**********************************************************
	*** Internet Go Server MacMahon 1993 tournament report ***
	**********************************************************

Introduction 
------------ 
This file is a report of the running of the Internet Go Server MacMahon
1993 tournament, and it describes things from my point of view (Olli the
Chief Organizer :-). 

There may be other things to discuss, but these are the random things
that spring to my mind. 

Overview
--------
The whole thing was a large international Internet go tournament, held
in honour of the selfless people who run, maintain and develop the
Internet Go Server (IGS).  Games were generally played in either the
Berkeley IGS or the French IGS, although the odd game or a few got
played on a real board. 

The tournament started Mid-February (14 Feb to be accurate :-), took a
month to run, and consisted of 8 rounds.  The tournament system was
MacMahon. 

Problems
-------- 
As everyone knows, problems always develop.  It was no surprise to me
that they would, but the source of the problems was. 

The problems were time, illnesses, time, pairing, time, unplayed games,
time, time and time.  By that I don't mean it took unexpectedly much
time to run the tournament, no way.  I fully expected it to be quite the
eater of time it proved to be. 

Most of the problems were related to time.  Time zones, time frame for
each round, deadlines, time limits for games, and so on.  But the very
biggest problem was that almost nobody carefully read through the
tournament rules that I published (found also in the IGS), and so not a
few problems happened since people did not know what, how and why. 

First and foremost, it became almost immediately very clear that *LOTS*
of players had no idea whatsoever how time zones affect things.  I
consistently announced things using GMT, and expected people to use the
IGS 'uptime' command to find out and calculate the time difference to
their local time.  Despite my announcing this possibility, perhaps 20%
never did so.  During the later rounds Jan Van der Steen (referee)
kindly constructed a time zone table, which I included in each draw. 
This reduced the misunderstandings, but surprising though it is, did not
remove the problem!

The deadlines and time frame for each round were carefully thought out
beforehand, and quite clearly announced.  Despite this, as late as on
7th round some people still had no idea that games were expected to be
finished and reported at latest at the deadline ("What?! The game
rejected? And we played it yesterday!" -- "Yes you did, but that was two
days after the deadline.")

The timetable for the tournament was carefully constructed to give the
organizers just a little leeway and also some time to make the draw (12
hrs for this phase) and almost invariably the last results *and* last
notes informing about people dropping out came in about _half_an_hour_
_before_the_next_draw_should_be_out_! Speak about people not
understanding tournament schedule!

Also, in each round there was one day period when players were supposed
to have contacted, agreed on when to play, and perhaps even have played
the game.  Even in the very last round I still had to ask players what
was the result: "What? I haven't heard about him yet."

All these lead to an uncomfortably large number of "games" each round
getting judged.  In the first round, I seem to remember it was two games
(~0.7%), but during the last round or two, almost 20% of games had to be
judged. 

Scope for improvement
---------------------
One thing became clear very soon: the schedule (two games per week) was
tight.  Around 5th round, it had become very clear it was too tight, but
for various reasons I decided to no change it (it was already taking a
month and that was about the upper limit I could possibly do :-).

For any future tournament I would thus suggest a time frame of a game
per week (minus one day for organisatorial things), e.g.  each round
starting on Tuesday and ending on Monday.  This would require multiple
groups of organizers, unless the members of the one group are confident
they can give about 1.5 to 2 months of their time to run the tournament. 
This multiple groups of organizers is a thing I recommend anyways, since
the strain and workload was just about the limit of what our group could
do. 

Communication was a problem.  In later rounds I took up the habit of
including about one screenful of text into the pairing clearing up
things still unclear to players.  This helped very quickly, but it would
have been much better had people carefully read the compleat tournament
rules...  also, quite a few of the players were (are) not familiar with
e-mail and its uses -- especially some of the strongest players. 

A related problem was that players had difficulties contacting their
opponents due to the reasons in above paragraph.  Also, not many had
included their e-mail addresses into their IGS 'Address:' field. 

A further thing I recommend is to make the rules for handling unplayed
games, unsuccessful contact attempts and cases where no suitable time
can be agreed strict enough, and _enforce_them_mercilessly_.  We lots of
times did not enforce them mercilessly and this also increased our work
(we knew this but did it anyway :-).  The reason why I was for it, was
that I considered the number of players too small to do it properly.  It
would have lead to even worse pairings; even now there were not a few
games where players of 2 or 3 MM difference were paired. 

And last but not least, I would recommend one to try and find a very
good pairing program.  I lost quite a bit of work several times, and
also in other ways the one in use increased my work.  There is one
critically important consideration, though, why I used it: it worked. 
It did not crash, and it did a very good job at pairing players. 

My story
--------
The whole thing started about two months ago, when I had quite a bit
more of free time.  There was some discussion about running tournaments
in the news and it seemed to me that lots of people, especially those in
the US, had little idea of what tournament systems exist, much less what
system would be suitable to such a large tournament. 

I first tried to leave the idea alone, but it would not die in my mind. 
After several days I decided to see whether it was at all possible and
asked another player for the pairing program.  I did some half-hearted
tests, and in another two or three days I decided to go for it. 

First I mailed tweet and tim to check the ground where I stood.  As
neither was against the tournament, I started recruiting people for
officials.  The very first I recruited was Tero (the mail agent), and
this was an important prerequisite.  I knew he had more free time than I
did and also would be online very regularly barring illness. 

In a week I had all the people recruited.  The lucky shot here was
recruiting Jim (referee), who soon proved to be one of the three most
important members of the organization (others were me and Tero). 

During the time I recruited the officials, I also wrote all the bits and
pieces of text that eventually wound up into the IGS help system.  Also,
the text in the motd (message of the day) was very quickly put up to
give any interested players time to consider joining.  About the same
time with the motd, I posted the first pre-announcement into the
newsgroup reg.games.go.  The description of MacMahon system (a lousy
piece of writing but I had no time to do that one properly -- everyone
would soon see what it was all about anyway! :-) followed soon after,
and when I had the tournament rules ready and okayed by Matti (the Chief
Referee) I immediately posted them. 

During the next week (the week before the start of the tournament) about
70 people had registered -- and all too few of them above 1 dan, despite
my posting two registration reports into the net.  When the registration
deadline was at hand, I started a recruiting campaign (Jim had already
recruited a few strong players, and he continued the campaign after I
went to sleep) to get some substance to the tournament.  Indeed, soon we
had 100 participants, and when it was time to actually start the
tournament 10 more had registered and were accepted despite them being
late. 

This was a very important lesson: few people seem to read the motd, and
news has a very limited audience among the very strongest players of the
IGS. 

There still remained the problem of communicating with the players.  To
this end I requested everyone for an e-mail address and this proved to
be a very lucky thing indeed.  It was soon resolved to use mailing list
as the primary means of distribution of pairings and results, and news
as the secondary one. 

The first round pairing was very worksome indeed.  I had to write in all
the players into the pairing program, and it proved a huge work to learn
all the quirks of the program.  Huge amounts of work were lost, and I
had to do the work about three times in all.  Nevertheless, the pairing
was out in time, and the experience here proved very valuable later. 

All in all, the first round went through with little to no problems,
since all the players were prepared to play and had needed to spend a
bit of effort to register.  The only problem was that several games
exceeded the deadline and had to be judged, but that was a problem that
could not be abolished during the tournament. 

The real problems started at the second round.  Not a few games were
left unplayed, and this caused a huge amount of work.  Lucky me, Jim was
extremely well present in the IGS and also very active, and so he ended
up doing most of the judgements.  Only in cases where time was of
paramount importance did I pick up the phone and call Matti, otherwise I
would usually sit in the IGS to see who was the first referee to show up
-- and very soon I learned that it was almost invariably Jim. 

Indeed, the problem soon became serious.  People started realizing just
how much time and effort they were required to spend to play the
tournament games, and several dropped out without ever notifying us. 
But as I was still green, I was not quite in favour of dropping people
out, which is what we should have done, and what we actually did in the
later rounds.  Also, each round several new players would join the
tournament to compensate for those dropping out. 

Nevertheless, the real crisis was yet to come.  When the second round
was just started, Tero fell ill.  This immediately placed a huge burden
on me.  I called Tero once I got the word of this, and got the use of
his account.  Nevertheless, managing mail, gathering results, running
the pairing, and coordinating things -- all these were suddenly my
responsibility.  Luckily enough I had completely stripped myself of all
judgementary power and stashed that to the referees; it would clearly
have been too much otherwise, even now I was pushed uncomfortably close
to the limit. 

When time came to start the third round, I spent the better part of a
day to try to get the pairing ready.  Once I saw that the pairing was
going to be late, I wrote a short notice and posted it into the news,
but did not mail it, since the mailing script had been done by Tero and
I was not yet familiar with it.  The pairing was finally only 1.5 hrs
late, but when I tried to mail it, the mailer script just produced empty
messages.  After another half an hour I got it out properly. 

Also the fourth round was my sole responsibility, since right after Tero
recovered from his illness, his monitor died.  This time I was better
prepared, though, and so this round was not late.  Nevertheless, when
Tero finally announced in midst of the fourth round that he's up and
about again, I was extremely relieved. 

Towards the end of the tournament things eventually smoothed out: Tero
handling the mailing, Jim doing anything that called for judgements (and
that was *LOTS*), and me doing the pairing and generally coordinating
all that was necessary.  (That is to say that I'm now running out of
both ideas and the time to write them down :-)

Acknowledgements
----------------
Without the IGS the tournament would not have been possible at all.  I
am deeply grateful to all the people who have selflessly done huge
amounts of valuable work with it. 

It was due to the enormous amount of work done by Tero and Jim that I
was able to run the tournament.

Without the advise and support of Matti lots more things would have gone
awry, and lots more trouble would have developed.  I am not sure I could
have run the tournament without him. 

My thanks to all the tournament officials not mentioned above (Jan van
der Steen and the Appeals Committee).  Even if they did not need to do
much, the pure fact that they were there left my hands free to handle
other things ("Well, if you don't like it, go ahead and make an appeal!"
:-)

Well, I guess that's about it.  My apologies for this report being very
late, but I've been quite busy with my work lately.