Log in

No account? Create an account
delirium happy

Just keep on trying till you run out of cake

Previous Entry Share Next Entry
puzzle me this
delirium happy
I remember when I was younger doing a type of puzzle that was like a crossword grid, except instead of having clues, all the squares had a number from 1-26, each representing a letter of the alphabet, with a few (normally three) of the letters given as a starting point. I was just wondering how well these would stand up to being done by frequency analysis (I think it's called. The way of breaking cyphers where you just compare the frequency of the symbols with the known frequency with which letters occur in the english language). I don't know whether you could more or less solve them in a fool proof fashion that way, or if the sample size wouldn't be large enough, and the words would be too random. Any ideas anyone? (oh, and does anyone know what that type of puzzle is called?)

  • 1
Sounds like a cryptogram to me. I think they're usually solved by looking at the 2 and 3 letter words, and sometimes the endings, (like "ing".)

- Smushy Alex

I know that's how they're usually solved, I was just pondering an alternative method.

And I don't remember them being called that. From the looks of things, they have several alternative names.

its called a codeword. u can get them in the daily mail. i used to do them loads. its funny u shud say about the frequency thing cos sometimes when stuck i will look at the position and frequency of certain letters and the letter S usually occurs most frequently so u can guess where that goes and i think the letter E occours more frequently than most of the vowels so u can usually put them in. u can usually tell where the Ys go too so there is some truth to the being able to predict it thing!

i cant believe i just admitted to reading the daily mail...

*geek mode off*

Nicky (via lipstickglove's friendslist) xxx

I don't remember them being called that either. Seems like they have several different names.

"Reference Crosswords" is the name I remember, but ICBW. And I recall doing them, at least initially, by frequency analysis - particularly on the letter E. I'd imagine there isn't enough material there for frequency analysis, but if you /were/ going to do it, I guess you'd have to count any square that was in two words twice, IYSWIM..

Hmmm. More thought and research called for here.

s/for frequency analysis/& to work reliably/
Oops. :-)

Damn you, now I want to go and buy a book of these things.
Well, mainly it's because I was reading the Code Book by Simon Singh recently and found myself thinking exactly the same thing. I've heard them called keywords before. I think they have lots of names.

Other similar beasts

Another puzzle format,but rarely seen in puzzle books and hardly ever in newspapers and magazines, are proper cross-number puzzles, where each A&D clue is arithmetic or
individually couched in a sub-puzzzle. The best have minimal number of clues. I recall an infmaous one which
was about a farmer and the sizes,perimeters and areas of his fields. Really hard and challenging.

Now inspired to google...

In Games magazine and all of the various spin-off puzzle books that they publish, these are called "Code Crosswords". I've got some right here on my desk, so if you're still interested (and didn't already investigate it), I'd be willing to do some geeky research with you. :D

I'm not sure how accurate this frequency analysis thing would be though. Part of the problem is that one rule of Code Crosswords is that every letter must be used at least once, and I suspect that makes the number of occurrences of each letter unproportional to the number of occurrences in everyday language. Also, I believe the intended method for solving these puzzles is to fill in the letters they give you, then look for partially filled in words for which there is only one obvious correct answer. Therefore, the puzzle will contain distinctly different words from normal vocabulary.

I'd forgotten all about this entry. I think that the name I knew the puzzle under was cross references, or something like that. And yes, the method you state is the intended method -- that's how I always used to do them. You're probably right about the difficulties in doing a frequency analysis thingy for them. Though I would be interested to hear how it works, if you feel like putting the time in.

I have an extremely large to-do list forming in a text file on my computer. It is a roughly categorized compilation of every stray to-do list I find in the process of cleaning my room. I shall now add "frequency analysis of code crosswords" to this list, and promptly forget about it until 2017. :D (By which I mean I really do want to try it out because I'm curious, but I just know I'll put it off indefinitely.)

  • 1