MITMH 2020: 2019 Aug

Careful: I'm pretty blasé about spoilers for this year-old hunt. If you haven't already solved the puzzles you wanted to solve, maybe don't keep reading.

Every so often, fearless leader Corey sent out a "State of the Hunt" email. Here are some excerpts from August's emails to let you know what the writing team was doing. (And to remind me, as I write this, what we were doing.)

The biggest workstream is puzzle authoring, and the good news is that we’re making solid progress. The bad news is that that solid progress is tracking us to being short by about 40 puzzles by October (or, equivalently, about 5 weeks behind puzzle authoring schedule). Yar’s cooking up a backup plan should we need it. Stay tuned. Or, write more puzzles.

We are just under 6 weeks away from our Big Test Solve. BTS will cover all the puzzles from a few of the lands (ie, it’s the complete “first part” of our Hunt). If you’re a puzzle author targeting Castle, Enchanted Forest, or Space, please do what you can to get your puzzle going!

If you’re attending Miskatonic or if you’ll otherwise be in Boston next week, please subscribe to #mu-mit-aug16 and check out the schedule.

Linda is surveying the team for their availability during Hunt 2020 weekend.

Big Test Solve. This is the big milestone coming up in less than three weeks. If you plan to test solve and have not said as much in Linda’s logistics survey, please e-mail Martin to let him know you’re coming. On the puzzle side of BTS, of the 27 puzzles, 15 are done or nearly done, 7 need work that’s well-understood, but 5 are still in writing. The Hunt server is progressing with more functionality but some aspects of Hunt (eg, how we handle hints) remain unanswered. Focused teams for hints and timed unlocks are working on these; contact me if you want to know more.

Art. Natalie is making art; see #art for some samples. The focus presently is art for BTS.

Puzzle HTMLification. Larry is driving the puzzle prod effort and many people have been pressed into production techs, converting puzzles into HTML targeting BTS. If you’d like to take part and haven’t been tapped yet, please contact Larry.

Students. Linda completed our student recruiting flyer and shared it with Puzzle Club. If you know students at MIT who might be able to help, especially during Hunt weekend, please put them in touch with Puzzles Club and/or share this flyer.

As Production Czar, I continued to HTMLify early puzzles. When I ran into a puzzle that did something that didn't fit our system, Doug and I put our heads together to figure out how the system should change.

But we were running into fewer weird cases. Maybe we were ready to let some more HTML nerds tinker with this stuff.

Rich and Kiki Bragg hosted a testsolving party. I don't remember what I did there.

The Miskatonic Game was coming up in the Boston area. Thus, many Left Out left coasters would be in the Boston area. Thus, for a few days there would be many more potential testsolvers for the Whirlwind mini-runaround.

I did some revising on Whirlwind to get it ready for in-person testing.

Also in my effort to prepare for Boston (where it was hot), I got a haircut.

[Larry with no haircut] [Larry with haircut]

I arrived in Boston a few days ahead of the Miskatonic Game. This gave me time to do many things.

I edited Whirlwind, let some testsolvers at it, edited it some more, let another batch of testsolvers at it. In hindsight, I wish I'd noticed some details in these tests. And I wish I'd done some more rigorous testing myself.

An iPhone user reported that his phone had never asked for permission to use his geolocation. I assumed that was a problem with just that one phone—for most people, everything seemed to work fine. What I didn't realize was that this was the day's only iPhone tester. Team Left Out has many, many Googlers and ex-Googlers; many have Android phones, few have iPhones. I wish I'd realized that in the year 2019, the default iPhone web browser still couldn't use the phone GPS/geolocation stuff.

I had a good time at The Miskatonic Game

Perhaps motivated by Big Test Solve deadlines, puzzle authors cranked out puzzles more rapidly. Testsolvers kept pace. As Production Czar, I'd been HTMLifying puzzles as they became ready. But soon there would be enough ready at once such that I could ask other folks on the Production (HTMLification) Team to do their thing: there'd be a puzzle for each of them.

(Many of the puzzles were pretty simple, HTML-wise. I could have played "hero" to try to do it all myself. But this was our chance to work out the kinks in our team workflow.)

Before The Miskatonic Game, I'd Slack'd a heads-up: after The Game, I'd ask folks on the team to HTMLify puzzles.

And after The Game, that's what I did: assigned puzzles to people. Checked in on how things were going.

Instead of me as sole dispatcher handing out puzzles, in hindsight, I wish I'd let folks sign up to be, uh, mini-Production Czars. Maybe the Enchanted Forest round could have had its own Production Czar, keeping track of the HTMLification status of all puzzles in that round. (I don't know how many people would be interested in the mini-Production-Czar role. Maybe just one or two, leaving me watching over most rounds.)

For 2019, it worked out nicely that I was sole dispatcher. But later I thought about succession planning: for the long-term health of the team, I wish I'd found out which folks were interested in a manager-y role; in future years, such a person might graduate from mini-Production Czar to main Production Czar. And then I could sit on my butt and eat burritos.

On the other hand, maybe that's not a huge concern. Plenty of Left Out folks are managers and directors and such in the "real world." Any of them could step into the role, no problemo.

I still have a text file on my computer where I keep track of the folks on the Production team and the puzzles they're working on HTMLifying. (Except that the Hunt's done now, so now there aren't any puzzles and the file just looks like

Laura ()
ChrisY ()
Eric ()
Jan ()
Rich ()
Tammy ()      


Anyhow, this batch of puzzles went pretty smoothly.

Sometime in here, Linda Holman and Ylaine started a shared spreadsheet of "special" puzzles. Most of our puzzles were just web pages, but some were special: Teamwork Time, mini-runarounds, physical objects to pick up at HQ… It seemed likely that our Operations folks would want to keep track of information about these specially in some central spot. This was true

Suddenly, I was getting a lot of puzzle feedback from Stribs (Jeff Stribling). Stribs was in charge of fact-checking. That is, he was managing quality control. You might think that testsolvers would catch all problems, but it ain't so. A testsolver who partially figures out a puzzle and sees the answer is SQ_EAMISH probably won't notice that the missing letter is, thanks to mistake in the puzzle, V. They'll just happily report that they got the answer.

A good fact-checker doesn't just leap to the answer. A good fact-checker goes through everything, notices that mistaken V.

If it were me running the fact-checking, I'd gather a team, tell the team to make sure puzzle authors had written up solutions, then have the team look over those solutions closely. But apparently I'm lazy (?) because Stribs chose another approach: he testsolved all of the puzzles. In groups and on his own, he tested everything. When that test convinced him a puzzle was 100%, great. If he thought that the puzzle still needed some checking in addition to the testsolve, he did that.

I lost track of how many times Stribs saved our butts. He found many, many errors. Thus, he swept away many, many red herrings that otherwise could have tripped up teams.

Our puzzle "pipeline" was something like:

With BTS' approach, there was a rush of puzzles coming through testing. Stribs sprang into action, testing and fact-checking many many puzzles.

We asked puzzle authors to write hints for their puzzles. In hindsight, we didn't get this right at first. Our first approach: Puzzle authors were asked to consider each "aha" in their puzzle. For each "aha", write two versions of a hint: a "nudge" and a "sledgehammer". E.g. a "nudge" might be "It's interesting that the text has so many hyphens" while a "sledgehammer" might be "It's interesting that the text has so many hyphens and periods, like dashes and dots."

This was a good first attempt, based on experience with other puzzling events. But BTS would teach us that it wasn't a good fit for Mystery Hunt puzzles. We learned that in Mystery Hunt puzzles, teams were likely to get the "aha" just fine, yet make mistakes in the execution. Mystery Hunt puzzles tend to be big, so there are plenty of places to make mistakes.

Towards the end of August, I was thus mostly HTMLifying puzzles (and gleefully watching over other folks HTMLifying puzzles). But I was also responding to feedback on my own puzzles. And I was also pitching in kind of randomly on some puzzles. Since I wanted to avoid the situation of too many puzzles showing up at the last minute, I looked over lots of struggling puzzles. If a puzzle was stuck because, e.g., the author couldn't find time to trim silent parts off the ends of some audio files, that was something I could do to move puzzles towards completion.

Just scramble, scramble, keep moving those puzzles through the pipeline.

Sep [>]

comment? | | home |