Report
by Roland Hübscher, Regional Contest Director
This year, we had two contests going on: Getting ready and staying
alive; and solving the problem set prepared by judges. As a result, we
have two sets of winners: The volunteers including the judges making
the impossible possible and the official contestants. This report is
about the adventures of running a contest for the first time, trying
lots of new things, and a rather naïve director letting the techies do
whatever they want to -- since he knew they can do it. For the results
of the official programming contest, please see the standings and the problem set. The clear winners of the
official competition and finalists are Georgia Tech's Yellow Jackets A
and University of Central Florida's Chimera.
Getting ready
As usually in the Southeast U.S.A. region, all the teams brought their
hardware to the site one day before the contest. Auburn's central
location allowed most of the 72 teams to get in at quite an early
time. So, that's good, right? Actually not. We had planned to network
all the incoming computers which, for the first time in the Southeast,
had Linux installed. Furthermore, we required the teams to bring
specific hardware and software configurations so that we could get
everything up and running in no time. Or so we thought...
But it would be no real challenge if everything ran smoothly, wouldn't
it? First, we had to wait for enough electrical power until 6
p.m. instead of 1 p.m. due to the planning skills of some facilities
supervisors. Of course, I should have known, Dean Benefield of the
College of Engineering told me the next morning before he welcomed the
teams to Auburn University. With all the power on, the rest should be
quite straightforward. I mean, all the computers were supposed to be
nicely configured as described on our web
site. How naive of us. There were even a couple of Gateway boxes
that were never opened before they saw the contest site. Really, the
carton boxes were never opened before they arrived. And then, of
course, once the machines started up they ran Windows 98 and asked for
the OEM number...
Staying up all night and fixing it was the obvious answer. But we had
to leave the building at 11:30 p.m. so the solution was to fix as much
as we could through the network from outside the building since about
half of the 72 machines were up. In other words, about 36 machines
were in a rather dire state. That plan didn't work either since the
power had to be shut off during the night. Finally our network people
started to seriously suffer. They had worked for hours; the two main
network gurus, Doug Hughes and Jason Campbell, for weeks. The
excitement had waned and Jason sat in a corner with the most
disappointed face you can possibly imagine. Then he got really mad and
decided with Doug, James and the half dozen skilled helpers to setup a
few Linux machines over night, and swap them with the culprits -- the
non-working machines, that is.
The network's up. Now we just need to write the software.
When I arrive on Saturday morning at 6:00 a.m. Doug, Jerry, James, Ben
and William had configured 15 contest-ready machines and were swapping
them with the sick ones. At 9:45 a.m. we let the students finally
touch the machines although some are still not on the network, and
Doug is still writing the last few lines of the programming contest
software. Didn't he learn that this can never work?!
We have been ready for the floppy-based sneaker net for quite a while,
but we -- the network guys and me -- want to run the network. The
judges are getting nervous and more and more skeptical. Joe Apple, our
extremely valuable and helpful programming contest software consultant
is worried but agrees to push the decision about sneaker or cable as
far back as possible. In the end, we pushed it even further.
The contest is supposed to start at 1:00 p.m. and the contestants
start moving towards the door of the contest hall. Finally, after 5
minutes of testing we are ready for the decision. (This is not
supposed to be a joke! We even called it "stress-testing.") Joe is
realistic: The network has no serious chance of doing the job all
afternoon. I don't want to be realistic. And our judges play along
although they will probably suffer just as much as the
contestants. (The truth is, they suffered more.) The solution: We
start with the network and switch to the sneakers as soon as it goes
down. I don't really know what the judges or Joe were thinking at that
time, but I'm surely glad they agreed to this solution.
The contest starts. Will it ever end?
The contest starts at 1:36 p.m. and runs smoothly for the first ten
minutes. That's no big surprise: There haven't been any submissions
yet. Three hours have past. The printers print. The volunteers run,
some are already limping. The contestants program and the judges
sweat. Their computers become less and less cooperative. One judge
gets wounded: it's a paper cut. The others scramble and Laurie White,
our invaluable chief judge, is in her element. It's wild, stressful
and chaotic. That's the way she likes it. The night before she was so
tense because things ran way too smoothly.
Wow! It's now over four hours into the contest and we had no serious
problem. The network people hang around, enjoying their victory. They
are relaxed. The volunteers run around with the paper slips and make
sure the judges have absolutely no spare second. The runners look
tired. Very tired, actually. But they just keep going. And so do the
judges.
But then, the god of crummy plugs strikes. The network is down. The
contestants get noisy. I run around like a headless chicken until one
of the judges tells me to stop the contestants from hitting the
keyboard. They have to turn off the monitor. Our technical volunteers
run around with cables. That worries me.
The network is up. I'm relieved and start up those 72 teams.
The network's down. "Turn off your monitors again, please."
The network is up. "No," says the other corner of the room.
It's time to think about what to do. We don't want to stop the contest
just yet, because too many teams have not solved a problem yet and the
top teams haven't solved enough either. Laurie White and I ask for
Ali Orooji's and Bill Poucher's advice. It's always convenient to have
the super-regional director and the world-wide director on site. We
decide to go with floppies and sneaker net.
The network is up.
But this time, we don't want to overload it. Here's the solution. The
first row of 24 machines goes up. One minute later the next 24
machines go up, and then the last 24 machines. The students are only
partially happy. The others have one or two minutes more time than we
do.
Not true. The contest will be finished row-by-row.
The line that the network team pulled through the women's bathroom
lets all teams work for the final 45 minutes just as planned.
First row: Do not touch the keyboard in five seconds, four, three,
two, one, zero.
Second row: Do not touch the keyboard in five seconds, four, three,
two, one, zero.
Third row: Do not...
Believe me, we all are truly relieved. The contestants. The
judges. The network people. Not me, I was never really worried.
|
