Listener No. 4378: Present Day By Horvendile

[Contrarian]

I'm surprised that there isn't a thread up for this yet. (Apologies if there is one)

I started on it this afternoon, having only returned to the UK this morning.

Just a couple of loose ends to tie up, but most of it was pretty straightforward.

I loved the idea of including the code-square within the grid itself.

Thanks Horvendile.

[fyellin]

@emcee, @scorpius. I'd love to know your technique(s) for solving playfairs without the use of a computer.

The problem is that computing makes it too easy in a way that can't easily be reproduced by hand. There are about 31k words in the BRB that have no repeated letters. If I know that the L of LISTENER encrypts to the same letter as the O of HORVENDILE, then I've reduced my possible keys to 350. These 350 words have nothing in common except that they all create appropriate squares. Only a few more such clashes, no matter how minor, and I can solve any single-word playfair.

[Cruncher]

23ac and 41 ac and Quinaplus Word Matcher were enough to sieve out the codeword/phrase for me. Would have struggled to do that by myself though. Now look forward to an awful lot of encryption. Certainly not too proud to use an online tool. Thanks above for the recommendation. Happy New Year to all as and when

[Scorpius]

Fyellin, if a code is not a dictionary term I don't see how generating lists of words or phrases with no repeated letters is going to help. Initially I did use computing power, but got nowhere with it, so decided logic rather than brute force was needed.

My approach was to work out the placement of two letters in the first line of the Playfair grid, one in the third line, three in the fourth line and all in the last line. On top of that, I deduced at least two additional letters that had to be part of the code. Lastly I was able to determine the columns in which two other letters must be. It wasn't hard then to see some likely possibilities, the first one of which I was able to confirm very quickly. There was some lucky guesswork; had the code been very obscure I might have had to resort to computing power.

I should stress that I did use computing power to solve some of the remaining clues that I hadn't solved, particularly the unclued entries.

[upsetter]

Some jolly clever and difficult clues to cold-solve, but we think this puzzle was more of an impressive feat of composition than anything else.

[Speravi]

I may be mistaken, but it seems to me that there are two possibilities for the unclued entry in the bottom row. They share the same second letter and the final five letters are the same, but both P*M***** and (the more obscure) S*L***** appear to fit the bill. Am I wrong?

[perseverer]

Speravi - are you allowing for the encoding?

[Speravi]

Perseverer, I was giving the possible answers prior to encoding - apologies for the confusion.

[emcee]

Fyellin,

It's possible to use the clashes in cells to produce an equation as the pairing across and vertically must produce the same letter when encoded. So, as I mentioned in my previous post, I had an 'equation' of the form R C (dot) = E (dot) S for each clash. The dots represent whether I should take the first or second 'blank' corner of the rectangle. Now R C (dot) = E (dot) S isn't very useful but R C (dot) = R (dot) E is useful as constructing the rectangle with R C as the leading diagonal and looking at the corresponding rectangle for R and E tells me that E and C must be in the same column.

It's then a matter of looking to see from the list of 'equations' formed, what to use next to build up the overall picture. I did this for about an hour and could then have a fair guess at what the codeword was (I was able to deduce the top 3 letters in the second column pretty quickly). The remaining unused clues were used to check that the resulting codeword grid was correct.

Solving this Playfair was a tougher challenge than the normal Playfair as we normally have a pair of letters and the corresponding encoded pair which makes the construction of the code grid much easier (and quicker).

Good luck!

[emcee]

Sorry, I meant R and E must be in the same column. (not C and E)

[fyellin]

@scorpius, @emcee: Thanks for the info.

@scorpius: I was purposely being vague about what the key actually is, and just giving an example of how only a few pieces of information are sufficient to solve a Playfair when you have a dictionary and a computer.

I gave my program six intersections (9D and six crossing words), and some general rules about possible keys. I started it running, and went off to see Star Wars.* When I returned several hours later, there were about 20 keys printed out. Only one really made sense. I tested it on a few more intersections and was happy with the result.

[*Seeing Star Wars on Christmas afternoon was a bad idea. The theater was packed!]

[tnap]

I spent the best part of a day following emcee's approach, with bits of paper everywhere - all to no avail. Then 2 pennies dropped - firstly, there's a much easier way of finding the codeword; and, secondly, I'd managed to find the wrong answer to 23A.

[olichant]

What a cracking year of Listeners it's been. It'll be interesting to see which one wins the prize for best puzzle. If it were down to me (which it certainly isn't) I'd find it hard to choose between Mixed Doubles, Triskaidekophilia, Quads and Russian Roulette...

Thanks to the setters and editors for all the work that must go into these puzzles. And hope everyone here has a happy New Year.

[Scorpius]

Fyellin, you're obviously far more skilled than I in the use of a computer. I wouldn't know how to begin doing what you did. What interests me is, would your method work if the code were, for example, FIVE GOLD BATS?

[fyellin]

@Scorpius: Excellent question.

My program only tried the possibilities I asked it to try. On its first run, it failed very quickly. I expanded my search, and on the second run it failed slowly (30min). I expanded my search even further and it found the answer. It would not have found FIVE GOLD BATS.

I did use the fact that I knew one specific less-than-common letter must be in the key.

I wrote this Playfair code years ago. I've made no attempt to make it efficient. It's written in python and I ran it on a MacBook Pro.

[1581960]

Hi tilbee, you’re not alone with AB site problems, slow response and lock ups are plaguing me as well, I’m also using IE and Chrome. Don’t think it’s a computer problem, AB is the only site I have trouble with.

[fyellin]

D'oh. All makes sense to me. I see that solving the Playfair was much easier than I thought.

[icynorth]

My first experience of a Playfair-encoded puzzle, and I'm not convinced it justified the many hours it took trying out different keys. The one that worked was, I thought, a bit 'loose', and could have been given some additional indication.

I enjoyed the logical challenge of completing the grid, but the final product just isn't pretty enough for my liking. Thanks anyway, Horvendile.

Oh, my favourite puzzle of 2015 was Conduit - the only one with a genuine PDM for me.

[aldanna]

We are nearly to the end of this puzzle, much interrupted by Life happening. I cold-solved most of the clues, and Himself did things beyond me to come up with the square. Just a couple to go now, plus the unclued entries. The opportunities to make errors in this one seem very numerous, though!

[petefoxes]

I had a feeling we'd have a Playfair to finish the year, and it was tough, and borderline enjoyable.

I started doing Listeners in 2013 and I managed 4/4 correct submissions. Last year I got 12/13. This year I have ?/51 (I didn't properly attempt Magic Cuboid - not knowing at the time that I'd finish all the others) and now I think I might stop to prioritise other things.

FWIW, my favourites this year were Mixed Doubles, Quads, Identity Crisis, and in particular Golf..., which wasn't highly thought of in this parish but to my mind had some excellently thematic clues.

[Philoctetes]

I am surprised more haven't commented on the effective proliferation of unches, normal or implied. As has been noted, 8d consists entirely of such beasts, and even such as 26d do not have any of the real benefits of a crossword. There are very few answers where cold-solving is not the most likely option.
That said, it took me about 40 minutes to work out the code on the basis of the few external answers. "Have His Carcase" by Dorothy L Sayers does contain an excellent description of how to do it.