I - A brief account of the facts
The 2002 South American Regional ACM Contest was held on Nov 8-9, 2002, in
eight different sites across South America: one site in Buenos Aires, Argentina;
four sites in Brazil: Porto Alegre, Recife, São Paulo-PUC and
São Paulo-USP; one site in Copiapó, Chile; and two sites in
Venezuela: Barquisimeto and Caracas. Altogehter 180 teams from eight different
countries have taken part in the Contest.
The judging was supposed to be centralized at São Paulo-USP, with
judges from Argentina, Venezuela and Chile joining their brazilian colleagues.
PC^2 was configured to direct all runs to SP-USP. During the warm-up on Friday
8, PC^2 came to a halt during a test with all sites but Buenos Aires. After
fruitless attempts to make it work we got ready for local judging, which
took enormous last minute efforts from site directors. On Saturday 9, for
the main sesssion, we decided to try at least partially centralized judging
in the brazilian sites in a move to keep some uniformity in judging. That
did not work either. After two attempts at bringing PC^2 back, SP-USP moved
to e-mail submission, while Recife, Porto Alegre had remote (PC^2) judging
in SP and SP-PUC had local (PC^2) judging. Meanwhile, Buenos Aires was having
its problems with PC^2 and decided to move to diskettes, understaffed for
that task, since it is a small site. Caracas had its problems too: after
a rough beginning with diskettes they managed to stabilize PC^2. The only
sites that were trouble-free were Copiapó and Barquisimeto.
The colapse of the SP-USP site was dramatic enough that required two major
interruptions of problem submissions; in one of these, students were actually
out of the contest room, without access to the problem booklets at all. As
a result, the contest was rescheduled to end 2,5 hours later than previously
agreed. In Buenos Aires the situation was also dramatic: teams were facing
judging delays of over one hour. The last run was actually judged hours after
the contest ended. Recife and Caracas had faced delays at the beggining but
then stabilized after an hour or so. All other sites seemed to be running
OK.
All sites except SP-USP had 5-hour contests. Buenos Aires progressed a little
further until communication with the central site (SP-USP) established that
only the runs submitted within the 5 hour limit would be considered.
II - The criteria for ranking teams
As a consequence of all these problems, it became clear to the steering committee
that it would be very hard to compare submission times of teams in two different
sites that had virtually different contests. Thus, the committe decided to
consider four groups of sites according to the delays they faced, each for
different reasons:
- group 1 has SP-USP, with about 40 teams, 7,5 hours of contest, two breakdowns
in PC2, then diskettes.
- group 2 has Buenos Aires, 18 teams, with very long delays in judging.
- group 3 has Recife-Brazil and Caracas-Venezuela (altogether 38 teams):
delays of about one hour at the beginning then switched to PC2 without problems from then on.
- group 4 has the remaining four sites, which had virtually no delays. These
added up to 83 teams.
The steering committee decided to assign one slot in the World Finals to
the first placed team of each group. That resulted in the following finalists
(in no particular order):
UBA-1 (Argentina)
AM/PM UFPE (Brazil)
PUC-Rio-1 (Brazil)
UNICAMP-1 (Brazil)
As a consequence of these criteria, it is not possible to obtain a consistent
global ranking. The ranking shown in the results
page merges together equally ranked teams in the four different groups
of sites: NO COMPARISON SHOULD BE MADE BETWEEN THE PERFORMANCES OF TWO
EQUALLY RANKED TEAMS; IN PARTICULAR, THE FOUR FIRST PLACED TEAMS ARE ONLY
THAT: FOUR TEAMS THAT CAME IN FIRST IN FOUR DIFFERENT CIRCUMSTANCES.
That is done only up to 8th rank, before the row filled with '*'s. After
this row, the table shows the teams with the times computed at their respective
sites, but they are not merged or ranked globally.
A final note on the times of the SP-USP site: they were taken from a complicated
formula that involved all the interruption periods and have no resemblance
with times from other sites; they are nevertheless uniform for the teams
at SP-USP.
We apologize for the delay in announcing the results, but we believe the
time taken was fundamental in reaching a minimally equitable result, given
the circumstances.
We thank the teams, coaches and all staff for their patience and high spirits.
We hope to count on their presence in the future, to keep the South American
Contest strong and a nice experience to all.
Ricardo Dahab
South American Regional Director
ACM-ICPC