Saturday 9 April 2022: Wordle

Saturday 9 April 2022: Wordle

Wordle is a word puzzle game, the aim of which is to guess the day's target word of five letters. Six guesses are permitted. Each guess must be a valid word of five letters. Invalid words are inadmissible. As there are tens of thousands of five-letter words, without some help, up to six guesses is an insignificant number, offering little chance of success, and the game would be purely one of chance. In response to each guess Wordle changes the background colour to each of the five letters in the most recent guess from white to grey (a letter not in the target word), yellow (a letter in the target word but not in the correct position in the target word) or green (a letter in the target word in its correct position in the target word). When the game is played adequately, the grid of 30 squares (six rows of five columns) progressively changes from a background white to initially mostly grey descending to increasing amounts of yellow and green. The target word, if guessed correctly is all green: a green row.

The skill involved in the game is to choose words for one's guess that reveal letters, and positions of letters, in the target word as efficiently as possible. A least insightful approach would involve mechanically and progressively filling five of the six rows with 25 different letters (and thus potentially revealing the use or otherwise of the 26th letter in the target word), no doubt a significant feat in itself, might reveal none of the positions of letters in the target word, and one would be left with an anagram of the 'yellow' letters to solve (for which there may be several solutions) for the final guess. I suspect that many Wordle players consider six guesses to be close to failure, whereas a mere three guesses is satisfying and a parsimonious two guesses something of an achievement. A better strategy is to take into account that  the letters 'A', 'E' and 'S' are to be found in many words, whereas the letters 'X', 'J' and 'Q' are relatively uncommon. A first guess, therefore, might contain five of the most commonly-used letters. At present, I use the word 'TEARS' as a first guess, and whilst there are occasions when none of these five letters are in the target word, there have been occasions when four of the letters are in the target word. The distinction between vowels and consonants becomes important, as every word in English must contain at least one vowel, so the absence of 'A' and 'E' in the target word means that a second guess will almost inevitably contain an 'I', 'O', 'U' and/or 'Y'. Some people use the word 'ADIEU' as a first guess. I do not currently do so for several reasons. First because my second guesses are tailored to respond to my first guess of 'TEARS'. Second, because vowels themselves seem to me to be less helpful than consonants at homing in on the target word; or to put it another way, I can more easily run out of guesses trying to identify a single consonant than when trying to identify the correct vowel(s): 'STADE', 'STAGE', STALE', STAKE', 'STARE', 'STATE', 'STAVE'. Third, and perhaps trivially, I consider 'ADIEU' to be a French word, and the version of Wordle that I play (the New York Times version) uses words employed in English. I know many French ('TASSE'), German ('SUPPE') and Dutch ('HEEFT') five-letter words, and suppose them to be inadmissible because they are not English words. I am also uncertain to what degree of informal (blasphemous, coarse, vulgar or offensive) language Wordle may stoop in its target word.

Once a valid guess has been submitted, it cannot be withdrawn. For players of a particular disposition, it is possible to spend hours attempting to identify the next best guesses. For more experienced players, therefore, there is not only kudos in guessing the target word in a small number of guesses, but also of achieving the goal in as short amount of time as possible, and these two aspects are in almost direct opposition to each other, and attempting to manage this conflict can be revealing, if only to oneself, of what kind of a person one is. For example, thinking in terms related to Myers-Briggs typology, some people are highly (or irritatingly) methodical, and others are highly (or irritatingly) intuitive; some people are able to manage significant amounts of uncertainty and (irritatingly) are not straining for closure, whereas other are (irritatingly) intolerant of uncertainty and are focused on rapid closure.

More advanced play might involve substituting as-yet untested letters in place of 'green' letters. For example, having established that the second letter is 'L', and that third letter is 'A', there remain three letters yet to find and place correctly. A simple choice might be to guess 'BLACK' (which might be correct, but might not be). A more advanced choice might be to guess 'BLOCK' (which is certain to be incorrect, but reveals whether the letter 'O' is a letter elsewhere in the word. I express this strategy in terms of not spending guesses on 'green' letters (letters already in their correct position).       

I use and build tools. I have been building myself a database of five-letter words. In itself, this is not especially helpful in solving the day's Wordle puzzle because most of the five-letter words in my ever-growing database are words that were either Wordle target words, or else inspired by Wordle target words. I do not know whether Wordle re-uses words, but I doubt it. Consulting my database, however, does allow me to consider visually (or audially) similar words.

Wordle has helped me to consider the relationship between a word and how the word is spelled. For the first time in my life, I am having to confront directly that I am mildly dyslexic. I have been reading from the earliest of ages, probably three years old. The circumstances of my learning to read were deeply unpleasant, and involved a lot of corporal punishment. By the time I was about six years old, I loved reading, and derived a lot of pleasure from the stories I consumed. However, I was, and remain, a slow reader. For me, there is a time lag between, on the one hand, reading a word, and on the other hand, its sense emerging. Not infrequently, the meaning of a passage may elude me, and I find it necessary to reread the passage, sometimes multiple times. One would not have thought that this might apply to Wordle. However, I can sometimes fail to recognise that a collection of five letters actually constitutes a word with which I am entirely familiar. In addition, there is an aspect of Wordle that is like trying to solve an anagram, at which I am not skilled. The required focus on individual letters disconnects me from making sense of the word, or even that the sequence of letters constitutes a word. The problem, my problem, is sequencing. I have long been aware that I have a problem with way-roundness: for example, but far from limited to, correctly attributing left and right. For not quite as long, I have also been aware of some difficulty in distinguishing the order in which two or more things happen (sequencing). For example, I may be able to hear two musical notes perfectly clearly, but when they are played at speed, I find it impossible to say which of the notes comes first. This is where I return to dyslexia: when I am reading, the letters in a word are all over the place. Although not startlingly skilled, I am not especially bad at spelling. It is, for me, the meaning of a word, not its spelling, that determines (in my mind) what the word is. Each word I read is meaning (or function) first, and this meaning (or function) then determines its spelling. Even simply looking at what is on the computer screen, the spelling of many words is made stable (i.e. not appearing visually jumbled) only by knowing the sense of what I have just written. Often, I am unable to ‘see’ (read) words in a sentence that do not contribute to the sense of the sentence. An easy example is that I might, in error, type a word twice, but find myself unable to see the repetition. When typing, I rely heavily on the automatic spelling check function, and the grammar check function. It is true that my eyesight has deteriorated over the years. However, the issue about which I have been explaining would apply even were the letters in 72 point. When it comes to Wordle, the only aspect that involves meaning is determining whether a trial combination of letters actually constitutes a valid word. Otherwise the puzzle is all about letters.

One of the several ways in which my dyslexia manifests itself is in an almost perpetual obliviousness to the possibility of a second occurrence of a letter, either as a double letter (e.g. ‘LL’) or more usually not (e.g. ‘STATE”). A second way is that yellow letters have a dual, but contradictory, significance: the yellow letter does appear in the target word, but does not appear in the target word in its current position. This contradiction is significantly interfered with by my difficulty with way-roundness: because yellow letters do not appear in the target word in their currently-guessed position, I repeatedly confuse the functions of yellow and grey letters. In my mind, I find myself hunting for words that contain one or more grey letters whilst intentionally omitting yellow letters.

As well as a growing database of five-letter words, I also have a very small spreadsheet showing the letters in order, and as I work through a puzzle, I paint each letter’s cell with its current colour (no-fill, grey, yellow, green) according to status. Wordle does this automatically on an onscreen keyboard, but I am unable to look at a keyboard and make any kind of sense of the sequence of letters. My spreadsheet, on the other hand, presents the letters in a single, vertical sequence relating to the frequency with which they appear in English words, with ‘S’, ‘E’ and ‘A’ at the top, and ‘X’, ‘J’ and ‘Q’ at the bottom. However, this  spreadsheet tends to amplify my obliviousness to a second (or even third, e.g. ‘SISSY’) occurrence of a letter. If I can be bothered, I might add double letters as separate entries on the spreadsheet, although getting hold of data about their relative frequency might be near impossible. The "if I can be bothered" is important, because it is clear to me that Wordle is nothing more than a game, to which there must be a limit on the investment of time and effort.

One of my principal strategies is to try out letters with preference always given to letters according to the frequency with which they appear in English words. This strategy has drawbacks, not least because the frequency with which a letter appears in an English word is dependent on the length of the word. I do not have a letter-frequency chart for five-letter words. There is also an issue of what constitutes a valid (Wordle) word. I understand that 2,500 words have been earmarked as potential target words for Wordle puzzles, from a list of 25,000 words that Wordle recognises as valid. I am uncertain how correct my understanding is. I use only Scrabble dictionaries (note the plural, because UK, US and Australian Scrabble dictionaries are not the same) to determine whether a word might be valid in Wordle. I am aware that Wordle ‘favors’ US spelling, although I do not know if British spellings ever make an appearance. Again, I understand that Wordle target words are rarely plurals, so ‘S’ appears at the end of Wordle target words less frequently than common in English.

If I can be bothered, I should like to add to my spreadsheet two further features: popular pairs of letters, e.g. ‘CH’, ‘CK’, ‘ND’, ‘RD’, ‘TH’, ST’, etc.; and frequency of each letter in a specific position, e.g. as a first letter, or a last letter – for example, ‘E’ is uncommon as a first letter, but common as a last letter. To this end, my database of about 1,200 five-letter words also includes a spelling out of each word as five separate letters. At present I can only use this feature to filter out words, just on the off-chance that the Wordle target word happens to be in my database. I have not yet worked out how to calculate the number of occurrences (and therefore the frequency) of a specific letter in a specific position. I guess that it will involve SQL, learning sufficient of which will demand a reasonably substantial commitment of time and effort.

One strategy I continue to employ, despite consistently proving to be utterly useless, is intuition. I use my intuition successfully in a variety of different circumstances, including when interacting with people, when cooking, when encountering computer software for the first time, and for choosing a geographical direction when out walking. However, with Wordle, my intuition has proved to be worse than useless because it is a persistent distraction. 

On the other hand, a more successful strategy is that I always use the same starting guess, and have written a partial list of second guesses depending on Wordle’s response to my starting word. This means that I am on my third guess within about 30 seconds. My average number of guesses is about 3.7. I have had only one ‘2’.

Should I bother? I certainly have better, more creative and more important things to do with my time. What would give me a great deal of satisfaction is creating what would amount to an algorithm that would get close to solving each day's Wordle puzzle for me. I have written some very complex spreadsheet equations for calculating annual projected use and cost of electricity, and for calculating when the heating oil storage tank will run out of fuel. These complex equations are extremely useful and afford me much satisfaction. Were I able to do the same for my engagement with Wordle, I might save myself some time and effort but still achieve the satisfaction.