December 2016 — This Squeamish Ossifrage Society site is moving to The move will be completed by December 31, 2016, at 10 p.m. MT. The site you are currently visiting is being dismantled and will no longer be supported.

Thursday, December 4, 2014

Bag o' Zyps

Here are three of Cryptopop's ever-popular, challenging Zyptograms. The first one is relatively easy, as far as Zyps go. The second and third are medium.

See if you can get all three. And then watch for two more Zyps -- even more difficult than these -- to be posted within a few days.





jdege said...

My first approach for aristos is usually pattern words. For a surprising number of cons, there is at least one unique word. There are very few that don't have a pair of words that have one unique joint solution.

The first problem was of that sort. There was only one possible for ujagupk/gauju. So it didn't take long.

The second problem wasn't much harder. There was only one possible for ndcmboc/notbod.

The third was nasty. My scripts kept trying to tell me that iqffao was "sappho", which made "soaps" and "saps" work, but made ioqfzagr "soap_h__", and that just didn't fit.

Cryptopop said...

Do you find the patterns manually or do you use a P/M program?

jdege said...

I used a sophisticated pattern-matching program called "grep".

And a canonized word dictionary I originally created after I first read Jon Bentley's "Programming Pearls".

Cryptopop said...

I'm familiar with the "grep" program. It's good but difficult to use. I discarded it long ago.

There are better programs such as Decrypto 8.5 which not only solves cryps with and without word divisions but can be used as a p-m program as well. I'd put its dictionaries up against any other program.

But what's the point? Solving sim subs manually is challenging, mentally stimulating and fun. Using a program, except for verification purposes and checking the uniquity of composed cryps, is a mechanical exercise. Where is the
satisfaction in that?

jdege said...

I've played with dekrypto. I was amazed at how fast it was compared to my own efforts at hillclimbers.

As for grep, I'm a strong believer in general-purpose tools, from which more specific applications can be built. The scripts I've built using grep are less automated, and require manual choices. And as a result, they sometimes find solutions when fully-automated programs run down the wrong path.

As for what's the point? Depends entirely upon what you are looking for. I sometimes use manual methods, sometimes my interactive grep-based searches, sometimes I resort to various heuristic programs. I've yet to find an approach that has worked on every cryptogram.

Cryptopop said...

As to your last statement I agree. For example, standard
cryptoquotes and cryptoquips are usually solved by finding high frequency or common pattern words together with application of general letter frequencies. Other cryps, such as my Zyptograms are usually solved by vowel placement techniques, digraph analysis, grammatical reconstruction, etc.

However, my point was that Zyps are puzzles, nothing more, nothing less. A puzzle is a mental challenge and to solve it mechanically takes the fun out of it. As far as actual code-breaking is concerned, it should be approached by whatever methods are available, expedient and pertinent.

jdege said...

From my point of view, the real challenge is creating a text that can't be solved using mechanical techniques- to intentionally use word choices that will send the mechanical methods down the wrong path. Many of the top-five aristos in each issue of the Cryptogram accomplish that.

So, for that matter, do your zyps.

And yes, I do try vowel placement, etc. Though in truth, that's something I use more when I don't have word separations.

For what it's worth, the automated programs I tried didn't produce anything like the correct text for #3. Neither, for that matter, did my scripts. What they did was to reduce the number of possibles for iqffao.