Wordle Lovers - Play the NYT Daily Game

[dwarde]

Wordle 1,126 3/6

 

[fresh_42]

Wordle 1,127 3/6


<--- optimistic instead of cautious

 

[Orodruin]

You don't understand what I'm trying to do here. The choices are irrelevant. The game Wordle is irrelevant as well. All we know is that the choices are equivalent. Let me ask it again:

• HEAD or TAIL?
• BLACK or WHITE?
• ITALIAN or BANANA?

Give me an answer to each of these questions then tell me how you came up with making that choice. If you say you "picked one at random", how did you do it? A coin toss? You took the first one? Did you go alphabetically?

How did you make the final decision and why?

Let me help you (WARNING: The following may influence your choice!):

• TAIL - I have never seen a double-headed coin but I know they exist and I never heard of a double-tailed coin, so I always choose tail even if no coin is involved;
• BLACK - I always preferred the darker things;
• BANANA - I find that word funny.

Now that everyone knows that, one could ask me "BLACK or BANANA?" And I would answer BLACK because the word is shorter.

This is how I picked my words. There's a reason behind every choice we make. There must be, no matter how senseless this reason may be. Even with "picking at random" we must choose a method to decide for us and there is a reason for selecting that method.

You clearly do not understand. I pick one randomly without thinking about it at all. I do not have any particular method. I just pick one. I do not care about what the words are, how they look, or how they place alphabetically. I just pick one. There is no conscious method to it unless you count “just take one without thinking about it any further”.

 

[Orodruin]

Wordle 1 127 4/6

 

[jack action]

You clearly do not understand. I pick one randomly without thinking about it at all. I do not have any particular method. I just pick one. I do not care about what the words are, how they look, or how they place alphabetically. I just pick one. There is no conscious method to it unless you count “just take one without thinking about it any further”.

You do everything to evade a simple question:

CAGE or BUTTER?

Don't even dare respond to this post without starting with your actual choice.

After you do that, explain how you picked it "randomly"? Did you use Eeny, meeny, miny, moe? Did you close your eyes and point a finger on the screen? Your brain must have sent some sort of command for you to do the actual "picking". Even if it was just "I picked the first one on the list."

The point is that, just like a computer, people cannot do "random", only "pseudorandom".

A pseudorandom sequence of numbers is one that appears to be statistically random, despite having been produced by a completely deterministic and repeatable process.

"Not thinking about it" - Of course you're thinking when choosing! When your hand will begin typing your answer, it will have to go towards "C" or "B". It will avoid any other letters. That requires some thinking.

I'm pretty sure the randomness would be different between throwing a die 100 times and writing manually 100 numbers selected "randomly" between 1 and 6.

 

[OmCheeto]

Wordle 1,127 3/6

 

[Orodruin]

You do everything to evade a simple question:

CAGE or BUTTER?

Don't even dare respond to this post without starting with your actual choice.

After you do that, explain how you picked it "randomly"? Did you use Eeny, meeny, miny, moe? Did you close your eyes and point a finger on the screen? Your brain must have sent some sort of command for you to do the actual "picking". Even if it was just "I picked the first one on the list."

The point is that, just like a computer, people cannot do "random", only "pseudorandom".



"Not thinking about it" - Of course you're thinking when choosing! When your hand will begin typing your answer, it will have to go towards "C" or "B". It will avoid any other letters. That requires some thinking.

I'm pretty sure the randomness would be different between throwing a die 100 times and writing manually 100 numbers selected "randomly" between 1 and 6.

BUTTER.

I chose pseudorandomly. Happy?

I just picked one of the words without thinking about it. Another time I might have chosen CAGE.

How hard is it to understand that there is no deep conscious strategy required to do this?

I do not understand your obsession with assigning strategy here. It would really not matter if I were to flip a coin or if I always picked the word coming first alphabetically.

 

[kuruman]

Wordle 1,127 3/6

 

[Herman Trivilino]

Wordle 1,127 3/6

 

[kuruman]

How did you make the final decision and why?

I think it is OK to ask the "how" but not the "why" when a random decision is involved. Asking "why" presupposes a causal relation that negates the idea of randomness. In my mind, the radioactive decay of a sample of unstable nuclei exemplifies randomness. One can base a random decision on it.

For example, I can take a radioactive sample that produces 10 counts per second on average. I establish the following "how" procedure: Count for 10 seconds. If the number of counts is above 100, choose candidate A; if it is below 100, choose candidate B." Poisson statistics tells me that the expected mean value of a measurement is $\mu=100\pm 10$ counts.

Now I make the measurement and choose B. If you ask me how I made this decision, I will describe the procedure above. If you ask me why I made this decision, I will say "there is no why, that's what came out of the established procedure."

 

[jack action]

Wordle 1,127 3/6

[ROATE]
[STAIN]
(I had to choose between SHAFT and SHALT)

In this one, I self-imposed the hard mode to maybe solve it in two guesses. The actual suggestions for the 2nd guess were SNAIL SLAIN SLING SLINK, with 23 possible answers. T in second place seemed to be popular, and the only word having an I (which seemed popular according to my previous suggestions) with this condition was my second guess.

For the third guess, EVEN THOUGH BOTH WORDS ARE STATISTICALLY EQUIVALENT (@Orodruin ), I correctly chose the answer because I thought it had more chance of being picked by the human that made the puzzle. Was I being smart or lucky? Who knows. But I had to make a choice somehow and this method was as good as any.

If I had solved it without the hard mode - using blindly SNAIL as my 2nd guess - I would have been left with 5 choices (STAFF STACK STAMP SHAFT STASH) and the only suggestion by the script happened to be the correct answer (because all of its letters are different and present in at least one other possible answer).

Again, this method based on the occurrences of letters for selecting the best guess didn't guarantee choosing the right answer, but it still was the logical choice in case it wasn't. And - as I said previously (@Orodruin ) - you have to necessarily choose a method to pick a word.

Proof that this "logical" method wasn't better than any other is if the answer had been STACK or STAMP, with this "logical" method, I would have had it in 4 or 5 guesses, which would not have been much better than picking any other choice as a 3rd guess (except the correct answer) which would have required 4, 5 or, at worst, 6 guesses before finding the answer.

Yet, I still had it in 3 today, not with one, but two distinct methods for choosing my guesses! And even though they did not use the same guesses as I did, so did @fresh_42 , @OmCheeto , @kuruman , and @Mister T . The "worst" result to date is 4 with @Orodruin which is still good. This cannot be only pure luck for everybody, by picking randomly between equivalent choices.

If you ask me how I made this decision, I will describe the procedure above. If you ask me why I made this decision, I will say "there is no why, that's what came out of the established procedure."

I may ask you why you chose this method, though.

 

[Orodruin]

And - as I said previously (@Orodruin ) - you have to necessarily choose a method to pick a word.

No, you do not. Your insistence on this is getting tiresome. Unless you count ”just pick one of them” as a ”method”. Just because you think you need a ”method” based on some unknown logic does not need there needs to be anything deeper than just ”pick one”. The ”just pick one” may be biased, pseudorandom, etc, but you do not need to specify any furtjer than that.

This cannot be only pure luck for everybody, by picking randomly between equivalent choices.

Thsg is completely disconnected. We obviously all have strategies for solving the puzzles. Most probably better than the average Wordle solver. We are generally not faced with equivalent choices.

 

[jack action]

Thsg is completely disconnected.

Because you keep saying all of these methods are equivalent - "just pick one", as you state - it is extremely relevant.

You started this by stating:

This is another fallacy. Once you know it is one of those two, neither word is more or less probable than the other.

If all "bias, pseudorandom, etc." methods were equivalent to a roll of the die (or "just pick one" if you prefer), the results for different players should also be random from anywhere between at least 3 and 6 guesses as I've shown; even while relying on hints from previous guesses.

The fact that the average PF Wordle player might be better than the average Wordle player doesn't exempt them from that randomness. The only explanation for them getting a better average is that the methods they use to "pick one" do not produce purely random results; they are not "just picking one" among apparently equivalent choices.

The "bias, pseudorandom, etc." method you select - as the evidence shows - matters. So, yes, I'm interested in knowing what they are. At the very least, it will be entertaining to hear what they are - unless one just stubbornly refuses to share them for some unknown reason.

 

[dwarde]

Wordle 1,127 4/6

 

[kuruman]

I may ask you why you chose this method, though.

You may and my answer is because it is an arguably random process totally independent of my actions. If I tossed a coin one might argue that there is bias in the probability, introduced by asymmetry, say in the coin's moment of inertia or perhaps the starting position of the coin.

In today's puzzle, after my seed word, I was faced with the same situation as I described in post #4,777. Five candidates, one of them being the filter that discriminates among all five. It turned out that today's word was not the filter so I got the answer in a guaranteed 3.

 

[fresh_42]

Yet, I still had it in 3 today, not with one, but two distinct methods for choosing my guesses! And even though they did not use the same guesses as I did, so did @fresh_42 , @OmCheeto , @kuruman , and @Mister T . The "worst" result to date is 4 with @Orodruin which is still good. This cannot be only pure luck for everybody, by picking randomly between equivalent choices.

My observation is that in most cases, we tend to get about the same number of tries. I often think "Hey, that was good / average /bad" only to find out that others did the same.

For a nerd website, we create really large data sets and we could test them on so many hypotheses. I was really surprised that some are reluctant to share even their overall result to determine $\mu$ and $\sigma$ over time and methods.

 

[Orodruin]

If all "bias, pseudorandom, etc." methods were equivalent to a roll of the die (or "just pick one" if you prefer), the results for different players should also be random from anywhere between at least 3 and 6 guesses as I've shown; even while relying on hints from previous guesses.

This is not correct and a misrepresentation of what I said. I was explicitly talking about the scenario where you have exactly two allowed words left. Your inference here does not follow.

In today's puzzle, after my seed word, I was faced with the same situation as I described in post #4,777. Five candidates, one of them being the filter that discriminates among all five. It turned out that today's word was not the filter so I got the answer in a guaranteed 3.

And this is to my point, the alteratives in this case are not equivalent. When one word (in addition to being a possible solution) is a filter that will allow you to get the correct word on the guess after while others won’t, it is a superior choice.

 

[OmCheeto]

And this is to my point, the alteratives in this case are not equivalent. When one word (in addition to being a possible solution) is a filter that will allow you to get the correct word on the guess after while others won’t, it is a superior choice.

This is how I now play. Now I'm curious how many words is the cutoff for when everyone else has to switch from 'just thinking it through' to 'writing it down'. 3 or 4 is my limit.

Btw, does anyone know if Daniel Tammet plays wordle? I would be curious of his scores.

 

[kuruman]

This is how I now play. Now I'm curious how many words is the cutoff for when everyone else has to switch from 'just thinking it through' to 'writing it down'. 3 or 4 is my limit.

Suppose that after 3 attempts the remaining choices are
BAKER
MAKER
TAKER
Faced with this situation, one strategy is a random guess. You write each choice on a piece of paper, put them in hat, pull one out (without peeking) and enter it. If it doesn't work, pull out another one. With this method, your score will be one of 4, 5 or 6. The expectation value for the score is $$\langle S\rangle=\frac{1}{3}(4+5+6)=5.$$ After many games, your average for this particular situation will be a 5 with equal amounts of 4 and 6.

A different strategy is to fashion a filter. That would be a valid word with TMB. So you try THUMB which discriminates among the three candidates and guarantees you the answer in 5. This is a cutoff of sorts. The expectation value is the same as in the random guess but the variance is zero. After many games, your average for this particular situation will be a 5 with no scores of 4 and 6.

The advantage of the filter method is evident if a fourth word, FAKER, is added to the mix. The random guess method presents the serious risk of not getting the answer with 1/4 probability. By contrast by using THUMB you are still guaranteed a score of 5 with no possibility of 6 or X. If you think about it, when there are two candidates left and you pick one, you are still using one of the candidates as a filter.

So my strategy has been to use filters right after the seed until I get the answer. I have played 708 games, my distribution is (1, 39, 349, 245, 32, 8, 0) with $\mu=3.43$ and $\sigma = 0.73.$ The personal pleasure I derive from this is in fashioning appropriate filters for each situation.

 

[fresh_42]

Wordle 1,128 4/6






I hate when they do this.

 

[OmCheeto]

Suppose that after 3 attempts the remaining choices are
BAKER
MAKER
TAKER

In this case I would play THUMB.

I was talking more in regards to if you had what I ended up with in game 1125 after my second guess:

ENTRY, NERDY, NERVY, NEWER​

As far as I can tell, picking either NERDY or NERVY guaranteed only one solution after that guess.

Likewise with game 1124:

DEITY, QUITE, UNITY​

Though in that case, I believe any choice I made would have guaranteed only one solution after the guess.

In both games I let my spreadsheet random number generator pick the words.

One thing I find most fascinating is something like game #1126. Both Wordlebot and I got it in 5, and everyone else at PF, aside from sbrothy who plays hard mode, got it in 3 or 4. AFAIKT, several people still play in un-augmented mode. I'm guessing that sometimes too much logic is a bad thing.

 

[kuruman]

I'm guessing that sometimes too much logic is a bad thing.

I agree. I don't spend too much time wondering whether there could be a better choice than the one I have already considered as being good. Life's too short to agonize over this.

 

[OmCheeto]

Wordle 1,128 4/6

 

[Herman Trivilino]

Wordle 1,128 5/6

 

[Orodruin]

Wordle 1 128 4/6

 

[jack action]

Wordle 1,128 4/6

[REALM]
[OINTS]
[SHEEP]

 

[kuruman]

Wordle 1,128 3/6

 

[dwarde]

Wordle 1,128 4/6

 

[fresh_42]

Wordle 1,129 4/6

 

[Orodruin]

Wordle 1 129 4/6

 

[fresh_42]

Wordle 1 129 4/6

You are cheating. Sphinx has 6 letters!

 

[Orodruin]

You are cheating. Sphinx has 6 letters!

Nonsense. It has 5 letters in Swedish!

 

[OmCheeto]

Wordle 1,129 3/6

 

[jack action]

Wordle 1,129 3/6

[POISE]
[LATER] 3 possibilities left: FACET TAKEN CADET
Chose this one because it seems more likely to be a selected word AND the letter D is more popular than F or K

 

[kuruman]

Wordle 1,129 3/6

 

[kuruman]

Nonsense. It has 5 letters in Swedish!

In Greek also. See here for an informative spelling and pronunciation of the word in various languages. It seems that the juxtaposition of sounds in this word presents a challenge for speakers of East Asian languages who have to separate the juxtaposed consonants with vowels.

 

[kuruman]

Wordle 1,129 3/6

[POISE]
[LATER] 3 possibilities left: FACET TAKEN CADET
Chose this one because it seems more likely to be a selected word AND the letter D is more popular than F or K

As you can see, word 2 returned to me the same pattern as yours (post #4,835) and I was faced with the same 3 choices as you. I chose as you did but for a different reason: the other two words had already been used.

 

[fresh_42]

Nonsense. It has 5 letters in Swedish!

I'm not a fan of replacing the Greek ph with an f. However, I'm not a fan of having to guess Speck in English either, so. Anyway, the pattern of a sfinx was nice.

 

[Orodruin]

I'm not a fan of replacing the Greek ph with an f. However, I'm not a fan of having to guess Speck in English either, so. Anyway, the pattern of a sfinx was nice.

Technically, that’s a Greek $\Phi$ so … one letter …

 

[gmax137]

Wordle 1,129 4/6

Chose this one because it seems more likely to be a selected word AND the letter D is more popular than F or K

I chose as you did but for a different reason: the other two words had already been used.

Well, if the other two possibilities have already been used, doesn't that show that @jack action 's approach is flawed? By the way, I have followed this approach myself for quite a while, but I'm beginning to see it isn't as justified as I had believed.

Since there are a finite set of possible solution words, and there is no "re-use" of solutions, then all three possibilities will eventually be used as the solution word. None of them really have a "higher probability" than the others.

 

[kuruman]

Well, if the other two possibilities have already been used, doesn't that show that @jack action 's approach is flawed?

I wouldn't call it flawed. Some people choose to ignore the additional information provided every day as a word is moved from the "not used list" and treat each game as a standalone with the same a priori probability every day. Others, like me, use the Bayesian approach and modify not only the a priori probability of the candidates but also the frequency distribution letters in the remaining candidates which is not necessarily the same as the frequency distribution of letters in the entire population of candidates. As someone said in an earlier post, it all depends on what you want to get out of this game.

 

[OmCheeto]

As you can see, word 2 returned to me the same pattern as yours (post #4,835) and I was faced with the same 3 choices as you. I chose as you did but for a different reason: the other two words had already been used.

And according to my list, we are quickly approaching the halfway point of the game:
≈ September 21, 2024

A point where I'm guessing that it will be noticeable that people who keep track of things consistently win, and those that don't, consistently lose.

hmmm.....

 

[dwarde]

Wordle 1,129 5/6

 

[Herman Trivilino]

Wordle 1,129 3/6

 

[kuruman]

A point where I'm guessing that it will be noticeable that people who keep track of things consistently win, and those that don't, consistently lose.

There will be a noticeable difference in the number of tries for a win between those who keep track and those who don't. I am one of the nerds who keep track. To document what you mentioned, I started recording my mean every time the percent of used words increases by 1%. I started at 44% and the next recording will be at 49%. By my reckoning we are now at 48.4% and I hope to establish some sort of baseline before the mean goes down noticeably. The plot so far is shown below. The drop is a statistically insignificant 0.4%.

 

[fresh_42]

I seriously doubt the hypotheses of both of you. My observation was, that the frequency of words in the English language significantly dropped by every new riddle. Whereas the number of possible words shrinks, and there are still hundreds unused, the probability to guess a rarely used word becomes lower so that the expectation value decreases rather than increases.

I mean speck? Seriously? That's a German word.

 

[gmax137]

I wouldn't call it flawed.

Yes, "flawed" bothered me a bit, but I couldn't think of a better word.

Some people choose to ignore the additional information provided every day as a word is moved from the "not used list" and treat each game as a standalone with the same a priori probability every day.

That's my approach. I don't have or keep any lists.

Others, like me, use the Bayesian approach and modify not only the a priori probability of the candidates but also the frequency distribution letters in the remaining candidates which is not necessarily the same as the frequency distribution of letters in the entire population of candidates. As someone said in an earlier post, it all depends on what you want to get out of this game.

Emphasis added -- I have thought about this also. Using the normal "crypto" list (E, T, A, O, I, N, S, R, ....) may not be accurate for the list of remaining solutions. Taken to the extreme where only one word is left, obviously.

 

[Orodruin]

AND the letter D is more popular than F or K

This is irrelevant at that point. It is not more common than F or K in the words you have remaining. I mean, sure, you can use it as an arbitrary strategy, but it will not help (or hinder) at that point.

 

[jack action]

doesn't that show that @jack action 's approach is flawed?

As it was already established in a few previous posts, all 3 possible answers statistically weigh equal value, no matter what; unless you have a list of previous answers (which I don't ... yet).

The point is that I cannot just "pick one". My brain needs to process some information before doing the actual picking. I could have picked the first one on the list. I'm just sharing my method, just for the fun of it.

@kuruman's comment seems to prove I was just lucky.

I'm just wondering how others do their "picking" because there seem to be a lot of lucky people around here.

EDIT: Actually, I lied when I said all 3 possible answers were equivalent because there is one answer that was worth less than the other two, and it did influence me, but I forgot about it when submitting my last post. Choosing TAKEN would have been a mistake as if it wasn't the right answer, it wouldn't have helped me choose between the last two. By choosing incorrectly either CADET or FACET, the answer becomes obvious.

 

[kuruman]

I seriously doubt the hypotheses of both of you.

Why? Maybe you misunderstood what I am saying. In today's word, after my first entry, there were 25 candidates that fit the starting pattern . After excluding the words that have already been used, the number of candidates fitting this pattern was reduced to 11.

My hypothesis is that if I have to choose from 11 candidates instead of 25, my score moves closer to being lower. Furthermore, because today's word is added to the used word list, the next time I see the same pattern (for the same first entry) the number of candidates will be reduced to 10 and then to 9 and so on.

As this reduction happens, it becomes easier to find a filter that discriminates among all candidates and hit the target with fewer tries. My method is not to guess the target word but to eliminate all but the target. For example, if my first entry returns (last letter E), there are 632 words with at least one E but not in position 5. After eliminating the used words, this number reduces to 327. Based on past performance (my mean is 3.4 tries), I expect to get the answer in a total of 3 or 4 tries. Since the last entry must be the target word, this means that I have to find one or maybe two filters to reduce the number of candidates to a final number of no more than 3 candidates. At that point I may have to guess.

So here is the core of my hypothesis: If I have 632 candidates, I will need more filters to apply sequentially than if I have half that many. More filters means a higher number of attempts and a higher score. Conversely, fewer candidates means fewer filters and lower scores.