Galactic Puzzle Hunt 2019: Reflection

Organizing a team

Teams for the puzzle hunt can be up to 10 people [1], so the first step was to attempt to find other people who are at least as insane as I am. I put out a tweet linking to a Google Form (and also messaged a few friends), asking (amongst a few other things) for a time commitment, with the intent that I’d select team members most happy to spend time on it.

As it happened, I only got seven signups (including myself), so that was my team formed. I thought that was a reasonable number, but it did have a lot of people only able to make a small time commitment, and I should have kept searching for more people. My perception was we had ~2.5 daytime solvers on average.

There were a few people who gave us some help during the hunt, so I think that means I need to promote signups more next year.

Having brought together some people, we needed a team name. A bunch of ideas we had (some of which followed the Antarctica theme for the puzzle hunt):

• Production Freeze
• By accepting this team name you agree to our EULA
• Fizz puzz
• Ice climbers
• TE-901
• UNTITL~1.GPH

But “How about ☃ repeated as many times as the registration form will allow?” was most liked. Confession time: I didn’t actually do this. Instead, I went for three rows of snowmen (as displayed on my laptop). So while we didn’t have the longest team name, we did have the most vertical space in the team list.

Perhaps next year we’ll figure out how to consume even more vertical space with our team name?

Team coordination

We used Slack and Google Docs to coordinate. In the Slack workspace, there was a #coordination channel which had login details, a periodically updated “best puzzles to work on” post, celebrations of solved puzzles, and occasional calls for help. Puzzle solving didn’t happen in that channel: instead, we created a channel per puzzle.

It’s hard to predict what technology will be best to solve a puzzle, but generally we always want a document to describing what we’ve tried and with partial progress (which eventually morphs into a “how to solve the puzzle” doc). All of these documents got put into a shared Google Drive folder, and each puzzle channel in Slack had a topic linking to the relevant doc for easy access.

A few of the puzzles needed more than a doc. We also used:

• Photos shared in Slack (for a physical jigsaw-like puzzle)
• Images shared in Slack
• Google My Maps in Drive (super-useful for visualising latitude/longitude pairs and looking for patterns)
• Google Spreadsheets in Drive (although a few times I started to create one, and then replaced it with a table in a Google Doc)
• Google Colaboratory in Drive (omg sharable Python code with notes!)

I was generally happy with the technology choices, although I need to figure out how to use Colaboratory in a way that involves less clicking and makes better use of my vim muscle memory.

Puzzle solving

Something I keep saying about puzzle hunts is that it’s important to have lots of people on the team, even if they aren’t spending gross amounts of time solving puzzles. What’s really important is that there’s a continuous stream of “have you tried <thing>?” ideas and “that looks kinda like <other thing>” observations: otherwise, you get stuck trying to solve a puzzle and then get frustrated at not having ideas, which isn’t a great mental state to be producing ideas with.

I’m very grateful to my team members for contributing their time: I think I spent less time in that stuck state than I have in previous years, but there’s definitely still room for improvement there.

There were also a few times where I think we had trouble because we didn’t quite know what “rules” the puzzle authors were playing by, so to speak. In one case, I wrote a bunch of Python code looking for patterns in some data because of a perceived hint in the puzzle title, which turned out to be a wild goose chase. There was also at least one hint spent on something where we should have just been able to get it out by being more rigorous.

Results

Our team solved the first 12 puzzles, which put us in 239th place.

• Out of 717 registered teams: top 33%
• Out of 505 teams who solved at least one puzzle: top 47%

I’m pretty happy with how we went: I only wish we’d managed to finish our half-solved metapuzzle in time!

For next year’s Galactic Puzzle Hunt (or any other ones for that matter), I’ve been thinking lately about doing some training sessions (where you solve a puzzle, but there’s zero-cost hints as soon as you get stuck). If this sounds interesting, hit me up?