The Nonlinear Library: LessWrong LW - (Not) Derailing the LessOnline Puzzle Hunt by Error
Jun 4, 2024
05:43
Link to original article
Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: (Not) Derailing the LessOnline Puzzle Hunt, published by Error on June 4, 2024 on LessWrong.
(spoiler alert: may meta-spoil future iterations of the LOPH, if you haven't already read other posts about it)
I knew early on that I wouldn't be able to finish the LessOnline puzzle hunt.
I contributed to solving two of the first six puzzles, each of which revealed the combination to a locked box. Each box contained wooden medallions for the solvers, plus a QR code. The QR codes led to the second layer of a larger meta-puzzle.
On discovering the existence of the meta-puzzle, I knew I would have to drop out. It was a shame, because I have never done a puzzle hunt before and I could see that the puzzlemasters had produced something amazing. But the opportunity cost of playing "for real" was just too high. There were too many other things I wanted to do, and the integral of my attention over time is not infinite.
So I stopped hunting. But, unexpectedly and hilariously, I ended up contributing to the game in a very different way.
Suspicion
During the Fooming Shoggoth concert I noticed (probably thanks to the song lyrics) that the parts of the puzzle I was aware of were all lockboxes. I mean, I knew that already, but now I noticed. And I knew from the contents of the two boxes I had opened that there was more to the puzzles than the obvious. The whole setup seemed like a suspiciously plausible metaphor for the AI Box Experiment.
I suddenly, strongly suspected that the puzzle hunt had a hidden narrative -- one that would end with the release of a rogue AI.
Sequence Breaker!
So I did what any Less Wronger should have done: I tried to warn others. No unbinding of seals, no opening of gates! At first I just told nearby hunters directly, but quickly realized it wouldn't work; if nothing else, I didn't know who was working on the puzzle hunt and who wasn't. I abandoned that plan and asked the front desk to print out three notes for me. The notes outlined my suspicions and warned the reader not to open whatever final box might be found at the end of the puzzle chain.
I taped Note 1 to my medallion, which I returned to its original box. In addition to the warning, Note 1 asked anyone opening the box to leave the medallion and the note itself for later hunters to see, and suggested they return their own medallion(s), just in case. I had no reason to believe it mattered whether the medallions stayed in the boxes, but I had no reason to believe it didn't, and it was an obvious thing to try.
I taped Note 2 to the table by the lockboxes. Note 2 asked anyone who shared my suspicions to sign it, as social pressure on others to not open boxes that might contain Very Bad Things.
Note 3 was a copy of note 2, and stayed in my backpack for contingencies.
After placing the notes, I moved on to other things. I'd volunteered to run a talk the following day on a whim, and preparation funged against sleep. By the time I had the slide deck put together it was 4am. Before going to bed, I checked on my warning notes.
Note 2 was gone.
I'd thought of that possibility, which was why I had a contingency copy. Now I had to decide whether to use it. Crap.
Decision Theory
So far I'd been intentionally open about what I was doing. I told multiple hunters about both my suspicions and the notes. I showed Note 1 to a volunteer when placing it. Note 2 was in plain sight. I even had a staff member at the front desk print the notes for me. My in-character warnings to hunters doubled as an out-of-character warning to the puzzlemasters: "possible derailment in progress". I didn't want to actually ruin whatever awesomeness they had planned.
I wanted to prevent hunters from opening the hypothetical final box if and only if that was the true win condition.
The problem was that I didn't know -- and couldn't know -- whether th...
Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: (Not) Derailing the LessOnline Puzzle Hunt, published by Error on June 4, 2024 on LessWrong.
(spoiler alert: may meta-spoil future iterations of the LOPH, if you haven't already read other posts about it)
I knew early on that I wouldn't be able to finish the LessOnline puzzle hunt.
I contributed to solving two of the first six puzzles, each of which revealed the combination to a locked box. Each box contained wooden medallions for the solvers, plus a QR code. The QR codes led to the second layer of a larger meta-puzzle.
On discovering the existence of the meta-puzzle, I knew I would have to drop out. It was a shame, because I have never done a puzzle hunt before and I could see that the puzzlemasters had produced something amazing. But the opportunity cost of playing "for real" was just too high. There were too many other things I wanted to do, and the integral of my attention over time is not infinite.
So I stopped hunting. But, unexpectedly and hilariously, I ended up contributing to the game in a very different way.
Suspicion
During the Fooming Shoggoth concert I noticed (probably thanks to the song lyrics) that the parts of the puzzle I was aware of were all lockboxes. I mean, I knew that already, but now I noticed. And I knew from the contents of the two boxes I had opened that there was more to the puzzles than the obvious. The whole setup seemed like a suspiciously plausible metaphor for the AI Box Experiment.
I suddenly, strongly suspected that the puzzle hunt had a hidden narrative -- one that would end with the release of a rogue AI.
Sequence Breaker!
So I did what any Less Wronger should have done: I tried to warn others. No unbinding of seals, no opening of gates! At first I just told nearby hunters directly, but quickly realized it wouldn't work; if nothing else, I didn't know who was working on the puzzle hunt and who wasn't. I abandoned that plan and asked the front desk to print out three notes for me. The notes outlined my suspicions and warned the reader not to open whatever final box might be found at the end of the puzzle chain.
I taped Note 1 to my medallion, which I returned to its original box. In addition to the warning, Note 1 asked anyone opening the box to leave the medallion and the note itself for later hunters to see, and suggested they return their own medallion(s), just in case. I had no reason to believe it mattered whether the medallions stayed in the boxes, but I had no reason to believe it didn't, and it was an obvious thing to try.
I taped Note 2 to the table by the lockboxes. Note 2 asked anyone who shared my suspicions to sign it, as social pressure on others to not open boxes that might contain Very Bad Things.
Note 3 was a copy of note 2, and stayed in my backpack for contingencies.
After placing the notes, I moved on to other things. I'd volunteered to run a talk the following day on a whim, and preparation funged against sleep. By the time I had the slide deck put together it was 4am. Before going to bed, I checked on my warning notes.
Note 2 was gone.
I'd thought of that possibility, which was why I had a contingency copy. Now I had to decide whether to use it. Crap.
Decision Theory
So far I'd been intentionally open about what I was doing. I told multiple hunters about both my suspicions and the notes. I showed Note 1 to a volunteer when placing it. Note 2 was in plain sight. I even had a staff member at the front desk print the notes for me. My in-character warnings to hunters doubled as an out-of-character warning to the puzzlemasters: "possible derailment in progress". I didn't want to actually ruin whatever awesomeness they had planned.
I wanted to prevent hunters from opening the hypothetical final box if and only if that was the true win condition.
The problem was that I didn't know -- and couldn't know -- whether th...