How Do You Solve?

[LLapp]

n-j-b, I agree. Those tips read like they were written by an artificial intelligence program. They're written using the themes that would be included in tips, but not a single one of them is the least bit helpful.

[Eureka]

I agree with both of you. It sounds like AI. As does the post on another thread, linking back to this one.

[xlr8r]

If you watch the YouTube videos of Bansai (RIP) solving puzzles you will see he starts with the first letter, the cursor jumps to the next open space then follows thru until it is solved. He does this without the the counts displayed below the coded quote and it was before the app gave you red letters when a letter is used with multiple code letters. You can also see his 2010-2012 record solves were between 5 sec and 12 sec which are reasonable times for Bansai. In a lot of cases nanrich or munchlet might beat him early on but in general he usually held the record for around 10 years.

I believe a 2 second solve is impossible for even a computer-aided human. I believe n-j-b is an app, an AI, that scans the quote screen as it arrives, solves the puzzle and applies it to the blanks. n-j-b has a 2 second record on the common Bob Hope quote about Banks lending you money: "A bank is a place that will lend you money if you can prove that you don't need it. "-Bob Hope. And I believe n-j-b worked down from 4sec to 2sec which would be in line with an AI learning the quotes' pattern and applying the solution the next time it came up. ChatGpt can solve these Cryptograms but it would take over 30sec to copy the coded quote, enter it into the the UI, then take the solution and enter it into the game. So a 2 second solve would have to be an app written in something like Javascript to gather the input coded stream as it is displayed on the screen, solve the cryptogram and enter the solution back into the game. And I ask ChatGpt to generate some code to screen scrape the solution screen and it provided a valid javascript solution so it could probably write the code to do what I'm suggesting. It is an interesting Technical exercise but that is not why I play Cryptogram, I want to work my brain so even if I wrote something I would not use it because it is unfair to the true savants like nanrich and munchlet or the Bansai legacy and some of the others who are obviously skilled players.

When I first saw this 2 second record I attempted to think of human ways to solve a quote in 2 seconds and I came up with something to try. I think I have had that Bob Hope quote 15 or more times and I decided that was the one i would work with because it is one I would most likely see again soon. So I came up with this stateregy - I determined the letters that need to be typed to solve it and the sequence they should be typed and memorized them [here they are if you want to try: abnkisplcethwdyoumfrv] (That is 22 letters and I am sure there are quotes that use only 10-12 letters that could be memorized and typed faster but who knows when you will get that quote again or have a trigger to begin entering the letters, AND ... n-j-b did this quote in 2sec). Then every time I saw a Bob Hope quote, any Bob Hope quote, I wouldn't even look at the coded quote I just typed the letters. I have had it a couple of times since and applied my solve strateregy and i'm not even close to 2 sec. It takes me over a second to see it is a Bob Hope quote and to start typing. Now I will admit I'm probably not as sharp or as quick as I was 20 or 30 years ago so the fact that I can't even get into single digits using this method is understandable at least to me. But the fact is, anyone who solves this Bob Hope Puzzle has to type 22 unique characters and hit return to solve it. For a 2 second solve that is typing better than 11 characters per second without any cognitive time to spare or effectively 660 letters per minute and if you take an average of 6 letter words that is 110 words per minute. Guinness says the fastest typer, Barbra Blackburn, types 150wpm and Google claims a Gamer using the typing measurement app monkeytype was clocked at 300wpm so I suppose it is possible but rare. Most of us 10-finger typers (or 9 , I only use one of my thumbs) type 40-50wpm (source google).

So n-j-b (or will0416), you are the clearly the current mastersolver(s), the Bansai and Nanrich de-throners and, if my theroy is correct ,the only players soon to be logging 1 second solves. My question is can you give evidence that my AI app theory is wrong? Can you explain your skill otherwise? Can you provide us with a monkeytype measurement that shows you can type 130-150wpm?

[xlr8r]

Ok, I apologize. After finding and watching will0416's video, I'm super impressed and concede he is a master and deserves the respect I have given Bansai and Nanrich (and the others in that league like munchlet and Deanna and gryhnd) all these years.
Even though I don't think the video shows a 2 second solve there were plenty of solves that were less than 5 seconds and more baron's in that video than I'll probably see in a lifetime - outside of when new quotes are introduced.
Impressive and can see where these skills might produce a 2 second solve. I'll have to concede n-j-b may be just as skilled - does n-j-b have a video like Will's?

[kb83]

I definitely use a mixture of bottom up and top down. Sometimes the word pattern is a give-away, and the letter patterns aren't necessary. Or the author plus word pattern does it. At the opposite extreme, it requires one word to start the process (e.g. 'everything' -- which is the same as 'popularity' ); or 'possibilities'; or 'difficulties'; or 'difference'; or 'he who'; or 'do not'; or 'those who'; etc. etc. Sometimes, nothing seems to work, and after a certain time, I give up, because I care about my average solve time-- if I allowed an unlimited amount of time, I could probably solve it, but it would ruin the average solve time. And then there are cases, where all the letters are put in, and it seems right, but it is not. I reluctantly give up, even though a brute force approach (trying every letter for the ones that have to be wrong) would work, but of course would torpedo the average solve time. Another technique I've used is simply to try to figure out where the 'e' or the 'o' goes, even by trial and error. That's the other thing-- trial and error can sometimes help, and often I just blunder ahead, and then keep correcting based on 'almost words' that become apparent, and a gradual metamorphosis into the right solution happens. This was hard for me to learn, because it is against my personality to guess, but sometimes it has been the only thing that has worked. And yes, 'ness' or 'less' or 'ress' at the end of a word is helpful, but sometimes the same pattern is a double e (employee, etc.) or a double l (baseball, football, or ...). And ' abbc' at the end of the word is sometimes 'ally' but also 'hood'. And there are alot of homomorphs (same structure-- don't know if this is correct word)-- for example people-proper-indian, which-order, between-boyhood, baseball-neatness, and innumerable others. nothing-perhaps. And regarding the unbelievably fast times, my best times or record times often come when I just make an initial guess and the guess is correct. There is a lot of learning too, in that these are a finite (though very very large) pool of quotes, so you begin to encounter the same ones again-- and then I tend to make the same mistakes again, but occasionally I remember and learn and improve my time. I also often do worse on subsequent encounters with the same one. It's fun and addictive.

[PenguinzForever]

I'm a relatively new solver, but I usually look for the word "the" first in any quote, usually because it goes from high to low to high frequencies and it's easy to spot. I also use "and" often with high high then low frequencies. When those don't work, I'll use word patterns like "there" or "that" or "without" and usually the quote solves itself. Another thing I do is look for sentence structures like "is a" as the second and third words of a cipher or word endings like "tion" or "ing". Word patterns, particularly "without" has helped me beat old records occasionally, one of which was by over ten seconds. I even beat my own record once by four seconds using the same word pattern.