dr4b | Liboncatipu, aka Microsoft Puzzle Hunt 7, aka How I Learned to Stop Sleeping and Avoid the Chess

Right, so I spent all weekend at Microsoft :)

I want to write so much about Puzzle Hunt, but I'm not sure I can really capture it properly. I had overslept, so I was all nervous about getting out there and being on this team of almost entirely total strangers (I knew Matt and Drew, but the other 7 guys were all people I'd never met before) and having them already hate me because I was late. Turns out I needn't have worried... while it was a little odd at first to just be in this conference room with a whole ton of strangers, Drew told me how the puzzles worked and that I should go find something to work on, so I did.

I saw a guy or two doing cryptic crosswords, and I'm good at those, so I sat down and started coming up with clue answers for them! That cryptic turned out to be really cool - first you solved the three cryptics (which had "RED" and "WHITE" word spaces on them, like the words "Egg White" and "E.B.White" would cross with only 4 and 3 squares), then you used the colors from one cryptic and the P's, R's, Q's, K's from another to get a chessboard - the letters from all of the cryptics to spell out "Find seven places where Alice can be and White mates in one", and then using those spaces on the THIRD cryptic you got the answer "Jalappa".

(There was an Alice in Puzzlehunt theme for the whole hunt... Alice was kidnapped, we had to solve puzzles to get her back.)

The first puzzle they got set the tone for the whole day, I think :) It was this one where we solved a sequence of numbers to read a bunch of food names, and also: "Call xxx-xxxx (I don't remember the phone number) Ask for Puzzle Special". So we call the number and it's the Overlake Pizza Hut. We ask for the puzzle special, they get our info and say "That'll be 12 dollars, we'll see you in 45 minutes." We're like "Was that for real?" A bit later, a pizza delivery guy shows up with a pizza for us! Mike pays 12 bucks and we open the pizza, and sure enough, there's a bizarre set of toppings on it... carrots, scallions, whatnot. Nothing weird in the box, etc... so we eat the pizza :) Turns the toppings were clues to the second half of the puzzle, plus one of the clues was "teriyaki sauce" although we ended up with "there's no sauce" instead and still solved it. Still. How cool :)

The other major "the world is in on this" clue of the night was... around 10pm we get a clue that says to go to Safeway as soon as possible and pick up the Sunday New York Times, the next puzzle is on page 89 of the Magazine. Guess what... it was the Sunday crossword, written by one of the guys who had made up Puzzle Hunt. He'd gotten Will Shortz to SPECIFICALLY put it in the NYT this weekend! And got Safeway to get ahold of it to sell at 10pm Sat night! It was SO COOL! So I got to do the crossword puzzle and then find some letters on it and do some more puzzles to it and so on. SO COOL.

In the meantime we had two of the funniest colorful puzzles that I wish I'd had my camera for -- one of them, they showed a hand with colorings on it and gave a keyphrase - it turned out you had to color your hand like that and use american sign language to spell out the letters and numbers in the keyphrase and given the hand positions for each character in the keyphrase, the markings on your hand would look like a letterform... take those letterforms together and you got a word. Whee! The other was this ANNOYING AS HECK Soma cube... we cut out from paper and built these 3-D pieces to make a magic cube -- the problem being you HAD to line it up so that all the letters faced the same way on each side, which was nigh-impossible. (Yes, we figured all that out before the clue came in.) Many of us played with it for a while and eventually Yang started marking the sides with colors as "which way can be up and which way canNOT be down" and after another hour he had the cube assembled and the directions to solve the rest of the puzzle!

There was also a few other 3-D puzzles; one was a dodecahedron where you had to figure out the pictures as characters from Alice in Wonderland and match them up in pairs and decrypt a message from that.

We also had a recursive wordfind... you had to find 18 words of a huge grid, then take the remaining 121 letters and refill them into an 11x11 grid and find some more words, then take the remaining 49 letters and refill THEM into a 7x7 grid, and the remaining letters after you found the new wordset made up the answer to the puzzle.

We had a great puzzle that involved solving 5 grids as 5 different games (Chess, Battleship, Connections, etc) and then overlapping things to get a grid with some letters in it. We could NOT figure it out until Brian walked up to the board like "Hey guys... look, upside-down it says MY ANSWER IS THE WORD ASTROICS". Of course, Astroics isn't a word. We rechecked our work and got Astronics. Whee! And we were second to solve it, woulda been first if we'd gotten Astronics faster :)

They also gave us a deck of cards which was an encrypted phrase as well. It took us WAY too long to get that one. (I had the letters, but not the order, sigh.)

I went home for a break, and when I came back some people had seriously been there all night working on puzzles, and they'd gotten a few more, like the pigpen cypher and whatnot. So I needed something to work on, and herein was my finest achievement of the weekend:

I solved a cryptogram in the International Phonetic Alphabet.

We'd gotten this thing of incomprehensible text which turned out to be IPA. But when I looked up IPA to try to just translate it to English, the text was utterly unpronounceable. The hint said it was a cryptogram. So I sat there looking for common words, repeated patterns, etc, and surfing the web to see the ways of representing those words to see if they matched up. I had two common two-phoneme words; one came 17 times and one 8, so I decided the 17 one was "the". Fortunately I turned out to be right. I also figured out "nineteen ninety" out of a bunch of the word pairs... And I thought I was wrong, but when Drew came over to look at it and I explained why I thought I was wrong I figured out that no, I was right! After that it was pretty easy to break the cryptogram, and start "reading" the IPA... fortunately after reading the first two sentences, the guys were able to Google the proper answer to the overall puzzle. Word.

The other fun puzzle I had a lot to do with was, they had given us an 8-minute mp3 that was basically like 25 songs spliced together... all of them had Alice In Wonderland themes to them (like stuff by Alice Cooper, Alice in Chains... songs like White Rabbit, Queen of Hearts, Alice Springs, Drink With Me, etc) but that wasn't the point. The point was that they all cut off at odd times, and you had to figure out the NEXT word in each song. Taking the first letters of each of those words gave you "Cowboy Junkies Sun Comes Up Its" and the last two "next words" were "blank blank". So the answer was "Tuesday Morning". Hee.

There were many other neat ones but I've mostly talked about the ones I had a hand in solving.

So anyway, the details on how it worked, though... No puzzle was solved on your first iteration. You always had to solve a puzzle to get directions to solve something else to get the actual answer to the puzzle. Some puzzles only had one or two levels of recursion, some had 3, 4, or more. Some puzzles required you to go all over Microsoft campus to various rooms, you'd usually solve a puzzle telling you where to go, then at that place you'd get more clues to bring back and solve the rest of the puzzle. There was even a Mad Hatter's Tea Party, and a caucus-race, and a curious croquet game that was all logic gates, and whatnot. Very cool stuff.

Since I am not a Microsoft employee I didn't go on any of the scavenger hunt puzzles though - I didn't want to get lost AND I didn't want to get locked out of buildings. So I pretty much spent all of my time in our conference room, or in the hallways between the conference room, the bathroom, the copy/printer room, and the kitchen where I kept taking more and more Diet Coke :) But it was ok. We had the conference room set up with tons of paper, pencils, markers, scissors, tape, about 7 computers, snacks... it very much felt like sitting in the CS Lounge or somewhere at CMU just working on a huge project with people.

And the guys on the team were great! Like I said, except Drew and Matt they were all strangers to me before this event, but I had a blast hanging out with them and making jokes and solving puzzles and whatnot. At like 3am we managed to end up in *5TH PLACE* out of the 51 teams and we all started cheering and screaming and stuff. Turns out there was another Puzzle Hunt team in the conference room next to us and so in retaliation they started cheering and screaming too :)

At the end we all went to the closing ceremonies, where we got to cheer for puzzles and have them explained by the creators, and see who won. Our final standing was 8th place and we were still pretty damn happy about it. We were one of the teams to make it to the final ending meta-puzzles, and so they read our team name out, but of course couldn't pronounce it. "Lib-un-cah-TIP-you?" "Close enough!" (It was apparently "Lee-bone-CAT-ee-poo", after a word that Matt had accidentally gotten as a wrong answer to a puzzle last year.) Hehe. It was still fun. Turns out that my friends Kevin (Babbitt) and Michelle were on the 6th place team... and I had just said to Drew, "I wonder if I know anyone here? I bet anything if I turn around I'll see someone I hadn't seen in a while" so he says, "Deanna, turn around" and sure enough, I see K&M smiling and waving at me. We had a good laugh over that one.

Yar. I am *so* in for next time's Puzzle Hunt :)

Flat | Top-Level Comments Only

I played with those damn cards for 24 hours before getting enough hints to solve it. We kept asking questions about it and finally got a reply that said "OK, we'll give you some more hints soon" and five minutes later we sovled it and sent a reply saying "that's ok, we'll just give you the solution :P" We stared at the cross-examine answer for ages before realising it was longitude and lattitude :)
We only solved 4 puzzles in the first 24hours, but solved fifteen in the last eight :)
Good times.

Damn! All of this just sounds so amazing. I wish there was something like that around here, but then again, I seriously doubt I'd be able to get past puzzle 1.

So, I've got a year to talk myself onto someone's team for the next one? :)

This event sounds like a really intense way to enjoy oneself. Thank you for letting us share it with you. I wish I could have taken part.
Congratulations on placing so well.

Was the NYT puzzle entitled Movie Sequels that were Never Made?
The Pittsburgh Post Gazette publishes crosswords from the NYT and the Chicago Tribune each week, and every week for months (since I started noticing) the two puzzles have had one solution word in common, but not this week. Just enough of a coincidence to start a conspiracy myth. ;)

Yes, Movie Sequels That Were Never Made (by... Bruce Leban? Something Leban, he was the captain of the puzzle team that ran Hunt I think)

That was the BEST part... how not only was it specific to our game (And did you notice the "Sleeping in Seattle" and the "Alice" inside jokes he threw in there?) but it was a GREAT crossword, I really enjoyed doing it! I can't gush enough about what a cool event it was.

Our paper didn't give the author attribution this week. Hmmm.

It sure does seem that crosswords published the same day have an unlikely number of words -- one, or two, but still -- in common. I should time-scramble all crosswords and see.

I would think that the odds would approach certainty in the way that it is almost certain that in any group of 50 people, two will share a birthday. The curious thing is that it has been reliably exactly one word, and not more.

Hm, yeah, if you're at the breakpoint where the expected number of collisions is 1, then the Poisson distribution says that P(two collisions) is fully half the size of P(one collision).

Let's see, we assume no collisions within a single puzzle, but that doesn't affect very much the N^2 pairs between two puzzles. We could make the totally bogus assumption that words are drawn uniformly from some pool. The NYT's max word count for a 23-by-23 is 170, which gives one expected collision for a pool size of 170^2 ~= 30000. I've heard 50000 and 100000 tossed around for a normal vocabulary, but non-uniformity effectively reduces the pool size. On the other hand, crosswords include proper names etc. that wouldn't be in that vocab count. The numbers are so squashy.

Your analysis is a good one, and yes, setting up the assumptions on which the calculation would be based is a really squiggy situation. I suppose it could be done statistically, based on archival data from previous puzzles.
The word pool is both broader and narrower than a random selection. On the narrow side, there is a small but frequent population of crosswordy words, such as etui, that show up in the puzzles considerably more often than in everyday conversation or writing. This group is small enough that I have made a list of them in the back cover of one of my puzzle collections.
On the broader side, there are the words that don't normally occur in a dictionary, contractions, compound words, phrases, celebrities, etc., and word forms that one doesn't often see (such as an adjective like crosswordy--see above--made from a noun because a y is required) but which can be formed in English according to rules, plus word parts like prefixes, standard word parts (eg ance, which makes a noun into an adjective), and abbreviations. SST is a favorite abbreviation because plurals (and past tenses) are more common. At the bottom and right of the grid, there usually words that have lots of Es, Ys, Ss and Ts to complete the crossing words.
In tough crosswords (and the NYT has this reputation) there are also rare, archaic, alternately spelled, and goofy words. This both broadens the word pool in quantity, and narrows it because it is a small group of special words.
And finally, there is the theme, which can broaden the base because regular words, phrases, titles, etc. have been mutated, and increase the likelihood of collision because the theme (such as Thanksgiving) is often seasonal.
With these and other observations that would relate to the style and background of the the puzzle creator pool that the NYT and the Trib employ, and to crossword puzzle protocol and etiquette, it might be possible to characterize the word pool, but I think I see a dissertation-sized problem.

Heh. I keep that list of crosswordese on the backs of various articles and then leave them scattered about.

The only approach that comes to mind is to gather a sample of crosswords, and see whether a power law seems to fit the top end of the frequency distribution. Then, um, extrapolate vigorously. My suspicious is that phrases and other oddities would make it fatter-tailed than power law, though. And even just to resolve the top end would take a hefty corpus.

Sounds like a quest with an answer, however, it also sounds like a bigger project than is worth the undertaking unless some basic principle were to be discovered.
Thank you for your observations, however, because your speculation as to method and likely outcomes is interesting in its own right.

Thanks for nudging me to do the collision analysis/handwaving. I've memoried this thread under my "overingenious" tag, but that's a good thing.

Liboncatipu, aka Microsoft Puzzle Hunt 7, aka How I Learned to Stop Sleeping and Avoid the Chess

no subject

no subject

no subject

no subject

NYT Crossword

Re: NYT Crossword

Re: NYT Crossword

Re: NYT Crossword

Re: NYT Crossword

Re: NYT Crossword

Re: NYT Crossword

And various northeasterly compass points, good grief.

Re: And various northeasterly compass points, good grief.

Re: And various northeasterly compass points, good grief.