The Bay Area Treasure Hunt (BATH) is pot-luck. 
Each team can bring a puzzle.  
Since each team on the hook to write just one puzzle, 
no-one stresses too hard.  
I'd volunteered to help out on 
BATH 3
before I found out 
that hunt would be different from past BATHs--it ran 
overnight, GC set up a double-layered meta-puzzle, it ranged 
outside of San Francisco.  BATH3 had been a lot of work--and I 
say that even though I pulled a scream-and-run before I 
did much work.
BATH&nbap;4, maybe as a reaction against the effort of BATH&nbap;3, 
had almost no GC involvement. In BATH1 and BATH2, 
GC had helped teams with playtesting and production. 
But in BATH&nbap;4, teams would have to do that themselves.  
BATH-mistress Alexandra 
sent out a request for comments--would teams be up for a game like this?  
A quorum of teams was indeed interested.
Our team, Mystic Fish, was thus on the hook to come up with a 
puzzle on the theme: Dichotomy.  Our puzzle really was a team effort.
Alexandra found an interesting location.  Someone came up with the
puzzle's mechanic.  My contribution was (surprise, surprise) a couple 
of little Python programs that searched our constraint-space and 
spewed out a first draft.
As part of the research for this, I visited our puzzle site, in
   the Mission district.  I was pretty happy that our site was in
   the Mission district: I was close to a good taqueria.  So when it
   started raining, I skittered over to this taqueria.  But my 
   intended taqueria (Pancho Villa) was full of drunken Santas!  
   Thus I had to go to another taqueria a couple of blocks away.
   In the rain.  This puzzling stuff is rough, I tell you.
We tried a playtest on a paper version of the clue.  This didn't 
go well.  Our puzzle was supposed to take about 45 minutes to solve.
It took about two and a half hours.  Fortunately, most of this time
wasn't spent on the puzzle itself.  Rather... when we met up, we 
found out that no-one had printed up the puzzle ahead of time.  One of 
Dwight's kids lived nearby, and we spent an hour and a half going 
over there, failing to find their printer, transfering PDF documents
from the internet to the hard drive to a thumb drive to a laptop to...
I lost track.  Then there was a printer, but it was in a box.  Then 
it was unpacked from the box but it was out of toner.  Then we gave up
and decided to go to Fedex/Kinko's to print the thing out,
but when we were halfway down the block from the apartment, I realized
I'd left my backpack behind and...
The playtest part of the playtest went pretty well, though.
And some of the stronger puzzlers on the team came up with some nifty 
refinements that had eluded my little Python script.
It was the evening of GC Summit 2009.  Chris Roat 
(of 
Team 
Longshots) and I were mooching a ride from Rich Bragg (of 
team Blood and Bones).
BATH4 was coming up soon.  Rich predicted it would go long.  "Sure each team
knows there supposed to take about 45 minutes.  But it's to their
advantage to go long."  Each team was bringing a puzzle.  A team didn't have
to solve their own puzzle.  So if your puzzle took about an hour to solve--that
was an hour that other teams spent that you didn't spend.  If that other
team's puzzle only took 45 minutes for your team to solve, you'd gained 
15 minutes on them.  The logical conclusion: it was a good thing there was
an ending time on the game; otherwise there would be a tendency for it 
to go into overtime.
The next morning, Alexandra sent out mail from some NPL-ish get-together
where she'd found a couple of playtesters.  Our refinements on the puzzle
were fun!  People liked them!  But they made the puzzle longer--closer to 
an hour than to 45 minutes.  I thought about that conversation with Rich 
Bragg.  I thought, Uhm, this is going to look bad.  While we mailed
back and forth over the implications of this, Alexandra noticed: it wasn't 
going to be too bad if teams' puzzles took an hour each--a couple of teams 
had withdrawn their puzzles.  So she sent out an announcement to the captains:
an hour was OK.
In hindsight, I guess Team Mystic Fish had an advantage here.  Alexandra,
the main organizer, was on our team.  If some other team came up with a puzzle
that took an hour, their first thought would be to simplify it--who would have
the chutzpah to ask Alexandra for permission to use a tougher clue?  But when
we had a too-long clue, we were able to notice that there was a solution.
Ah well--we finished waaay late (I heard we were DQ'd on time), so I 
don't think anyone would have accused us of jockeying for time.
There have been rumblings of re-using this game's puzzles in a BANG or 
somesuch.  So I guess I shouldn't go into details.  But there were some 
high points.
- We went to a Peet's coffee to solve 
    Taft on a Raft's 
    elegant puzzle; and that puzzle 
    mentioned Peet's coffee.  We were so tickled by that that we 
    brought along our receipt as a souvenir.
 - coed astronomy 
    had a very environmental puzzle--and it was at a site
    that I walk past pretty often.  In fact, I had noticed that this was 
    a data-rich site. I had once just stood at the site, staring at its data, 
    trying to figure out how to turn it into a good clue.  And I failed.
    When it came time to pick our "favorite puzzle", I lobbied for this 
    one--because they'd succeeded well where I'd given up.
 - The Smoking GNU's puzzle 
    did a couple of impossible things.  
    
- 
    I watched the strong puzzlers 
    on the team solve this puzzle--and as they got started on extracting
    the answer, I was thinking That can't be how you extract the
    answer--that's too constrained.  They could never have constructed
    the puzzle to make that happen.  Fortunately, I didn't have a 
    better idea, so I kept my mouth shut.  And that impossible method 
    turned out to be right.  
    
 - They found an interesting game-ish site in a neighborhood which
    I thought contained no such sites.  Really, I thought that this 
    neighborhood was a waste of space, an sort of suburb that had somehow
    snuck inside the San Francisco city limits.  (Of course, New Yorkers
    say the same thing about the sunset district, where I live, so I 
    shouldn't put on airs.)  And yet... and yet... that was a good site.
 
 - We noticed that other teams weren't calling us up to ask for hints
    on our puzzle.  They just called up to confirm answers.  Whew, that
    was a relief.
 - The Burninators put together a fun puzzle that was very environmental.
    "Environmental", you will recall, is gamist jargon for "You gotta be
    at a particular place for the puzzle to work".  In the case of this puzzle..
    This was in a place which, every time I go there, I promise it will be
    the last time.  It's a place with mellow background music.  The 
    data-gathering segment of this clue thus involved listening to a lot of 
    mellow background music.  Pretty soon I had given up on concentrating on
    the clue.  The only way I knew to keep my morale up was to sing 
    "I Like Choco Bars" to myself constantly. I was no help on this one.
 - We used a lot of hints for the Longshots' puzzle, though it turned out
    that we were doing things correctly.
 - Team Loxi's puzzle was laugh-out-loud funny and then the final extraction
    step was elegant and really dichotomous and--.  I hope someone either runs
    a game with these puzzles or else publishes them somewhere.
 - The game was ending as we drove towards the Presidio.  Really, we should
    have driven to the end-party at Bocce restaurant.  But instead we stopped
    at Debbie's house.  Debbie is the nice lady who was watching Alexandra's
    dog during this event--and who happened to live a couple of blocks from
    where we opened up Dr Clue's puzzle envelope and realized we needed a 
    table to work on.  By rights, we should have driven back to the 
    restaurant--but we all like puzzles, and darned few of us were going to 
    stick around for the after-party.  So we hung out at Debbie's house and
    pushed pieces of paper around for a few minutes.  And then Alexandra got 
    a phone call from some teams at the finish line.  Bocce restaurant had
    forgotten about our reservation.  Teams were milling around the bar.
    So we packed up the puzzle and headed to the restaurant to do damage 
    control.  But by the time we got there, it seems the restaurant people
    had found an area to put us, after all.
 
That's a lot of high points.  This was a good day.
   I really liked this format.  It meant some uneven-ness in 
   puzzle quality, but it also meant some wild variety.