I have seen lots of articles/ research papers about the seed word. I’m not a scientist and I don’t have access to big powerful computers. So, my approach is more empirical. I go by the frequencies of various letters.

Static triad
A static triad is a predefined group of three five-letter words (seed word, second word and third word) that include as many of the top 15 letters frequency wise as possible without any overlaps.  One example is SOARE, CULTI and NYMPH.

Seed word: The seed word could include the top five letters, frequency wise anywhere in a word, or it could include the most frequent letters in each of the five positions, as long as a valid word could be formed with these five letters.  We could classify the seed words into two broad classes – 1) containing three vowels (example SOARE) or 2) containing two vowels (example ALERT).

Second word:  I try to ensure the second word has the remaining top five letters from frequency perspective anywhere in the word. None of the letters in the second word should overlap a letter in the first word since that’s a wasted opportunity.

Third word: The objective of the third word is to include the vowels missing from the first two words and capture consonants that are high on the frequency distribution, but don’t overlap with the first word or the second word. I consider Y to be the 6th vowel.

Dynamic Triad
A dynamic triad is one where the choice of the second word (or third word) depends a lot on what is revealed by the seed word (seed word and the second word).

If the seed word reveals enough information, then I don’t use the predefined second word. For example, if a seed word identifies two vowels, regardless of which positions they are in, I try not to search for the remaining four vowels since most of the Wordle words contain only two vowels. In such a case, the second word could be better used to search for letters that are not in the seed word and/or to confirm the positions of the letters identified by the seed word if their positions are not confirmed.

Similarly, the choice of the third word very much depends on the context and there is no hard and fast rule that it should always be the third word of the predefined triad.

Eliminate-not-Enumerate (EnotE) strategy
I deploy eliminate-not-enumerate strategy, especially in the case of orthographic neighbors. Orthographic neighbors differ in only one character.   One example is COWER, LOWER, MOWER, POWER, ROWER, TOWER, etc. If the last four letters have been confirmed and we are looking for the first letter that the WordleGod has selected for this particular Wordle, EnotE is the perfect strategy.

Under the enumerate approach, one starts with the first word (of the orthographic neighbor set).  If that’s not the right one, enters the second one; if that’s not the right one, moves on to the third one and so on and so forth. It’s highly likely that there is a good chance one is going to run out of six guesses. It’s also possible that LadyLuck may smile on him/her and the very first word from the orthographic neighborhood set he/she enters happens to be the target word selected by the WordleGod.

Under the eliminate approach, I try to look for a word that would eliminate most of the possibilities, if not all but one possibility. It’s not always possible to find such word, but it’s worth a try. This way one is assured of solving the puzzle in two additional guesses, one for the elimination word and one for the target word.

Whether one uses the enumerate approach or the eliminate approach comes down to one’s risk profile. If one is a risk taker, he/she probably would like to use the enumeration approach. Since I am not a risk taker, I use the elimination approach.

A “bonus” opportunity is when one can find an elimination word from within the set of possible answers. In such a case it’s the best of both words. There is a chance that the possible word one enters as the elimination word happens to be the target word. In such a case one would have solved the puzzle and saved the extra guess. But if that doesn’t happen to be the target word and it ends up eliminating all but one then the next guess is the target word.

Sometimes in the list of possible words, one may find a word or two that from his/her judgment have very low probability based upon the frequency charts. Then one should try to find an elimination word without these low frequency words, if by excluding them it becomes easier to come up with a strong elimination word.

I have posted several posts that explain these different scenarios I have described above. Please feel free to look at them. If you have a situation that is not covered, you may send me an email message and I’ll try to find an example and post that also.

So, to summarize while seed word is very important, many times, the second word is as important, if not more important to solving a Wordle puzzle in few guesses as possible.

Also, from my experience, if a seed fails to reveal any of the five letters, it’s not a “bad” seed word. The fact that the five characters of the seed word don’t appear in the target word reduces the universe of possible words by a very large percentage. I will cite one example at the end of such a scenario.

That’s all about the seed words. One thing to keep in mind is since we are dealing with probabilistic outcomes, it’s quite possible that one may go through a string of zero or one revelation by the seed word. But that is how probabilistic solutions work. It’s like tossing a coin. If one is looking for heads, it’s quite possible that they may get three or four tails one after the other.

Some people have the tendency of alternating between different seed words. I think that breaks the probability-based model’s underpinning. There is nothing wrong with that as long as one understands that the probabilistic models work better when the sample size is large enough.

There is no universal agreement as to which particular word is the best seed word. Professors in the areas of combinatorial mathematics, operations, research, physics etc. have run models and/or simulations and have come up with different seed words that for them are the best. So, your job is to pick one that you like and then stay with it for a “long” period of time.  If a month is too long, definitely for 2 weeks.

Another thing to keep in mind is that academics use large computing power and have access to the entire set of 2,315 Wordle words to arrive at the “best” seed word.  My empirical approach relies on the Wordle words disclosed to date.

Example of a “bad” seed word
Wordle Puzzle dated July 12, 2022

Happy Wordling!