, and the strategy winning (failing to choose the word) is less obvious.
I still got it though. Those ones that rhyme with other words, especially if I blow through some of my guesses before finding most of the letters, are quick to answer but are typically the ones where I lose.
- Uncommon letters (I tend to mentally skip over them because they're uncommon.)
- Words that differ from the "usual" pronunciation for a pattern, like having a glottal stop. When I'm pronouncing some pattern mentally in the usual way I can end up skipping past a valid word because I'm mentally pronouncing it wrong.
Even if you don't play hard mode it can be difficult to eliminate even 3 less common letters or so in one shot.
It likely does mean you’ll have to sacrifice a lower average guess count in favor of avoiding losses.
To each their own.
I was in the mood for either a crazy 1-guess win or a longer road otherwise.
Game gets lame if you’re always using “statistically engineered” guesses.
I actually tired of the multiword Wordle "clones" for the reason that you say. I found you pretty much had to do a 3 or 4 word opening sequence and then mostly solve anagrams. Otherwise you end up wasting too many guesses trying to solve one word that has a lot of options.
As you said, to each their own. But the original point is, don't blame the word selection when "your own" doesn't work out.
I thought I would share some background about the game, some of the iterations I went through, and some of the future features I would like to add.
Background: Like many others, I have been playing Wordle daily. Recently, my Wordle win streak hit 99. I was admittedly very careful with my guesses on Day 100, not wanting to ruin my streak. Then I began to wonder…even if I wanted to lose intentionally, could I do it? Obviously, I could make intentionally bad guesses, but that would take the fun out of it. So I decided to build Don’t Wordle.
Iterations I have gone through:
The first version of the game was simpler and much worse IMO. I did not display the “Valid Words Remaining” at the top, and there was no concept of “Undos". Initially, I actually displayed the Wordle word at the top throughout the entire game! To be honest, that doesn’t change the difficulty of the game that much. However, given that the spirit of the game is not to guess the Wordle word, it seemed like the right decision to hide it.
I created the “Valid Words Remaining” feature when I realized how challenging the game was. I kept getting stuck and wondering whether there were even any words left besides to Wordle word. When I saw how fast the count shrinks, I felt the game was a lot more interesting.
I then added the concept of “Undos” when I realized the game was still too challenging. It’s particulary lame if you’re playing the game for the first time and you get eliminated after just 1 or 2 guesses.
Current Tech Stack:
-Route53 + Cloudfront + S3 + Create React App
Unexpected challenges:
1. The animations. I obviously copied all the concepts from the original Wordle, but I failed to appreciate the complexity.
2. The nuances of the repeat letter words.
3. What makes a Wordle square yellow? The answer is not so simple
4. Trying to get www.dontwordle.com and dontwordle.com to take you to the same URL (either www.dontwordle.com or dontwordle.com) and still support TLS and only use the tech I mentioned above. I actually still don’t have it working perfectly. I know of a solution, but it’s overly complicated
Features I would like to still add:
1. I have heard from multiple people that it’s annoying how you could get the game in an unwinnable state without realizing it. For example, maybe 20 valid words remain, but there is no valid sequence of words remaining to finish the puzzle. While the “Valid Words Remaining” feature is nice, it would also be cool to have a “Valid Solutions Remaining” feature
2. I would like to build a brute force solver that you can watch attempt to play the game in real-time. I have built something similar for a crossword puzzle and really enjoyed watching the computer try to fill the grid. I think it would be cool to do that same thing with this game.
Question I don’t yet know the answer to:
Do any words exist that would not have a single valid solution in this game?
Otherwise, your best approach is sadly a second cloudfront distribution with a different bucket (or bucket path) that just hosts a redirecting index key because you end up fighting stupid AWS design decisions.
Not that I've ever wasted stupid amounts of time on this particular problem...
Good game to add to my current daily routine (heardle, Wordle, quordle, worldle), but man is it hard!
> Otherwise, your best approach is sadly a second cloudfront distribution with a different bucket (or bucket path) that just hosts a redirecting index key because you end up fighting stupid AWS design decisions.
> Not that I've ever wasted stupid amounts of time on this particular problem...
Doesn't this document[0] cover dealing with ssl on cloud front? Plus Amazon's Cert Manager? Or is it missing something?
Pair that with an redirect/alias of www=>@ and you should be golden?
[0]https://docs.aws.amazon.com/AmazonCloudFront/latest/Develope...
I have also spent _too much time_ trying to get this to work. Gave up and used nginx instead.
> it would also be cool to have a “Valid Solutions Remaining” feature
That would be awesome. I suppose the biggest challenge would be to enumerate possible solutions in tree form to allow DFS or BFS searching.
Thanks for sharing!
If you use Cloudfront with an S3 bucket as origin for _dontwordle.com_, just create a second Cloudfront with an S3 Bucket for _www.dontwordle.com_ and enable "static website hosting" to redirect to "dontwordle.com" on the second bucket.
One other suggestion, is maybe the win condition shouldn't be exactly the same as in wordle. Instead of only allowing one game per day but using undos to make it easier, maybe allow 5 attempts per day instead? As a player I think it's simpler to be able to retry from the beginning if I lose rather than try to figure out the optimal time to use my undo.
As it's very unlikely you have every word in your list.
I mean, you start by playing almost the same as standard Wordle, just to guess the answer, without entering it. Then undo everything. Next, figure out a solution which avoids that word. Finally, type in your entire solution.
This could be a bit grindy, and not in the spirit of the game, but it seems like almost assured to win, or am I wrong?
Edit: If there are fewer undos allowed, this strategy gets more difficult
Well, I would say the question is poorly posed. The answer to the question as posed is simple as can be if the guess contains no duplicate letters. But if the guess does contain duplicate letters, the question as posed doesn't necessarily have any answer. It may - if all duplicates should be yellow, or if some of them should be green while all the rest are yellow, there's only one valid way to color the guess. But in the general case, you have some number of green squares, some number of yellow squares, and some number of dead squares, and you can swap yellow for dead without affecting the validity of the coloring.
So the question you should be asking is not "should this square be yellow?" but rather "how many yellow squares should this guess receive?". Then you can just choose appropriate squares to color yellow arbitrarily.
Well done!
Is it possible that in an earlier word, you had tried using an O in the 4th spot? That should be the only time that error message appears.
One thing I had wondered--but not enough to figure out what the code does...
Let's say you enter Mamma for example. And there is one M in the word but in the second position. Which M square does Wordle choose to turn yellow? (But maybe it doesn't matter?) There may be other cases like that as well.
Suppose the target word is “polar” and you guess “banal”.
A submitted word is marked in 3 passes.
0. Initial state: (“banal”, “polar”)
1. Zip together the letters of the guess and target words, and loop through the zipped list. When both letters at a position are the same, replace the guess letter with the symbol for “correct” (green) and remove the target letter from the target word. State: (“banGl”, “polr”)
2. Loop through the target word’s letters, replacing the first occurrence (if any) of each one in the guess word with the symbol for “present” (yellow). State: (“banGY”, “polr”)
3. Replace any remaining letters in the guess word with the symbol for “not present” (black). State: (“BBBGY”, “polr”)
Return the marked guess word.
---
Other examples:
0. (“mamma”, “amaze”)
1. (“mamma”, “amaze”)
2. (“YYmmY”, “amaze”)
3. (“YYBBY”, “amaze”)
---
0. (“nieto”, “otoño”)
1. (“nietG”, “otoñ”)
2. (“nieYG”, “otoñ”)
3. (“BBBYG”, “otoñ”)
edit: fixing formatting issues. Sorry, first time posting a comment on here.
The remaining grey Ms tell you the same thing (that those positions don't have an M).
I don’t think it matters, it should just default to the first M.
Don't know about Route53 + Cloudfront combination but your #4 challenge would be super simple with a webserver like apache or nginx.
The wordlist can be found on GitHub or by viewing source, so you can probably just discount 90% of the words in your guesser before you even start.
I don’t know if this is a bug or intentional, but playing today, I was eliminated on the last guess when there was a valid word that was not the wordle word. However, I had used that word previously and it had forced an undo. This came as a surprise, because I knew the word was safe and planned the 5th word to leave that word available.
edit: well, I solved it by adding a breakpoint to handleWordFinishedAnimating and printing this.state.targetWord . (The whole game is run in front-end.)
Well, yeah. That was a word.
Sort of. You could argue that Super Mario Bros. is 99.9% losing for most people. But within those losses are little personal victories, like getting past the first Goomba.
I hate games that try to remove all loss, assuming people can't handle it.
examples include arcade machines/roguelike, gotcha/lootboxes/gambling, crosswords
psychologists have found that the most effective reinforcement strategy is a combination of continuous and random rewards. For example, in World of Warcraft, enemies have a small amount of gold, but only a chance of carrying a good item.
[0] https://www.nytimes.com/interactive/2022/upshot/wordle-bot.h...
Seems like the perfect popular game: be easy enough for most everyone to get, but just hard enough to give a feeling of mastery.
I had a similar feeling when playing Guitar Hero way back. It was pretty easy to get the hang of at a medium level, but it was just hard enough that it felt like an accomplishment. Plus the "output" you got from playing the game well (a complex, professionally played song) was wildly disproportionate to to your inputs (a bit of well-timed button mashing that had a far simpler in structure). The mistakes screwing up the song sealed the illusion of skill.
Another point of confusion is that the first row says 2318 possible words, but there only seem to be 2309 words in the dictionary in the code
[EDITED to add:] Not exactly the same; with Antiwordle the idea is just to last as long as you can, rather than having a limit of 6, and it doesn't have undos. Also, UX-wise, I don't think Antiwordle tells you how many possibilities remain (win for Don't Wordle), and it doesn't have such slow tile-flipping animations (win for Antiwordle).
The gist is you’re trying to find the worst sequence of (hard mode) guesses possible for any word in the wordle set.
Their site shows submissions of up to 16 words and claims 17 is impossible [2]. I’ve been trying to do the same, but have yet to cut the search space enough[3].
Ex: calculating the worst chain for jazzy (with a 1000 word subset of the full word list) took an m1 core 1 hour and resulted in the ten-word chain:
civic, buggy, woody, array, leaky, mammy, fanny, happy, tasty, jazzy
[1] https://www.janestreet.com/puzzles/eldrow-index/ [2] https://www.janestreet.com/puzzles/eldrow-solution/ [3] https://github.com/roshangiyanani/wordle/blob/main/wordle/el...
It was tough to come up with any sequence that didn't have a 50/50 chance of losing on the last guess.
So again, overall it's cool, but too hard (for me) to seem fun, and it's a very different game than Wordle even though the primary mechanism is the same.
A feedback: the “you lost” screen passes too quickly, I couldn’t read the word of the day.
I like this one better, the undo button is a nice touch.
I still failed as I came to a 50-50% chance at the end (I was choosing between "f" and "v" to start the word, got the rest of them right).
Though I guess to be fair Wordle doesn't force hard mode, so it seems this version shouldn't either.
Cool idea though.
First guess. No, no, no, no, no, no, no.
The application won’t let me do anything even mildly suboptimal, and it’s not fun at all (aside from a brain exercise).
There are two different Wordle word lists--the set of legal guesses (roughly 13k) and the set of words in the solution list (roughly 2k). The latter list is curated to be a set of fair, interesting words, excluding plurals and more obscure words. Most Wordle solvers use knowledge of what words are in the Wordle solution database to pare down the set of "possible words". But with Don't Wordle, the "legal guess" list ends up being the more interesting one to show "remaining possibilities" for.
https://scoredle.com/ is a site that shows possible remaining words using the larger "legal guess" list, so you can see that you could have also guessed "SOCAS".
> Another point of confusion is that the first row says 2318 possible words, but there only seem to be 2309 words in the dictionary in the code
There were 2318 remaining possible words AFTER your first guess, which eliminated ~11k possibilities.
EDIT: this isn’t it - I missed the C in COMMA!
Once you finish the game, if you have any rows with 25 or fewer valid words remaining (shown to the right of the row), the word count becomes clickable. It will open a modal that will show you the valid words remaining after that stage of the game. I went ahead and entered your same sequence of words, and the the other word remaining was SOCAS. That is a valid word in the NYT official Wordle game.
>Another point of confusion is that the first row says 2318 possible words, but there only seem to be 2309 words in the dictionary in the code
There are different word dictionaries in the code.
https://en.m.wiktionary.org/wiki/bocca
Edit: The sibling with SOCAS is right.
There are two dictionaries, and the second contains 2309 entries which are found in the 12974 entries of the first.
WHIZZ seems like it could be a very strong opening move.
VEXED looks pretty promising too.
Even worse if your guess had them in the right location!
A decent strategy for guaranteeing that you can win is:
1. Play good Wordle moves (SOARE, RAISE, CRANE, SLANT/PRICE/DOUGH/BALMY[1], HANDY/SWIFT/GLOVE/CRUMP[2]) so that you can determine what the secret is. 2. Undo each of those so you have no restrictions about what you can play. 3. Figure out a 6-word sequence that you'll be able to play that doesn't include the answer. (I started out with some of those double-letter words, another word that used the remaining vowels, then worked backwards from the end with progressively fewer green letters.)
Though apparently if you type in the secret word, you can't undo that, so it may be valuable to re-do the analysis from the links but looking for words that can't possibly be the answer.
links:
1. https://i-kh.net/2022/04/22/my-lazy-wordle-strategy/ 2. https://github.com/norvig/pytudes/blob/main/ipynb/Wordle.ipy...
What if you scored like wordle but based on undos? 1 undo used, 2 undos used... Up to 5 undos used. This incentivizes the user to try and not undo if possible.
I think the eliminated score method is kind of redundant with the wordled scoring, just make the player guess the 1 word left and be wordled?
Really cool and fun idea, we had a great time playing with it and probably will add it to our daily rotation.
But you're smart to be generous with the Undos while introducing the game, because it helps users get familiar with it
... as if it isn't hard enough already :-D
I think in general the rule is this: - Any Ms in the correct position turn green (duh) - If there are X Ms remaining in the answer, turn the first X (non-green) Ms in the guess yellow
thanks for heardle! have you tried nerdle?
I didn't want to spoil at the time, but yesterday you would have been unlucky with both these word choices!
It was some tinkering around with bucket name (has to match the subdomain.domain.tld) and the cloudfront origin settings. But it works flawlessly. Feel free to ping me if you need details.
See links below.
coindex.de www.coindex.de
<script>document.location.href = https://dontwordle.com</script>Not everything has to be pass/fail, and most things in life aren't.
If you keep in mind that it is not, indeed, a real guitar, the sense of accomplishment doesn't have to be false. When I full combo a fairly difficult song, especially if it's one that I've gotten close but not quite got, it really feels like an accomplishment. Also, I'm active on the modding side of things for GH1/2/80s/RB, and that is where my true sense of accomplishment has rested. I've been involved with the mod projects Guitar Hero II Deluxe, Rock Band 2 Deluxe, and Rock Band 3 Deluxe, all of which have been wildly successful within the GH community.
I came up with it on my own. I derived it from some introspection on how Guitar Hero made me feel, and realizing it tricked me into feeling a false sense of competence, like I could actually play those songs well.
Even their games, though, although they are famously and genuinely difficult, do try and guide the player to success. They just do it in somewhat subtle ways by trying to kill you a lot. I’ve been watching a let’s play series on YouTube of Dark Souls I & III for a while (“Souls Academy”), and one of things I enjoy about it is how they often talk about what skills particular bosses or game areas are trying to teach you, so that you can be successful with later, harder challenges.
But you’re completely right that these games aren’t “hard not to finish”. They’re extremely hard to finish!
But then for some reason every single other major boss is very gimmicky and does not draw upon this. Radahn is a (fun) clusterfuck with yer boys. Rennala just has projectiles and no defenses. Rykard is... just poorly designed and can't hit you if you're in melee range. Fire Giant is a gimmicky horse battle. Astel is sort of similar to Rennala. Hoarah does his own thing with grabs and jumpable stomps. Malekith and Malenia are so awkwardly fast that you need to be proactively positioning rather than reacting to their telegraphed moves for the most part.
Then you've got at the endgame Morgot and Radagon who, for some reason, seem to be not only simpler, but significantly slower opponents than Margit.
Not a complaint really. I thought boss design was great and entertaining. But definitely a bit unusual. Sekiro's boss progression was amazing and the best I've ever seen. It's a really great inflection point when Genichiro swings by to make sure you understand the game's not gonna fuck around anymore.
Kind of like the Matrix.
Fun game but some very terrible UX decisions that seem spiteful at times.
Soulsborne games succeed because of exactly this. And you might notice that they’re drifting towards mainstream and that’s actually causing them to shed all the little details of what makes a Soulsborne game.
Elden Ring was ridiculously forgiving and I found it boringly easy. But it made me go back to Demon’s Souls.
It’s so player friendly and lacks most of the hard lessons previous titles teach.
As far as I can tell it’s just time consuming not hard.
On the converse is his upwards slash. That thing has basically no indicators and comes out faster than you're likely able to dodge... almost. If you fight him for a while it becomes apparent that when he wants to do this move he will walk towards you and otherwise will never do so. I couldn't tell if this was an intentional mechanic. Malenia later does something similar and it makes her fight much easier when you can understand her intentions before she begins doing any of her moves.