I enjoyed the experience, net, but it was about 10 hours too long and about 30% too hard. Fixing either one would have been a big improvement, too: less time to stare out unsolvable puzzles, or more tractable puzzles to mow through for the duration. *shrug* The fun thing about playing with the team I did (Left Out) is that it's playing _with_ people whom I usually play _against_ in puzzle hunts. We had 12 - 18 people in the group in San Francisco and about the same in Boston at MIT. The San Francisco group was a lot of fun, especially for being able to learn how other people approach solving.
The puzzles were structured in two layers. First, there was the "little black book" which contained contact info for many people and a puzzle attached to each. The people grouped together and their contact info + answers combined to form a phrase. Submitting these phrases to GC released Dossiers about certain of these people. Within a dossier were several puzzles as well. The answers to these puzzles combined with information in the dossier and allowed us to accuse that person of committing the crime (killing Dr. Awkward; see the official MIT Mystery Hunt page for the backstory). So, two layers of metas.
By about midnight Saturday night we'd solved enough LBB puzzles to reveal all the dossiers and nearly all of the dossier puzzles. (Plus, it seems, the dossier puzzles were revealed to us based on time, even if we didn't solve all the LBBs). One thing that was weird was that it seemed at this point there was no reason to solve any more LBB puzzles. So we tried working on the others instead, and the metas. We managed to solve two or three metas, and then to backsolve several feeder puzzles as a result, but that left about 10 metas to go. By the time the event ended we were probably 24 - 36 hours behind.
The puzzles were individually less fun than I've seen in the best puzzle hunts. Some puzzles required specialized knowledge but didn't make that very clear; for example, we had four dozen celebrity photos to identify, but knowing that they were all guest stars on the show 24 would have helped greatly. We figured this out but wasted a lot of time on the brute force approach. A much more frustrating effect was that some puzzles first required a difficult, but approachable, effort, but then finished not with the answer directly but with a phrase that was still obscure. For example, one (otherwise) fun puzzle required reassembling comic strips from individual panels, noting that certain letters were missing. These elided letters formed a message: CHARACTER WEARING SUSPENDERS. This would be a fine message if one of the characters in the puzzle wore suspenders, but no, no such luck. We did the fun part in about 30 minutes but didn't get the right answer for another 10 hours, because we just didn't guess right. There's lots of characters who wear suspenders, comic and live-action, but none seemed more right than the others. The actual answer also didn't clarify why it was chosen among the others. We encountered many such puzzles: we did the hard work, got a message, and instead of being rewarded with a solution we were rewarded with an impossible task and failure. This was gradually demoralizing.
Overall the puzzles also just seemed to offer less purchase than other puzzle hunt puzzles. There seemed to be a lot of specialized knowledge needed, or just being able to see into the puzzle creator's mind. I expected some number of these puzzles, but also puzzles where the task was obvious but perhaps just lengthy ("world's biggest crossword puzzle"). We didn't see many of those. We also didn't see many puzzles where a big group of people could solve them more rapidly than one or two people could solve them.
Just to be clear, I'm not complaining here, I'm trying to record my state of mind before it's gone with sleep. 8) Creating puzzle hunts is hard, and I'm trying to put my finger on what lessons could be learned to make them more fun for players. The lessons I would take from here: after a long and difficult process in a puzzle the answer should be right there, it shouldn't require "just one more easy step." Because the easy step may not be easy. Also, a puzzle that requires domain-specific knowledge should make this clear. It's not fun for a player to spend hours trying to solve a puzzle that they were simply never going to be able to solve for lack of knowing about X.