Log in

Back from hunt - The Revolution of the Moon
Links WikiWikiWeb / Harvard School of Public Health / KPIG -- my sation / Smith -- my former(!!!) college
Fri, Jan. 14th, 2011 07:37 pm
Back from hunt

A longer post will follow with more of my hunt experiences, but first, before I forget it all, I want to share a minor *cough* frustration I encountered.

So the actual answer to the puzzle is here: http://ihavetofindpeach.com/puzzles/mega_man/recombination/answer/

I got handed this puzzle in, it turns out, a completely solved state. That is, they had done exactly what they were supposed to do except that they only did it once. They even got "ALLINCLAPC", which was described as "ALLIN followed by garbage, like the other clues". Note that you are doing exactly the same thing you did before, when each clue came with a bunch of garbage letters at the end that you ignore, and it very specifically has an end point that does not connect with something to form a complete cycle. I have no idea how the C migrates to the front of that clue. All in all, I have no idea why we were supposed to prefer "ALLINCLAPC" to "ALLIN" which followed the pattern established up to that point, but I wasn't involved in any of this so there might well be a good reason.

I just got handed an already-solved puzzle and told that this answer wasn't the answer, the obvious extraction mechanism wasn't the extraction mechanism and here are 5 numbers, 5 words and a bunch of unused information.

First, there was the title. It could refer to taking letters from multiple words and combining them into one word, but that is what we do in very nearly every puzzle. It's sort of like getting a college titled "Black And White" and having it refer to the color of letters on the paper. However, "recombination" and "juggling" are two of three terms for non-destructive genetic mutation, along with "reshuffling". So to me, this puzzle was clearly cluing DNA. Up, down and sideways. Not only that, but my five words all have A,T,C,G.

This is when I went off into apparent lala land.

First, I got one set of [A,C,T,G] by indexing into the words with the number of the puzzle where they are juggled, with the fourth word producing "R", which I couldn't decide if it was a clue that this was RNA (you'll see), or just ignored. That makes sense. Then I have all these extra, repetitive, unused letters at the end of each letter. If I counted them I, and an independent counters, got consistent-but-different-from-reported-in-the-answer numbers, in the same range of 3 to 7 as the original numbers (though non-unique this time), which give another set of [A,C,T,G], with one clue producing no letter at all. I liked the dual strand idea and that I had some concrete connections.

So now I have two sets of four words in pairs. Three words were tied to two other words, two words were tied to only one other word. It did not give an ordering, except that CATNIP had ended the cycle earlier, so I stuck it on the bottom while not being completely satisfied with that if it came up later.

Now this digression down lala land would probably have been abandoned and we'd eventually have backsolved it, except that this is what you get when you put these together in the given configuration:


One diagonal. A 5-nucleotide sequence: ACCGA, which uses letters not used while putting this together (not necessary, but still aesthetically pleasing.) This appeared absolutely clear, unique and well-clued in my brain at that point in the hunt (it even kind of looks like half a double-helix.)
It's not clear if it is RNA or DNA, but since I had transcribed the sequences according to the DNA, which is how RNA is usually produced, I figured RNA was the stronger position (plus, the extra R might be cluing, who knows) . It you extract every possible amino acid in that sequence (ACC, CCG, CGA), you get TPR, which is a human gene that causes thyroid cancer when it experiences recombination errors. However, there is no indication of where to go from there (of course, since I was never supposed to get here in the first place.) I had no indication of Latin vs. English name, for example. I called in thyroid cancer, just in case it was simple and poorly constrained.

I really kind of wanted to anagram, since then I'd have juggling, recombination and reshuffling. THYROID CANCER wasn't going to help, and neither were the three-letter versions of those amino acids. So I produced four sets of three letters by extracting forwards, then backwards, then the transcribed sequence (that is, as though this were DNA instead of the RNA I had assumed it was) backwards and forwards. I liked this even more because it meant it didn't matter that it didn't clue up-from-down or RNA-from-DNA strongly; those orientations just didn't matter! I took the three-letter abbreviations of their amino acids and then discarded (I think it was, don't quite remember) all the adjacent-and-stacked letters. Which makes sense, because in recombination errors only the changes matter; if you swap an A for an A it is irrelevant. From these poor, over-worked five letters I had wrestled free a set of seven letters that anagram into ORAL STREP.

Finally! Success! The only unused information one discarded R instead of almost half the information in the puzzle! Recombination, juggling and reshuffling! It even fit the pattern we'd seen in the meta (everything up to that point (that is to say, two puzzles) had been a two-word disease or plague of some kind; ORAL STREP sounded unpleasant to me and was two words, if a little short.)

When I called this answer in, though, in the guy on the phone said, "you really should try solving this forward; it is a good puzzle." Clearly he saw that we'd almost had it, time passed and then I'd called in THYROID CANCER and figured that I was now randomly choosing two-word disease phrases as awkward as "ORAL STREP"? Whatever his reasoning, however well-intentioned his "stop guessing" chiding was, it felt like a punch in the gut.

I had just spent about 45 minutes following a train of logic that if it weren't so clearly correct as I went along, and elegant and rewarding, would have been frustrating all on its own. Then I had spent at least another hour and a half, probably more, attempting to wring an answer from it before hitting on the backwards&forwards idea. I thought that, after two hours post-work-doing, when the obvious answer was reported to be wrong I was no longer obliged to think it was a good puzzle. In fact, he appeared to think that instead of devoting hours to his his "good" puzzle, I might as well be backsolving.

So I went out and cried my frustration off in the hall and then unproductively obsessed for a while, and went to sleep and woke up and obsessed for a while longer until I finally gave up in a depressed, frustrated huff. When the answers went up I immediately went to look, hoping to discover that I was just too stupid to figure out the obvious and elegant thing I should have done instead. Instead, it was the most unsatisfying conclusion to a puzzle story thus far: it was already solved, but the clue phrase looked like garbage, so I invented an entire puzzle on top of it. Well, I think my version of the puzzle was good-er than that clue phrase, so there condescending Mr. You-Might-As-Well-Be-Guessing-Randomly-As-Trying-To-Solve-This-Puzzle.

So, in conclusion, the theme of this hunt for me was pretty much "you might as well be backsolving." And don't get cancer *or* oral strep, because neither of them is the answer, apparently.


Mon, Jan. 17th, 2011 02:53 pm (UTC)

I have the power to punish at least one of the writers. If it was he that was mean, I will do that for you, just say the word. :-)

Mon, Jan. 17th, 2011 07:19 pm (UTC)

Well, I realize he didn't intend to be mean; he just had incomplete information. They actually had a box that said "I am backsolving this", which wasn't visible to the people on the phone (presumably so they wouldn't say things like this? I'm not sure), but the issue could probably have been avoided if it was.

ReplyThread Parent
Mon, Jan. 17th, 2011 07:55 pm (UTC)

As much fun as it is for the running team to mock the backsolvers (and I am sure you have heard the ACME story about it), if the person on the phone can’t see the “yes, I backsolved it” button, it seems only polite to ask the solver if they did, in fact, backsolve before they pick on you for backsolving.

So, let me know. I have access to Plot. ;-)

ReplyThread Parent
Tue, Jan. 18th, 2011 01:59 am (UTC)

FWIW: It wasn't visible to us from the phone queue. It was visible in the previous answers log (as gray italics), which is an entirely different page.

Also, god damn is that a huge tangent to get launched into because of one letter. *forwards this post to team*

ReplyThread Parent
Tue, Jan. 18th, 2011 08:53 am (UTC)

Oh, good, that saves me the trouble of forwarding this post to your team. :-) Actually, I did write to Noah and Aaron, not so much about the details of this solving story but to suggest that they pass along to Codex some ideas about HQ doing sanity checks for people who have found themselves deep in a red-herring wormhole on a puzzle. I would have included the URL for this post if I'd realized it was written up, so it's perhaps just as well that they'll see it via your mail.

ReplyThread Parent
Tue, Jan. 18th, 2011 09:34 pm (UTC)

I'm one of the authors of this puzzle. I can't imagine I can say anything that will prevent you from hating me forever, but I'm extremely sorry you got so frustrated working on it. The red herrings you found are truly astonishing, though, and I do kind of wish we'd written the puzzle you describe (though I'm sure our editors would have rejected it because it sounds too constrained to be writable).

All in all, I have no idea why we were supposed to prefer "ALLINCLAPC" to "ALLIN" which followed the pattern established up to that point

Basically, there wasn't a good reason. What we were thinking was that generally siteswaps are cyclical; for example when you juggle 441 you generally do 441441441441... So if you juggling ocarina, grail, illogic, conceit, and catnip in 75364 and repeat the pattern, you'll get ALLINCLAPCALLINCLAPCALLINCLAPC, so the exact start position doesn't matter. The problem is that the siteswaps in the rest of the puzzle aren't cyclical, so as a solver you're not necessarily thinking about it that way. Originally we were going to have the sequence of numbers be a cycle as well. In fact, the set of objects that are being juggled are already designed for this, and all you have to do is replace the cauliflower with "catnip", and everything else works and the sequence cycles. The problem with that was that you don't know that you should start at the 7 in the siteswap and not somewhere else. If you interpret the pattern as 64753, you get the string INALLPCCLA.

Much too late, we realized that the answer to this puzzle should have been UPKEEP (which ended up going to Letter Bank). Not only would that have been more thematic than CLAP, but we could have had a length-6 siteswap that gave the string ANSWERUPKEEP. Then we could have had a full cycle of objects, and it would be easy to deduce that you should start the pattern in the position to get ANSWER, and then if you continue it, which is clued by the cycle, you get UPKEEP, which you would hopefully not mistake for garbage letters. But, this would have meant rewriting almost the whole puzzle from scratch, including filming 6 new videos, and we just didn't have time for that.

Wed, Jan. 19th, 2011 02:40 am (UTC)
Commenting before reading.

I can see that Jen responded, so I'm guessing that other constructors/running team members might also. To be clear, I'm responding now before reading any of the comments, and I apologize ahead of time if I'm unknowingly offending anyone who commented farther on.

In general, I found this Hunt to be technically well-done but somewhat unsatisfying, because so many of the puzzles exhibit qualities like the trap into which you fell. In particular, I'd say that at least 50% of the puzzles I saw followed the pattern: ``have an insight, gather some data, consider it, and then iterate through the 20-odd different unclued, unhinted random extraction mechanisms until you find one that seems to work''. This is a strategy that works well for large teams that devote 5- and 10-person teams to each puzzle, but it makes things totally frustrating for smaller teams, because there is no insight, no `ah-ha!', and no `breakthrough' - there's just ``grind it down until it works; repeat''.

In general, I feared (without evidence) that too much attention was given to polishing the meta structure, and not enough to polishing the actual puzzles. If forced to guess, I'd say that time was the key factor here -- if someone were to take the 2011 hunt, and spend a week or two adding clues to 90% of the puzzles, it would be a truly beautiful Hunt; possibly the best I've seen in 20 years of off-and-on Hunting. As it was, it was a fine Hunt, but on that was really very heavily skewed to large and/or heavily experienced teams (like mine, to be clear).

Hopefully you got to have some fun on other parts of the Hunt. If not, you could always dig through the puzzles you haven't seen, and make J check answers for you on the way. :)