Wordle Lovers - Play the NYT Daily Game

[jack action]

That’s not how science works. You are the one making the hypothesis that you can select the correct answer more often than random selection would. You have yet to actually prove this.

• This is how science works.
• Yes, I am making a hypothesis.
• I am in the process of proving it. (Will it work? Nobody knows.)
• This is why I'm showing my selection process at every game. I do not state facts, I state my reasoning for selecting my words.
• It has gone really well up until now. (Can you imagine that my program suggested - quasi-randomly^[1] - FALSE as a filter word for the 2-word list EASEL and FALSE? It is far from being the first time I won that bet.)
• I cannot go faster than NYT publishes a new puzzle because it is only with them that this hypothesis would work. (Because the basis of my hypothesis is that a [very biased] human selects every solution.)

---

[1] The program removed all letters already found in each word and ordered the remaining letters by count. In this case, all letters were removed from EASEL and only F appeared in FALSE. It then parsed the list of possible answers (2 words, in this case) to see if a word contained all the remaining letters (in this case, only F) and, obviously, FALSE was the only possibility.

If there weren't any words in this list, it would have parsed the complete list of words to find appropriate filter words for as many as possible remaining letters, starting with the most popular ones.

 

[Orodruin]

I am in the process of proving it. (Will it work? Nobody knows.)

Then don't make statements like this

I'll keep selecting the right answer more often than not until proven otherwise.

until you have proven it conclusively with reasonable methodology at a reasonable confidence level.

 

[dwarde]

Wordle 1,229 4/6

 

[jack action]

Then don't make statements like this

until you have proven it conclusively with reasonable methodology at a reasonable confidence level.

So I'm supposed to NOT use a methodology to prove a hypothesis BEFORE I can prove the hypothesis works? How do I do that? It doesn't seem very scientifiky. Again, I don't make statements, I think out loud and share with people.

Just to inform you in advance, if somehow I find out along the way that it doesn't work, I'll change my methodology without any notice. Whether it works or not, don't expect an article published in a scientific journal either.

All I know is that when I'm stuck with selecting between two words, my human brain must have a thought process to do so. I would gladly ask you how you do it, but I won't go down that road again.

This is for fun. I don't think PF will be sued because someone followed my methodology to solve NYT Wordle and lowered their statistics because of it.

To quote you:

Your insistence on this is getting tiresome.

 

[Orodruin]

So I'm supposed to NOT use a methodology to prove a hypothesis BEFORE I can prove the hypothesis works?

No, that is not what I said. Please read what I am saying instead of making allegations. What I said was that you should not claim it is on others to prove your hypothesis wrong. What you are supposed to do is to collect sufficient data to support your claim before you unequivocally state that you are definitely selecting the correct answer more often than others.

 

[ChaseLess]

Wordle 1,230 3/6

 

[jack action]

What I said was that you should not claim it is on others to prove your hypothesis wrong.

I don't claim my hypothesis is right. But how can someone claim it is wrong without ever trying it first or looking at any sort of data beforehand?

What you are supposed to do is to collect sufficient data

This is what I'm doing, You are the only one making a big deal out of it every time I present the new data.

to support your claim

I don't claim anything. I just state facts: 1- Here is how I selected my guess, 2- Here is the result.
I don't pose any judgment about it.

You are the one making claims: I say that I noticed a pattern of more positive words being preferred over negative words and you claim it is not correct and I should use words that are more common to test my hypothesis ... but that is not my hypothesis! If you want to prove that hypothesis, do it, but I don't see that pattern emerging in the puzzle solutions.

When you say:

That if anything is conjecture.

Of course it is since, according to Wiktionary, it is pretty much a synonym for hypothesis:

• (formal) A statement or an idea which is unproven, but is thought to be true; a guess.
  I explained it, but it is pure conjecture whether he understood, or not.
• (formal) A supposition based upon incomplete evidence; a hypothesis.
  The physicist used his conjecture about subatomic particles to design an experiment.
• (mathematics, linguistics) A statement likely to be true based on available evidence, but which has not been formally proven.

before you unequivocally state that you are definitely selecting the correct answer more often than others.

I never made such a claim (the keyword being "unequivocally"). But it does look promising as time goes by. I cannot be "lucky" all the time. "Being lucky" would even disprove your theory of pure randomness, even if my pattern hypothesis was wrong.

The key is to identify the correct biased human pattern. Like trying to identify the tell from your opponent in a poker game. This is why the same people win the poker tournaments over and over, even though the cards are dealt randomly.

Of course, if you are convinced that the selection is purely random, maybe not even made by a human at all, you do not look for patterns. When I play on other websites, I don't use that technique at all.

For example, I was using the word CLASH as my seed word for a while as it seemed to help me find the answer more quickly. It gave good results for a few weeks and then I didn't notice any real differences. The hypothesis was disproved (or it had run its course?), so I abandoned it.

When I'll stop seeing the pattern of positive words being selected more often than not, I'll do the same.

 

[Orodruin]

I don't claim my hypothesis is right. But how can someone claim it is wrong without ever trying it first or looking at any sort of data beforehand?

Nobody has done so. I have kept asking for evidence that is true since you keep claiming so.

This is what I'm doing,

No, it is not. I quoted one of your posts explicitly claiming that you are selecting the correct answers at a higher rate than random. This is what I am arguing against. I have also questioned your assertion that a word you selected should be more likely to be selected in some instances.

Of course it is since, according to Wiktionary, it is pretty much a synonym for hypothesis:

The problem is not that it is conjecture. The problem is that you are presenting conjecture as fact before you have shown it.

 

[Orodruin]

Wordle 1 230 4/6

 

[kuruman]

Wordle 1,230 3/6

 

[gmax137]

Wordle 1,230 4/6

 

[kuruman]

I say that I noticed a pattern of more positive words being preferred over negative words . . .

The last ten words before today's were
SPOON SHOUT GOOFY BOSSY FROWN WREAK SANDY BAWDY TUNIC EASEL

Three of these
SPOON TUNIC EASEL
are objects and, as a human, I would consider them neutral. They could be positive or negative depending n one's personal experience and should be exempt because human experiences are unique to individuals. For example, if I were born with a silver SPOON in my mouth, I would consider it positive, but if my low-class stepfather spanked me with a wooden SPOON to discipline me as a child, I would consider it negative.

Of the remaining words I would consider,
SHOUT as neutral: It is impolite to SHOUT when talking to someone but it is necessary to SHOUT "Fore" as a warning when your golf ball is in a trajectory to hit someone on the head.
GOOFY as negative. Someone who does GOOFY things is not all there.
BOSSY as negative. Nobody wants to be ordered around by BOSSY people.
FROWN as negative. It shows disapproval when one's remarks are met with a FROWN.
WREAK as negative. One WREAKs havoc or vengeance neither of which is good when it happens.
SANDY as positive. It evokes SANDY beaches, relaxation and good times.
BAWDY as negative. It evokes indecency and lack of respect.

So there you have it. Of the last ten words I, a human, would consider only one positive.

I understand that this is only a small sample of the 2300 (or so) words on the list. The proper way to proceed is to first classify all of them as positive, neutral or negative. Then plot on the same graph the unselected percentage of words in each category as a function of time. All three plots start at 100% on day 1. If the theory that a human preferentially selects positive words is correct, then on day 1230 (today) the fraction of positive words should be (within error) well below the other two and the random selection expected percentage $(2300-1230)/2300=46.5\%.$

However, I will not attempt any of that because I am not interested in proving (or disproving) that a human preferentially selects positive words for the daily puzzle. This is too much work for something that doesn't interest me and is, therefore, not fun to me.

 

[Orodruin]

that a human

It is not even a human in general. It is one specific human - the Wordle editor - who may have all sorts of personal bias we are unfamiliar with.

 

[gmax137]

SANDY as positive. It evokes SANDY beaches, relaxation and good times.

If the Wordle editor is a New Yorker, SANDY could be a negative reminder of the hurricane.

the Wordle editor - who may have all sorts of personal bias we are unfamiliar with.

yes.

In defense of @jack action , I think he uses the positive/negative only to decide between possible solutions, so it is a relative judgement: which (of the two or three) choices is "more positive."

 

[OmCheeto]

Wordle 1,230 2/6




Very lucky seed left me with only 1 possible answer.

 

[Orodruin]

If the Wordle editor is a New Yorker, SANDY could be a negative reminder of the hurricane.

Am I the only one who doesn’t like sand between the toes?

which (of the two or three) choices is "more positive

Even so, if it is the case you should be able to split the word list in positive/negative and then check the relative rates. If positive is more likely relatively then the relative rate should be higher.

 

[jack action]

Wordle 1,230 3/6

[DRONE] 8 left
[DAILY] filter - 1 left
[WEIRD]

---

No, it is not. I quoted one of your posts explicitly claiming that you are selecting the correct answers at a higher rate than random. This is what I am arguing against. I have also questioned your assertion that a word you selected should be more likely to be selected in some instances.

I'm tired of this. I decided to waste my time and did the math to show you what I knew informally in the hope of ending all of this.

Going back to my previous posts where I'm sure I used the method of most-probable-word-to-be-selected-by-a-human-according-to-my-very-subjective-criteria I found 45 puzzles out of the last 102^[1] puzzles I did:

Where I was right (34 total):
1129-1130-1132-1134-1135-1136-1141-1142-1144-1178-1185-1187-1188-1189-1190-1191-1192-1194-1195-1196-1197-1198-1199-1201-1202-1203-1204-1206-1210-1218-1220-1221-1222-1229

Where I was wrong (11 total):
1137-1140-1182-1193-1200-1207-1209-1216-1217-1225-1227

That is a 75,6/24,4 chance of getting it right which is better than 50/50 for pure randomness.

It may change over time but for this sample of 45 puzzles, I'm satisfied enough to continue exploring this avenue - thus my statement (as opposed to claim):

I'll keep selecting the right answer more often than not until proven otherwise.

This is as much math and data gathering as I will do to "prove" what I know informally. If and when I reach a 50/50 ratio, I'll know it by "feel" and stop using this method or at the very least modify it. That is what "proven otherwise" means in my statement (again, not a claim).

Logging my word selection process in this thread is to help me get a "feel" as I don't want to do formal data gathering (in a spreadsheet, for example). I know I'll get crazy about it if I do. Doing it this way gives me a good sense to help me keep track of my progress informally such that it stays fun and doesn't feel like a job. If someone doesn't like my comments, they can be ignored as they are the equivalent of words written on a napkin, for my eyes only.

---

[1] It includes 32 puzzles (1146-1177) where I stopped commenting, out of respect for the OP, after the OP made this post:

And I personally dislike "x possibilities left" comments. I don't want to know who is cheating.

I started again when he announced he left PF. So I don't know how many times the method was used during that period and how often it was successful or not.

 

[Orodruin]

I'm tired of this. I decided to waste my time and did the math to show you what I knew informally in the hope of ending all of this.

Doing the actual science to prove your claims is never a waste of time. At least not if you want people to believe you. This is a science forum after all. It is literally part of my job to be sceptical until statements are proven beyond reasonable doubt. That includes doing the math.

Where I was right (34 total):
1129-1130-1132-1134-1135-1136-1141-1142-1144-1178-1185-1187-1188-1189-1190-1191-1192-1194-1195-1196-1197-1198-1199-1201-1202-1203-1204-1206-1210-1218-1220-1221-1222-1229

Where I was wrong (11 total):
1137-1140-1182-1193-1200-1207-1209-1216-1217-1225-1227

Ok, compiling data. That is good, but to make the statement solid you will need more than:

That is a 75,6/24,4 chance of getting it right which is better than 50/50 for pure randomness.

These are your sample mean and expectation from randomness. In themselves, they do not hold any statistical power. What you need to compute is the probability of this (or a more extreme result) occurring by chance if it were random. Essentially all you need to do now is to look at the CDF of the binomial distribution. It is literally the least tedious step in all of this.

 

[dwarde]

Wordle 1,230 3/6

 

[gmax137]

Am I the only one who doesn’t like sand between the toes?

Nope. Have a listen to Ali Baba's Camel, by the Bonzos

EDIT:

 

[jack action]

to prove your claims

I'm not claiming anything.

At least not if you want people to believe you.

I don't.

but to make the statement solid

I don't care. I'm just having fun.

In themselves, they do not hold any statistical power.

Why am I not surprised that not only do you think that but you also need to share this opinion?

What you need to compute is the probability of this (or a more extreme result) occurring by chance if it were random. Essentially all you need to do now is to look at the CDF of the binomial distribution.

The data was always available to everyone and I have now identified it for you. You are more than welcome to do any analysis you wish to prove your claim.

It is literally the least tedious step in all of this.

I'm glad it will be easy for you to do.

Finally, the message may be easier to understand with a song:

​

 

[gmax137]

Let's please not get this thread closed, ok?

 

[Orodruin]

I'm not claiming anything.

That in itself is a claim. And you have definitely claimed that you are selecting the correct answer more than the expectation from random. I have already quoted several such instances so I will not bother to do so again.

I don't.

Then don’t get upset if they don’t.

Why am I not surprised that not only do you think that but you also need to share this opinion?

Because that is how science works and this is a science forum.

The data was always available to everyone and I have now identified it for you. You are more than welcome to do any analysis you wish to prove your claim.

That’s not how science works. You are so close yet you - yes you - choose hand waving instead of making a scientifically sound claim. That is what bothers me. I couldn’t care less if your actual hypothesis is true or not.

 

[Orodruin]

Let's please not get this thread closed, ok?

Apart from the snide remarks of @jack action the discussion is relevant to Wordle. Proving the hypothesis would be relevant. They are so close.

 

[Orodruin]

Wordle 1 231 2/6

 

[jack action]

Wordle 1,231 3/6

[IRATE] 12 left
[HEXYL] filter - 1 left
[SIXTH]

 

[kuruman]

Wordle 1,231 3/6

 

[OmCheeto]

Wordle 1,231 3/6

 

[dwarde]

Wordle 1,231 3/6

 

[gmax137]

Wordle 1,231 3/6

 

[kuruman]

Wordle 1,232 4/6



Anagram!

 

[jack action]

Wordle 1,232 4/6

[BLAST] 78 left
[IRONE] filter - 2 left: SNOOP SNOWY
[SNOWY] most probable
[SNOOP]

most-probable method count update (Success/Fail): 34/12 - 73.9%/26.1%

 

[OmCheeto]

Wordle 1,232 4/6

 

[gmax137]

Wordle 1,232 5/6

CHAIR
LOOSE this turned out to be a productive guess
STOOP
SWOOP "more positive" than -->
SNOOP

 

[dwarde]

I started out with one of those new fangled seed words, but it was a lucky guess that solved the puzzle.
Wordle 1,232 3/6