# Sample commentaries on puzzle creation

### Update: Wrap-up posts have been approved by the community and may be posted as answers at their puzzles. No need to first post them here anymore.

Examples of wrap-up posts may be found by searching for keywords such as wrap up making poser.

Original post (with minor edits) follows

This  is  was a temporary sandbox for posting sample commentaries on , to whet the idea described in the post about sharing and rewarding what went into making a good puzzle.

The objectives of this sandbox are were to:

1. Confirm the usefulness of this idea
2. Standardize on a convention to caption these special "self-answers" to differentiate them from regular "solution" answers
3. Formulate content guidelines, e.g. level of detail, what people find interesting, things to avoid ...
4. Formulate style guidelines, e.g. spoiler-tagging so as not to reveal the solution (as with regular answers)

Post a behind-the-scenes commentary about how you created your puzzle(s) as an answer to this question (one answer per puzzle). Provide a link to the puzzle itself in your answer. Other than that, post as you would if it were a self-answer being posted on the puzzle's page itself.

Comment on others' commentaries to provide specific feedback about what aspects you found interesting, or didn't.

There is a separate thread on discussing what formal standardization should be used in such wrap-up posts. These are not yet agreed on, so the sandbox answers might or might not adhere to some of them.

When the idea is finalized, a separate meta-post explaining the purpose and listing the guidelines can be created, and sample commentaries from below can be copied into their actual puzzles (assuming that we don't decide to disallow bumping old puzzles).

• Not sure if I like the idea of posting explanatory self-answers, but I do like this meta post. Sandboxes for community feedback and discussion are great (as long as people aren't prevented from posting puzzles on the main site because of them) :-) Sep 7 '16 at 20:13
• @randal'thor See my answer on the original proposal post. I think the key is, that this self-answer should only be added when the puzzle has an accepted answer. Sep 8 '16 at 10:46

This is an answer that would potentially be posted on my most-popular puzzle, Hearken, now, and listen close!

Meta-note (this is in response to the question above, and not part of my actual answer):
I have included spoilers in this answer, in response to point #4 in the original post above. This is mostly to demonstrate that it is ineffectual to do so. The post becomes extremely fragmented (or, alternatively, becomes one huge spoiler). Note that these answers will never be the accepted answer, and thus will never be the first one displayed. The will also always be posted after the solution has been found, so in my mind spoiler-tagging them is unnecessary and cumbersome.

This is not a solution to the puzzle, but rather a commentary on how this puzzle came to be, for those who are interested.

CAUTION: This answer contains spoilers for the puzzle.

Being familiar with

I thought it would be fun to create a puzzle containing them. I thought I could create a story where each paragraph

left out a certain letter,

and those letters combined to provide the solution.

But then I realized that, especially with less-common letters, it wouldn't always be obvious

which letter was being excluded.

This led me to the realization that every paragraph would have to

contain every letter of the alphabet, except the one being excluded.

Being an aficionado of rhyme and poetry, I decided to create the puzzle in verse.

I initially thought that the puzzle might be too difficult without any hints, so I decided to build a hint into the puzzle by making the

first letter of every line combine to provide the hint.

Partway through, I realized that would be too difficult, so instead I just made it

all the uppercase letters,

which proved to be a little more manageable.

I came up with a phrase that I thought would be a good solution to the puzzle, and then got to work.

I found some online resources, including:

• A rhyming dictionary
• (this was useful in helping to find rhymes to complete each stanza)
• A thesaurus
• (this was useful when I needed a word, but it contained the "forbidden" letter; I used a synonym without that letter instead)
• Lists of words containing Q, X, Z, and J (these were helpful in finding not-too-obscure words that would "use up" these letters so that I could get every letter into the stanza)(words like "just, enjoy, equal, quiet, quite, quick, fix, text, excel, lazy, gaze, puzzle)

• I set up a spreadsheet in Excel that would

count the number of each letter in a paragraph, and tell me which letters did not yet appear.

Then I started composing the rhymes. I was somewhat constrained by the requirement to

have the uppercase letters spell out the hint,

but I could be a little flexible in that regard by abusing punctuation -- especially semi-colons.

I didn't have a clear idea of what the poem was going to contain when I began -- I just kind of "winged it." I noticed after about 4 stanzas that the subject matter was becoming a little redundant, so I started trying to change it up a little, by leaving some oblique hints.

When beginning each new stanza, I tried to

get a number of words containing less-common letters like J, Q, V, X, and Z

into the first couple of lines, so that I wasn't struggling to

cram too many letters in at the end.

When I got to the second stanza, I realized I had a challenge ahead of me, because I needed to

include every letter except U, which means I needed to figure out a way to get Q in there without an accompanying U. There are a few words that have Q without U, but none is particularly common, and they would stick out like a sore thumb.

I eventually came up with the idea of

referring to the letter itself, in the common phrase, "mind your P's and Q's."

I think it worked out quite well.

It probably took about 4 hours over the course of an evening to construct all 11 stanzas. I revised things several times until I was satisfied with the content, metre, and rhymes.

I learned afterward that I didn't need to build the hint into the puzzle (although I still think it added something to the puzzle). It was originally solved without the hint, which was only discovered after the fact. People on this site are good! (Although in retrospect, it probably wasn't as hard as I thought, either.)

• Can I just say, this puzzle was effing brilliant. (I especially liked it because I've used the same idea - not on PSE, in a piece of creative writing years ago - of writing a poem containing exactly 25 letters of the alphabet, the missing one being a clue to a vital piece of information.) Sep 8 '16 at 19:11
• Good example of how wrap-up posts should look like. I don't mind the spoilers, but they are indeed not needed - as long as the initial startup of the post (which is formalized) is long enough to prevent "accidental reading" and the "final solution" isn't visually popping out when glancing over the post. Sep 9 '16 at 6:53

# Wrap-up: the making of Reassemble the riddles!

This is not a solution to the puzzle, but provides the puzzle poser’s thoughts on its creation. This type of answer has been approved by the community.

Caution: This post may contain spoilers.

I had these four riddles written down so long ago (at least a year) that it's hard to recall exactly how the idea took shape. But it all started with a riddle composed by Avigrail for the Riddlers' Den (this riddle is now #28 on the site1) which contained the lines

No hands at all
Yet I wave to you.

When I first saw this riddle, I assumed that a "wave" without hands would probably refer to a different sense of the word "wave", so I immediately thought of the sea. But it eventually turned out that this part of the riddle referred to a dog.

This got me thinking about how exactly the same lines and clues could mean different things in the contexts of different riddles. I wondered if I could create a set of lines which could be put together in two different ways to form riddles with totally different solutions. With this in mind, I created the following short riddles with the solutions sea and dog:

1. A faithful friend,
With a nose for all.
I can wave to you,
Without hands at all.

2. Most of the world
Can feel my salt.
I can wave to you,
Without hands at all.

Well, now we've got two duplicated lines and four unique ones. How else can we use those four lines which haven't yet been reused? Looking at the first line, "A faithful friend" immediately made me think of a spouse, so I cobbled together the following riddle with the solution spouse, which reuses one of the lines from the 2nd riddle as well:

1. A faithful friend,
Contracted to love.
Most of the world
Ends up with one.

Then, looking at the "salt" line with salt water still fresh in my mind, I immediately thought of tears. Conveniently, tears are found close to noses, which were mentioned in the 1st riddle. So I cobbled together the following riddle with the solution tears, reusing lines from the 1st and 2nd riddles:

1. With a nose for all
Of us to run down.
Those in woe or joy
Can feel my salt.

At this point, I walked away from the idea to do something else. The four riddles quoted above remained, unchanged and gathering metaphorical dust, in my repository of puzzle ideas for over a year, because I wasn't really sure what to do with them.

Eventually the 16th Fortnightly Topic Challenge came along. I spent a while pondering what I might be able to contribute to this, as I'd never written a question before (my only encounter with the tag had been by answering one). Since is my top tag and best claim to fame, it was natural for me to wonder if it would be possible to combine these two tags in some way. But how?

I can't remember exactly how my thought process went, but somehow I recalled the line-sharing riddles I'd written way back in 2015 and had the idea of providing a list of just the lines and requiring solvers to reassemble the riddles. I was doubtful at first whether this would really count as , but other puzzlers persuaded me to give it a try. I still needed a few days to work out exactly how to construct the puzzle, as I wasn't sure exactly how much information to provide: should I just list the lines and say "form these into some riddles", or say how many riddles, or say which lines were duplicated? I dithered over it for so long that I posted another question and a few more puzzles first, but eventually Reassemble the riddles! was posted and solved and reached the first page of the Hot Network Questions. For the record, my final version of the four riddles was slightly different both from those above and from Sconibulus's solution:

1. "A faithful friend,
Living by nose.
I can wave to you,
Without hands at all." = DOG

2. "In most of the world
You can feel my salt.
I can wave to you,
Without hands at all." = SEA

3. "A faithful friend,
Contracted to love.
Through woe or joy,
I'm bound to you." = SPOUSE

4. "Living by nose
Which we may run down.
Through woe or joy
You can feel my salt." = TEARS

Honestly, I'm still not sure whether I got the balance of difficulty right: I was expecting it to last a few days, but it was solved within ten minutes. Whether I made it too easy or Sconibulus is just too clever, I have yet to work out.

1 I'm not sure whether I'm allowed to link to the Riddler's Den here, since it's a site I co-own. I included a link because it seems to be relevant, enabling you to see the original riddle that inspired the puzzle being discussed here, but feel free to edit out the link if you feel it constitutes excessive self-promotion.

Wrap-up, the making of: The Case of the Missing rebus

This is not a solution to the puzzle but provides notes from its poser. This type of answer has been approved by the community.

Caution: This post contains information about the solution.

Often (almost always?) construction of a puzzle presents a more challenging puzzle than does the resultant puzzle. Sometimes the puzzle of construction itself can neatly infect the finished puzzle.

In this case, leaving a part of the puzzle incomplete improved it. The original idea was a two-dimensional pattern of rebuses whose numerical and verbal interpretations were distractions from each other, for a Stroop-like effect. This would’ve been acceptable for posting, with multiple clues from two angles, but 3 rebuses remained to concoct with only enough ideas for 2 of them.

$$\kern-2em \small \color{#274}{\boxed{ \quad \begin{matrix} \\[-1ex] & \kern -4em\raise2ex\color {black}{\boxed{\matrix{\sf \raise.4ex{same}\\[-1ex] \sf \raise.4ex{2~ideas}} }} & & & \kern-1em\rm LLL \\[-1ex] & & & & \kern-1em\rm LLL & & \kern-1em x \, y \\[-1ex] & & & & & & \kern-1em z \:\: t \\[-1ex] \rm XCXC & & & & & & \\[-1ex] \rm XCXC & \raise -2ex\color {black}{\boxed {\matrix{\sf\raise .4ex{same}\\[-1ex] \sf\raise .4ex{2~ideas}} }} & \kern-1em \raise-5ex \color{black}{ \boxed{\matrix{ \sf\raise.4ex{same} \\[-1ex] \sf\raise.4ex{2~ideas}} }}\kern-2em & & & & & & & & \kern-1em\raise-8ex{\rm Sn} \\[-.6ex] & & & & & & &\kern-1em ~ 2e/6 & & & \kern-1em \rm Sn \\[0ex] & & & & & & &\kern-1em e/2 {-} \!\; e/6 & & \\[-1ex] & & & & & & & & & \kern-1em\rm TT \\[-1ex] & & & & & & & & & & & \kern-1em\rm ZZZZZ \\[-1ex] & & & & & & & \raise{-1.7ex}\strut & & & & \kern-1em\rm ZZZZZ \\ \end{matrix} \quad }}$$

💡 So why not just be lazy and leave one rebus blank, as the pattern is well established without it, and direct the challenge toward filling it in? This would have serendipitous benefits.
•    Add meaning to solving the originally conceived puzzle.
•    Specify the challenge more precisely than “what’s going on here?”
•    Provide an opportunity, with ready examples but no pressure, for solvers to be inventive.

💡💡 Aesthetically unsatisfying gaps in the midst of the layout could also be filled in with possible places where the missing rebus might belong. This had a surprise benefit as well.
•    The numerical and verbal patterns must be explicitly solved, not merely sensed.

These 💡💡💡 added natural levels atop a puzzle that already had nine mini-puzzles by which to gather momentum while solving.

I've posted wrap-ups for several puzzles already. I stuck them in as answers. I haven't done Riddled and Dismembered yet, so I'll post it here.

There was a meta post that asked for riddles whose answer was flower. I did one and it was ok but I felt as I did it that it was "same old". I wanted to do something different. The feeling that I hadn't done justice to the flower theme made me stick with that as a subject.

I had the idea of hiding the parts of a flower in a poem. I had done something with that basic idea for the chess fortnightly challenge. It had to be something different. I googled "parts of a flower" to see what I had to work with. At first sight it seemed like a pile of crap. Weird-looking, awkward, stand-out words: "pistil", "stamen", "anther", "ovule", "style", "filament", etc.

I was about to go in search of another idea when the phrase "top is tilted" popped into my head. That seemed like a pretty natural and nonobvious way of hiding "pistil". I had a picture of a petunia in my mind with its bell-like flower tilted toward the sun. Bell-like, the bell of a trumpet, trumpet shaped... "Trumpet almost" sounded good to me as well. It was a start.

At that point, having the picture of the petunia in my head, I figured I was writing a poem about a flower with the parts of a flower hidden in the poem. I liked the idea. I started looking for similar pairs of words to hide flower parts as backups in case I couldn't work the first two into the poem. Ironically, what decided me to try to do six lines all with the same device was when it occurred to me that I could do "anther" with anything ending in "ant" followed by anything beginning "her" like "important here". In the end, I used a different word break. "Ovule" and "ovary" straight out sucked. "Style" led to "nasty lemon" and other unworkable options.

Eventually I had a pretty good list of possibilities.It was a drag that "stamen" and "filament" both contained "amen". I tried whenever possible to use hiding words where the letters have a different sound.

I started putting phrases together and looking for rhymes. There wasn't a lot of space left over. In particular there wasn't room to say "I'm a petunia" or "I'm a flower". That's when I had the idea that I should have had in the beginning: make it a rhyming riddle instead of just a poem.

I was fairly happy with the outcome. It reads as an actual, coherent rhyming riddle. "Form", "mourn", and "adorn" don't quite rhyme but somehow I quite like the half-rhyme effect. Likewise, the imperfection in the meter of the second last line is, to my ear, more of a pleasant variation than a jarring switch.

I do rather regret the fourth line. I was listing the different ways we use flowers: to make someone happy, to say sorry, to add a pleasant scent, or to express sympathy. I think I should have left the last for the final two lines where the mood changes. If I could go back and change it, I would make the last three lines:

Your grief I lament. Slain, I will
Your last amen attend and mourn.

Or something like that so that the third fourth lines are more of a unit.

I was tempted to post it without instructions and say: "No, that's not the solution" if they just solved the riddle. I believe the puzzle as posted is better but maybe people would have enjoyed figuring that out. What tag to use was a problem. I think I originally only used "riddle", thinking that steganography was too much of a giveaway.

There were a lot of bits to put together. Coming up with a title was the hardest single step. In the end I think the title had a lot to do with its popularity. Short, dramatic, intriguing (hopefully). It was solved fairly soon after posting and most of the up votes came after there was an accepted solution. Looking at it as objectively as I can, it seems to be pretty good as a rhyme, not bad as steganography and not very good at all as a riddle. Somehow, this seems to have been the right combination.

I believe my biggest takeaway from this one is that a puzzle doesn't have to be killer hard to be popular. Working on the rhyme, meter, and sound or look can help a lot.

Since it was asked for, here are some puzzles with wrap-ups already: My Puzzling Girlfriend, A Puzzle That Dims a Tornado, No It's Not a Polynomial, Thinking Outside the Box, A Puzzle That Doubles Back Obscurely

• Hadn't seen this one, so glad it got resurfaced here. Similar idea to the one I posted in that same flower thread, but executed about 1000x better. Very nicely done. Making meter/rhyme and a coherent riddle all work while successfully hiding things in this way is hard work. Sep 12 '16 at 5:48
• @Alconja I really should have included a shout-out to you in the write up. I definitely had your riddle in mind: the idea that the flower dies to honor the dead and the idea of contrasting a flower's use in courtship to its use in mourning were both stolen from... I mean inspired by your work. Sep 12 '16 at 7:01
• :) honoured you found inspiration in that one, but, I meant my other entry, where I hid flower names in each line, which is structurally more similar to yours. Though as I said, no where near as good, and it almost completely loses it's "riddle" nature in lieu of nearly pure steganography. Sep 12 '16 at 7:08
• @Alconja Huh! I didn't see that one. That is close. I probably wouldn't have done mine if I had known about it. Sep 12 '16 at 7:58
• Glad you didn't see it then. :) I guess great minds just think alike. Sep 12 '16 at 8:14
• Just noticed, Hugh Meyers, seems that you never posted this as an answer at the puzzle itself. This belongs there by now and it's never too late.
– humn
Aug 22 '17 at 14:00
• @humn Done. I wasn't happy with this write up so I redid it with the official template. Thanks for the prompt! Aug 24 '17 at 20:07

Sample for: The box with the curious inscriptions

# A closer look inside the box

How the puzzle was sculpted

## The idea

This was one rare occasion of a surprisingly short jump from thinking, "Hmmm... So what's my next puzzle going to be about?" to an actual idea within a few minutes. I looked in the mirror and saw that my T-shirt had a picture of an:

ice cube

A short train of thought later, the basic idea had taken shape.

## Building it up

The next day, I started making a list of suitable clues (with a bit of help from Wikipedia), and eventually ended up with this table (with only the first and last columns initially):-

. ratio *42 vol% clue

pyramid 1/3 14 33.33 food
cone pi/12 11 26.18 ice cream
cube 1 42 100 sugar / ice
sphere pi/6 22 52.36 bio / hydro // public / celestial ... [*]
cylinder pi/4 33 78.54 gas / oxygen

#prism 1/2 21 50 ? #roof / #wedge / #Dove

[*] Full list: bio / hydro // public / celestial / Web // social / litho / meso / exo / of influence

By suitable I mean that the word associations had to be common enough to eventually strike a chord, yet not as individually strong as, say, "Egyptian". The spheres was a bit tough because all choices seemed a bit sub-par, especially with many of them just being prefixes and not proper words. The "bio gas" juxtaposition helped with that.

## Changing the type of puzzle

With the series in place, I decided I didn't want to just do another "What links these words together?", and so at least a supplementary attribute needed to be attached.

I chose a number which spoke volumes about its object. For this it was necessary that the different number of variables in each be normalized in some way. Thus by necessity did they all come to be inscribed in the "box" of the lot.

The initial idea was to use the percentage (fourth column) - rounded to an integer value - but that seemed uninteresting. Taking the LCM of the denominators in the second column (after substituting a well known rational approximation) yielded the third. And then it was clear which one was going to be "the answer" to be replaced with a "?".

## Weaving a story, title and hints

With this much, I was ready to go ahead and post the puzzle as a simple table. Maybe...

food ice cream sugar bio gas
14    11    ?    22    33


Can you find the missing number?

"Meh, sure I can... I think...
(And if I can't, well it probably wasn't an interesting concept anyway.)"

Some kind of flavour seemed to be in order, not just to make it interesting initially, but also to be able to unreel hints gradually.

After some thought, the title was born.

Now the perfectly natural thing to expect of a "box with curious inscriptions" is that it would posses great mystic powers and be unearthed on a moonlit night in some ancient ruins, preferably in Egypt! But more wordplays appeared on the horizon and consequently the box surfaced in a modest library instead (still had to be the basement). ... And not to mention that "bio gas" and "ice cream" would sound ridiculous on a box of great mystic powers:

... With quivering hands, Le Carre brushed aside the sand and read the inscriptions. "Good grief, Higgins!" he exclaimed, "It's the dreaded curse of the lactose intolerant Pharaoh!! Now we must find the answer, or face a fate worse than death!!! ..."

I can't recall at what point I came up with this second table, perhaps it was a residue from the "building it up" stage:-

square 14 perfect // Times
circle 11 inner / magic / vicious / full // crop / traffic / great ...
triangle 7 love // Bermuda

But I worried that it might make the puzzle much easier too solve, so I saved it for a backup hint.

Speaking of backup hints: In hindsight, maybe I could have waited a little longer for the puzzle to be solved, or gone ahead first with one of my other backup hints - which involved conversation bits between the librarian and the police such as:-

• Discussing how the case was taking shape
• The police saying that if enough measures had been taken then she wouldn't have had to ask such questions
• "But officer, why would an educated person do such a thing?!"
"Well, ma'am: Some rise by sin, and some by virtue fall."
[cough, cough]

But I thought they would be as oblique as what had already gone before. So I finally chose to use my knowledge of the Dewey system from my days of being a librarian in school, which only had the effect of placing the subject one Google search away.

• Great to see the fun pieces for this puzzle that got left out (of the box, so to speak).
– humn
Sep 12 '16 at 19:45
• Just noticed, KeyboardWielder, seems that you never posted this as an answer at the puzzle itself. Never too late.
– humn
Aug 22 '17 at 14:01

I'm posting this not because this is my best puzzle, nor the one I'm most proud of, but because it grew from a germ of an idea and almost shaped itself and I learnt a few important lessons from it.

# Wrap-up: The Making Of << Something for the Real Fans >>

This is not a solution to the puzzle, but provides the puzzle poser’s thoughts on its creation. This type of answer has been approved by the community.

Caution: This post may contain spoilers.

### Inspiration

This is the third puzzle I posted on this site, over a year ago now. At the time there had been a few cipher-based puzzles posted, using ciphers I had never seen before, so I was interested in building one into a puzzle.

Of course several ciphers require some sort of key, so what to choose and how to lead the readers to it?

I have always been a fan of the film Jumping Jack Flash and the line "the key is the key" was firmly ingrained in my memory. This led me to wondering how I could use a musical key from some other piece of music as a cipher key.

### Creative Steps

I needed a piece of music that would fit the following criteria:

• Reasonably well known
• No key-changes in the piece
• No cover versions or alternative arrangements in a different key

It was this last criterion that made me think: "how can I guarantee that any version of the piece of music would be in the same key?"

"Air on a G string" seemed, to me, to be an obvious choice. The fact that the Air came from a suite (JS Bach's Orchestral Suite No3 in D Major) that was all in the same key led strength to my feelings that this was the right piece of music to choose. So now I had a piece of music to tie into the puzzle.

Looking for ways to tie Bach back to the Rolling Stones, I came across the fact that there was an brand new museum exhibition (called Exhibitionism) about to go on tour. I searched to see if there was a Bach exhibition anywhere and Bach House (Eisenach) jumped out at me. I then had the idea to add a layer of complexity to the puzzle by encoding the geographical coordinates in the cipher.

All I had to do then was to come up with clues to link to the film, the music and from the physical museum to the mobile museum exhibition.

### Resources

I used an online Vignere cipher programme (I can't remember which site) to encrypt the text. As it happened I made a very poor choice, as the programme I used introduced a one-letter shift. Of course, when I checked that the code was decipherable I used the same tool and everything came out fine.

Len decrypted the text within 24 hours and drew my attention to the fact that the vignere was incorrect.
I quickly replaced the cipher text.

### The Solution

There was less than a week from the first posting of the puzzle to the final solution. Of course, when you're fairly new to the site and eager to get feedback and have someone solve your puzzle this can seem like a long time. As a result I added in one hint which made the puzzle a little easier to crack. Maybe if I was posting this puzzle today I would wait a little longer before adding a hint.

### The Title

I started off with the title "How good is your film knowledge?", but realised soon after posting that I was sending the wrong message - the puzzle wasn't really about movies. It relied on one fact about one movie.

I changed it to "Something for the real fans".
This related to the fact that you would need to be real fans of the Rolling Stones to know about the exhibition. You would also need to be real fans of the film for the names of the characters to register.

I still think that the title might have put some people off the puzzle, but even now I'm not sure what title would work well instead. Any suggestions?

### Takeaway

The first and most important lesson I learned was that when you use an encryption tool off the internet that you haven't used before, make sure you check the results using a different site!

I didn't want to make the puzzle too tricky, so I used a simple cipher with a fairly long piece of encrypted text. As it turns out, this made it far too easy for people to crack the code first, before working out the meaning of the clues. It also made the puzzle visually unappealing as it meant there was a large, ugly code block dominating the text.

Overall I enjoyed the process of putting this one together and I was glad to see a good, comprehensive solution provided, but I've produced better!

• Just noticed, Gordon K, seems that you never posted this as an answer at the puzzle itself. This belongs there by now and it's never too late.
– humn
Aug 22 '17 at 13:59

Meta note: This puzzle’s solution/answer did not mention how it was solved, which leaves the puzzle looking like a sneeze cipher and has bugged me ever since.
So I’m now thinking of making this wrap-up out of the worksheet originally provided as a hint in the form of a partial answer (a maneuver that has evolved into self-initiated community wiki posts that accompany some of my puzzles.)

Wrap-up, the making of: Some string of letters

This is not a solution to the puzzle but provides notes from its poser. This type of answer has been approved by the community.

Caution: This post contains information about the solution.

The mystery encryption scheme was devised to require trivial effort to encode and decode, with a twist intended to defy most automated deciphering techniques. The twist is that...

...each original message letter can be encoded as 26 different encryption-letter pairs, so that a message with 200 letters can generally be encoded without repeated pairs. A longer message can deliberately repeat pairs for less common letters, such as consonants early in the alphabet, to serve as distractions from inevitably repeated pairs that will represent frequent letters, such as E and T. [Anyone know a name for this scheme?]
The pairs used in the puzzle’s cipher were intentionally selected to provide clues.

As the accepted answer does not mention what led to solution, here is how clues were meant to work. Among other considerations, solving any letter-sequence cipher includes two initial steps.

Step 1. Convert each letter into its alphabetic position number.
Step 2. Count how many times each letter/number appears.

A chart that combines these two steps for the cipher in question reveals an odd distribution.

                                                                 z
z
z
z
z
z
z
z
z
a                                                 z
a     d                                           z
a     d                                           z
a     d           j                               z
a     d e         j                               z
a     d e         j                               z
a     d e       i j                   t           z
letter         a b   d e   g h i j                   t           z

alphabetic #   1 2 3 4 5 6 7 8 9 10  .   .   .   .  20          26

count          8 1   7 4   1 1 2 5                   2          17
:.................:                   :...........:
Almost all letters are               The only outliers
early in the alphabet.               are 20th and last.


The noticeable clustering was meant as a compelling clue while a subtler clue is that the total number of letters is 48, which reeks of doubling, and that the letter counts divide into neat halves.

letter         a b   d e   g h i j                   t           z

count          8 1   7 4   1 1 2 5                   2          17
:...............: :...............................:
The left portion        The right portion also
includes 24 letters.       includes 24 letters.


Compare all that to how this distribution could have easily been made more cryptic.

                       e f g h i j k l m n o p q r s t u v w x y z
letter         a b c d e f g h i j k l m n o p q r s t u v w x y z

alphabetic #   1 2 3 4 5 6 7 8 9 10  .   .   .   .   .   .   .  26

count          1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2


The actual distribution was meant to draw attention...

...to the letters j, t and z, along with their suggestive alphabet positions of 10th, 20th and last.

The next two steps were meant to follow naturally.
Step 3. Highlight the interesting letters.

                  zz  z jj z  j zz  zz  z jz  zj  t z tz  zz  zz
b  ea d  i ai a  dd  ad e  ad  ga e a  ed  ha  d  

A very even dispersion emerges, meant to strongly suggest...

...letter pairing, which further reveals that each pair includes exactly one j, t, or z.

      | z|z | z| j|j |z | j| z|z | z|z | z| j|z | z|j | t| z| t|z | z|z | z|z |
|b | e|a |d | i| a|i |a | d|d | a|d |e | a|d | g|a |e |a | e|d | h|a | d| 

Step 4. Tally...

...the letter pairs.

                                              z a             z d
z a             z d
z a             z d
z a             z d     z e
j i     t a     z a             z d     z e
j d     j e     j g     j i     t a     z a     z b     z d     z e     z h

10 4    10 5    10 7    10 9    20 1    26 1    26 2    26 4    26 5    26 8 

This was meant to be the “Aha!” moment, where the distribution can be recognized as having characteristics of a typical letter-count distribution, especially if...

...z = 26 is taken as z = 0.

       z a           z d
z a           z d
z a           z d
z a           z d    z e
z a           z d    z e                                   j i     t a
z a    z b    z d    z e    z h     j d    j e     j g     j i     t a

0 1    0 2    0 4    0 5    0 8    10 4   10 5    10 7    10 9    20 1
1      2      4      5      8      14     15      17      19      21
a      b      d      e      h       n      o       q       s       u 

This may indeed represent a 24-letter message as it has 10 unique letters, including 4 vowels.

Step 5. Rewrite the cipher...

. ...in terms of these pairs.

       bz ze az dj ji za ij az zd dz za dz ej za dz jg at ez at ze dz zh az zd

b  e  a  n  s  a  s  a  d  d  a  d  o  a  d  q  u  e  u  e  d  h  a  d

beans       as     add    a   do    ad        queued         had 

Terse as the puzzle statement is, it does say:

Don’t be satisfied with a nonsensical string of words, which will be just halfway to the answer.

This reinforces the progress so far and leads to completion by repeating steps 4 and 5.

Meta note: I think writing this in somewhat of a chronological order works better than trying to fill each section with that specific content. In my case these sections came kind of in that order, although there is some overlap. I added a section "finishing touches" in this case to handle how I made the puzzle presentable. I also decided to spoiler the image, because it is more easy to see that than the text in the post itself.

This is not a solution to the puzzle but provides notes from its poser. This type of answer has been approved by the community.

Caution: This post may contain spoilers.

### Inspiration

I lurk every now and again on Puzzle.SE, and found the questions interesting. However, a guess-the-digital-format question gets boring fast. I decided that I wanted to encode a sentence... in an unconvential way.

In my mind, there are two over-used onces:

• Various ciphers on natural text, usually yielding a similarly sized string.
• A base-x encoding, where we obfuscate the text by using the ascii numbers, and interpret it in a different format.

I thought of how we write things, and how text can have shapes. That's where I started my puzzle.

### Creative steps

The first step was to decide on how to encode these shapes. I thought about L-Systems and Turtle graphics in particular. You can encode any shape in them, as long as you define the steps well. However, this would be a puzzle, which means that I had to obfuscate the steps involved in this. If I would have used a string such as b[LF[F]RF]Fa, the puzzle would have been solved within an hour.

After some looking around for easy shapes, I remembered that we could create letters with 7 segmented displays. After some searching around, I settled for this sheet:

Next up I created a simple Python program that took an arbitrary string and converted it to turtle graphics. After I was satisfied with that, I had to decide on the actual sentence people had to decode.

The sentence had to be something iconic - something people knew was correct when they found it - and something that ideally made some types of analysis harder.

### Logistical steps

Now that I had a phrase, I decided to encode the characters I used to make turtle graphics to numbers. I had insider knowledge that the last two characters would always be the pen-up and forward event. In fact, this was a common delimiter between characters.

I deduced that from the numbers alone, it would be nearly impossible to find the turtle graphics. Especially because people did not know what they were looking for. However, due to the nature I generated my characters, there was a variable:1 substitution cipher that could be used. I briefly considered manually replacing some of the silly sequences, such as "penup forward penup forward" with a more efficient sequence, but this would break the substitution cipher that existed, and more importantly, would mess up any analysis people could do on the anonymous states.

In the end, the intended way of solving the problem was by simply analysing the string, finding the 2-character delimiter (30), then finding a suitable substitution after struggling a bit with characters that natively contained this sequence. Afterwards, knowing what the sequences encoded, I found it much more likely people would find the turtle graphics, as the steps were very basic for turtle graphics, and similar characters produced similar sequences.

### Finishing touches

In the end, I still needed to present my puzzle. I wanted my audience to be able to find the strictly necessary information easily, but I did not want to present it on it's own. I could add some meaningless flavour text, but I decided to present it in the context of an alien species instead that would actually not be able to decipher the code. For additional humour I decided to present the aliens as beings that did see themselves as superiour beings, being able to do everything better than "meatsacks". In the end it turns out they are not capable of reading, and kind of "as blind as a bat". Furthermore, they identify someone that apparently played a fairly horrible video game as "some kind of hero", unable to distinguish between the virtual reality of video games and the actual reality. I also added some remarks about kittens and reputation.

### Evolution

In the end, the puzzle seemed to be too difficult. People did some analysis, but did not get the breakthrough I hoped for. People seemed to return to trying to interpret this as base-7. The first hint was a subtile one where I gave some backstory, while also hinting that 0 thru 6 was not anything mathematics related.

Some useful information was posted after that hint, and I decided to wait it out a bit. People still were messing around with base-7, so I posted a more obvious hint about what 0 through 6 meant. It also contained some hints that would make the logical leap between sequences and the turtle graphics more managable.

Looking at the answers I got, the only thing people seemed to lack was data. I decided to encode another string, transcribing part of the second hint in it, and posting the sequence. This resulted in a rapid breakthrough.

### Takeaway

I think this puzzle was slightly on the difficult side. Things I have learned:

• Don't encode distinct states without an order, without any relation to math, with numbers. People will try to use base-x for it, and waste a lot of time. I would probably try random characters.
• People liked the role-playing as an alien overlord in the comments. It might be worth repeating such a thing if appropriate.
• I am unsure if hiding hints in "research notes" (basically flavour text) is a good idea. In the end it was my intention to subtily push people in the right direction, but it might have been to subtile.
• People on Puzzle.SE can make crazy logical leaps. I should probably start my puzzles with just a little more information.
• These wrap-ups just make me like their puzzles all the more. Speaking of turtle graphics and programming language variety, I hope you noticed Dragons: Myth or MathJax?. Turtle graphics were also considered (but turned out to be a false lead) in solving a different 7-segment puzzle.
– humn
Sep 17 '16 at 19:41
• Just noticed, Sumurai8, seems that you never posted this as an answer at the puzzle itself. This belongs there by now and it's never too late.
– humn
Aug 22 '17 at 12:36
• Thanks for the heads-up. I have now posted it as an answer. Aug 22 '17 at 14:18