The puzzlehunt has ended, thanks for playing!
Puzzlehunt 2012

Reset Puzzle State

Head Count

vamp pivot havoc havoc prog pat adopt puma gap putt hump fop cap camp amp wapato pooh top gpo thaw cow chaff quod hub tech jams beds jots header chef high fmc jacm sniff fund judd nags find jed lew jon ftc adjured jock be farm jed vii deject jed jeep kate comb jet put adject dock ferber vied waved eve rivet vii itch vito rid tva rid vicar cramp


Do some stats!



Joe Seaton (js845)


The solution to this rather opaque looking puzzle is actually quite head-smackingly simple. The trick is in not being fooled into thinking the words (and yes, they are all words) are important.

I imagine most people entering this competition have at least heard of frequency analysis. Applying this technique to the gibberish above results in the following:

Frequency in puzzle2651920251251315143111519201115209135000
Frequency in English (scaled)2301123566171902117192150171825807050

Clearly something is strange here. Not only do the letter frequencies not line up with what we expect for English, the most frequent letter, A, appears significantly less than E, making simple substitution ciphers unlikely. Actually the solution is hinted at by this frequency - 26, the number of letters in the English alphabet. By treating the frequencies themselves in the table above as letters, taking 1=A, 2=B and so forth, we get the string "zestylemoncakestaketime" or to expand out this strange phrase, "zesty lemon cakes take time".

I'm quite glad only one team worked this one out, since it took quite a long time to actually produce a satisfying text for this puzzle. The awkward part here was generating a sample text that was all valid words.

The strange answer is the result of my attempts to construct a sentence that would translate into something not too far off the appropriate letter frequencies. Even slight mismatches quickly resulted in having far too many of particular letters. Even the sentence used resulted in far too many 'j's and 'p's. Perhaps generating the puzzle text would have been easier if the answer generation had been automated too.

Initial attempts at generating question sentences by depth-first search proved far too slow, even for one-off use. The eventual solution used a bizzare greedy algorithm which back-tracked and removed words probabilisticly. Results could then be picked through for the most satisfying text, namely the one with the least obscure words. While ugly, it served it's purpose - several teams confessed that they treated it as a word game, even after the hint was released.


zesty lemon cakes take time

Correct submissions

The 5th Element 2012-06-16 15:46:57.155664